metradar 0.1.0__py3-none-any.whl

metradar/retrieve_cmadaas.py

@@ -0,0 +1,3126 @@
+# _*_ coding: utf-8 _*_
+
+# Copyright (c) 2020 NMC Developers.
+# Distributed under the terms of the GPL V3 License.
+
+"""
+Retrieve the CMADaaS 大数据云平台 using REST API with pure python code.
+
+refer to:
+  http://idata.cma/cmadaas/
+"""
+
+import os
+import fnmatch
+import warnings
+import json
+import pickle
+import hashlib
+import uuid
+from datetime import datetime, timedelta
+import urllib3
+import urllib.request
+from pathlib import Path
+import numpy as np
+import pandas as pd
+import xarray as xr
+from tqdm import tqdm
+import nmc_met_io.config as CONFIG
+
+
+def get_rest_result(interface_id, params, url_only=False,
+                    dns=None, port=None, data_format='json'):
+    """
+    Get the http result from CAMDaaS REST api service.
+    2021/01/13, 采用hashlib.md5对输入参数进行加密, 则在url中不会出现pwd关键字, 数据传输更加安全.
+
+    :param interface_id: MUSIC interface id.
+    :param params: dictionary for MUSIC parameters.
+                   可以直接指定'serviceNodeId', 'userId'和'pwd'三个参数, 若没有, 则从 .nmcdev 配置文件中读取.
+    :param url_only: if url_only = True, return url string.
+    :param dns: CMADaaS数据库DNS地址, 若不指定, 则从 .nmcdev 配置文件中读取.
+    :param port, CMADaaS数据库DNS的端口, 若不指定, 则从 .nmcdev 配置文件中读取.
+    :param data_format: MUSIC server data output format.
+    :return:
+
+    :example:
+        params = {
+            'serviceNodeId':'NMIC_MUSIC_CMADAAS',
+            'userId':'test',
+            'pwd':'123456',
+            'dataCode':'SURF_CHN_MUL_HOR_N',
+            'elements':'Datetime,Station_Id_d,Lat,Lon,PRE_24h',
+            'times':'20200910000000',
+            'dataFormat':'json',
+            'limitCnt':'10'}
+        data = get_rest_result('getSurfEleByTime', params)
+    """
+
+    # set MUSIC server dns port
+    if dns is None:
+        dns  = CONFIG.CONFIG['CMADaaS']['DNS']
+    if port is None:
+        port = CONFIG.CONFIG['CMADaaS']['PORT']
+
+    # construct complete parameters
+    sign_params = params.copy()
+
+    # user information
+    if 'serviceNodeId' not in sign_params:
+        sign_params['serviceNodeId'] = CONFIG.CONFIG['CMADaaS']['serviceNodeId']
+    if 'userId' not in sign_params:
+         sign_params['userId'] = CONFIG.CONFIG['CMADaaS']['USER_ID']
+    if 'pwd' not in sign_params:
+        sign_params['pwd'] = CONFIG.CONFIG['CMADaaS']['PASSWORD']
+
+    # data interface Id and out data format
+    sign_params['interfaceId'] = interface_id.strip()
+    if 'dataFormat' not in sign_params:
+        sign_params['dataFormat'] = data_format.strip()
+    
+    # add time stamp and nonce code
+    sign_params['timestamp'] = str(int(datetime.now().timestamp()*1000))
+    sign_params['nonce'] = str(uuid.uuid1())
+
+    # construct sign string with hashlib md5 code
+    sign_str = ""
+    keys = sorted(sign_params)
+    for key in keys:
+        sign_str = sign_str + key + "=" + str(sign_params.get(key)).strip() + "&"
+    sign_str = sign_str[:-1]
+    sign_str = sign_str + '&sign=' + hashlib.md5(sign_str.encode(encoding='UTF-8')).hexdigest().upper()
+    sign_str = sign_str.replace('&pwd='+sign_params['pwd'].strip(), '')
+
+    # construct url
+    url_str = 'http://' + dns + ':' + port + '/music-ws/api?' + sign_str
+    if url_only:
+        return url_str
+
+    # request http contents
+    http = urllib3.PoolManager()
+    req = http.request('GET', url_str)
+    if req.status != 200:
+        print('Can not access the url: ' + url_str)
+        return None
+
+    return req.data
+
+
+def _load_contents(contents):
+    """
+    Extract information from contents.
+
+    Args:
+        contents (string): [description]
+
+    Returns:
+        [type]: [description]
+    """
+    if contents is None:
+        print('Return None.')
+        return None
+    
+    try:
+        contents = json.loads(contents.decode('utf-8').replace('\x00', ''), strict=False)
+    except Exception as e:
+        print(e)
+        print(contents)
+        return None
+    
+    if contents['returnCode'] != '0':
+        print(contents)
+        return None
+    
+    return contents
+
+
+def cmadaas_obs_convert_type(obs_data):
+    """
+    Convert observation dataframe to numeric types.
+    
+    Args:
+        obs_data (dataframe): data frame of observations.
+    """
+    for column in obs_data.columns:
+        if column.upper().startswith('STATION'):
+            continue
+        if column.lower() in [
+            "province", "country", "city", "cnty", "town",
+            "data_id", "rep_corr_id", "admin_code_chn"]:
+            continue
+        if column.upper() == "DATETIME":
+            obs_data[column] = pd.to_datetime(obs_data[column], format="%Y-%m-%d %H:%M:%S")
+            continue
+        obs_data[column] = pd.to_numeric(obs_data[column])
+        obs_data[column].mask(obs_data[column] >=  999990.0, inplace=True)
+
+    return obs_data
+
+
+def cmadaas_get_obs_latest_time(data_code="SURF_CHN_MUL_HOR", latestTime=12):
+    """
+    Get the latest time of the observations.
+    
+    Args:
+        data_code (str, optional): dataset code, like "SURF_CHN_MUL_HOR", 
+                                   "SURF_CHN_MUL_HOR_N". Defaults to "SURF_CHN_MUL_HOR".
+        latestTime (int, optional): latestTime > 0, like 2 is return 
+                                    the latest time in 2 hours. Defaults to 12.
+    Returns:
+        datetime object, the latest time
+
+    Examples:
+    >>> print(cmadaas_get_obs_latest_time())
+        2020-03-12 03:00:00
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'latestTime': str(latestTime+8)}    # 由于服务器是世界时, 因此需要+8才能与当前时间比较.
+
+    # interface id
+    interface_id = "getSurfLatestTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    time = pd.to_datetime(data['Datetime'], format="%Y%m%d%H%M%S")
+
+    return time[0]
+
+
+def cmadaas_get_obs_files(times, data_code="SURF_CMPA_RT_NC", out_dir=None, pbar=True):
+    """
+    Download surface observation file.
+    
+    注意cmadass设置了ip访问次数, 如果下载的文件数量过多, 则会返回"-5004 Reach the hor access line."
+    表示单位小时内检索过于频繁，请降低检索频次. 另外, 也要注意下载的文件大小, 如果文件只有几K，表示下载不成功, 删除重新下载.
+    
+    Args:
+        times (str): times for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'; or
+                     time range, '[YYYYMMDDHHMISS,YYYYMMDDHHMISS]'
+        data_code (str, optional): dataset code. Defaults to "SURF_CMPA_RT_NC".
+        out_dir (str, optional): download files to out_dir. if out_dir is None,
+                                 the cached directory will be used. Defaults to None.
+        pbar (bool, optional): Show progress bar, default to True.
+
+    Returns:
+        the list of download files path.
+
+    Examples:
+    >>> out_files = cmadaas_get_obs_files('20200311000000,20200312000000')
+    >>> out_files = cmadaas_get_obs_files('[20200308000000,20200312000000]')
+    """
+
+    if out_dir is None:
+        out_dir = CONFIG.get_cache_file(data_code, "", name="CMADaaS")
+    
+    times = times.strip()
+    if times[0] == '[':
+        # set retrieve parameters
+        params = {'dataCode': data_code,
+                  'timeRange': times} 
+        interface_id = "getSurfFileByTimeRange"
+    else:
+        # set retrieve parameters
+        params = {'dataCode': data_code,
+                'times': times}
+        interface_id = "getSurfFileByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # loop every file and download
+    out_files = []
+    files = tqdm(contents['DS']) if pbar else contents['DS']
+    for file in files:
+        out_file = Path(out_dir) / file['FILE_NAME']
+        if not out_file.is_file():
+            urllib.request.urlretrieve(file['FILE_URL'], out_file)
+        out_files.append(out_file)
+
+    return out_files
+
+
+def cmadaas_obs_by_time(times, data_code="SURF_CHN_MUL_HOR_N",
+                        sta_levels=None, ranges=None, order=None, 
+                        count=None, distinct=False, trans_type=True, 
+                        elements="Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM"):
+    """
+    Retrieve station records from CMADaaS by times.
+    
+    Args:
+        times (str): times for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        distinct (bool, optional): return unique records. Defaults to False.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+        obsdata = cmadaas_obs_by_time('20181108000000')
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'times': times,
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+    if distinct: params['distinct'] = "true"
+
+    # Interface, refer to
+    interface_id = "getSurfEleByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_time_range(time_range, data_code="SURF_CHN_MUL_HOR_N",
+                              sta_levels=None, ranges=None, order=None, 
+                              count=None, distinct=False, trans_type=True,
+                              elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve station records from CMADaaS by time range.
+    
+    Args:
+        time_range (str): time range for retrieve,  "[YYYYMMDDHHMISS, YYYYMMDDHHMISS]",
+                          like"[20160901000000,20161001000000]"
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        distinct (bool, optional): return unique records. Defaults to False.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> time_range = "[20180219000000,20180221000000]"
+    >>> sta_levels = "011,012,013"
+    >>> data_code = "SURF_CHN_MUL_DAY"
+    >>> elements = "Station_Id_C,Station_Name,Datetime,Lat,Lon,PRE_Time_0808"
+    >>> data = cmadaas_obs_by_time_range(time_range, sta_levels=sta_levels,
+                                        data_code=data_code, elements=elements)
+    >>> print "retrieve successfully" if data is not None else "failed"
+        retrieve successfully
+    >>> obs_data = cmadaas_obs_by_time_range(
+            "[20160901000000,20161001000000]", data_code="SURF_CHN_MUL_DAY_N",
+            elements="Station_Id_C,Station_Name,Lat,Lon,Alti,Datetime,PRE_Time_0808")
+    """
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'timeRange': time_range,
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements,}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+    if distinct: params['distinct'] = "true"
+
+    # interface id
+    interface_id = "getSurfEleByTimeRange"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_time_and_id(times, data_code="SURF_CHN_MUL_HOR_N",
+                               sta_levels=None, ranges=None, order=None, 
+                               count=None, trans_type=True,
+                               elements="Station_Id_C,Datetime,TEM",
+                               sta_ids="54511"):
+    """
+    Retrieve station records from CMADaaS by times and station ID
+    
+    Args:
+        times (str): time for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. Defaults to "54511".
+    
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_time_and_id('20170318000000')
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'times': times,
+              'staIds': sta_ids,
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleByTimeAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_time_range_and_id(time_range, data_code="SURF_CHN_MUL_HOR_N",
+                                     sta_levels=None, ranges=None, order=None, 
+                                     count=None, trans_type=True,
+                                     elements="Station_Id_C,Datetime,PRE",
+                                     sta_ids="54511"):
+    """
+    Retrieve station records from CMADaaS by time range and station ID
+    
+    Args:
+        time_range (str): time range for retrieve,  "[YYYYMMDDHHMISS, YYYYMMDDHHMISS]",
+                          like"[201509010000,20150903060000]"
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. Defaults to "54511".
+    
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_time_range_and_id("[20160801000000,20160802000000]")
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'timeRange': time_range,
+              'staIds': sta_ids,
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleByTimeRangeAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_rect_by_time(times, limit, data_code="SURF_CHN_MUL_HOR_N",
+                                sta_levels=None, ranges=None, order=None, 
+                                count=None, trans_type=True,
+                                elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve station records from CIMISS in region by times.
+    
+    Args:
+        times (str): time for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        limit (list):  map limits, [min_lat, min_lon, max_lat, max_lon]
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_in_rect_by_time('20170320000000', [35, 110, 45, 120])
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'times': times,
+              'minLat': '{:.10f}'.format(limit[0]),
+              'minLon': '{:.10f}'.format(limit[1]),
+              'maxLat': '{:.10f}'.format(limit[2]),
+              'maxLon': '{:.10f}'.format(limit[3]),
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleInRectByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_rect_by_time_range(time_range, limit, data_code="SURF_CHN_MUL_HOR_N",
+                                      ranges=None, order=None, 
+                                      count=None, trans_type=True,
+                                      elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve observation records from CIMISS by rect and time range.
+    
+    Args:
+        time_range (str): time for retrieve, "[YYYYMMDDHHMISS,YYYYMMDDHHMISS]"
+        limit (list):  map limits, [min_lat, min_lon, max_lat, max_lon]
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> elements = ("Station_Id_C,Station_Id_d,Station_Name,"
+                    "Station_levl,Datetime,Lat,Lon,PRE_Time_0808")
+    >>> time_range = "[20160801000000,20160801000000]"
+    >>> data_code = "SURF_CHN_MUL_DAY"
+    >>> data = cmadaas_obs_in_rect_by_time_range(
+            time_range,[35,110,45,120], data_code=data_code,
+            elements=elements)
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'timeRange': time_range,
+              'minLat': '{:.10f}'.format(limit[0]),
+              'minLon': '{:.10f}'.format(limit[1]),
+              'maxLat': '{:.10f}'.format(limit[2]),
+              'maxLon': '{:.10f}'.format(limit[3]),
+              'orderby': order if order is not None else "Datetime:ASC"}
+    #if sta_levels is not None: params['staLevels'] = sta_levels    # getSurfEleInRectByTimeRange not support this parameter.
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleInRectByTimeRange"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_admin_by_time(times, admin="110000", data_code="SURF_CHN_MUL_HOR_N",
+                                 sta_levels=None, ranges=None, order=None, 
+                                 count=None, trans_type=True,
+                                 elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve station records from CIMISS in provinces by time.
+    
+    Args:
+        times (str): time for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        admin (str, optional):  administration(or province code), sperated by ",",
+                      like "110000" is Beijing, "440000" is Guangdong. Defaults to "110000".
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_in_admin_by_time('20200206000000', admin="110000")
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'times': times,
+              'adminCodes': admin,
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleInRegionByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_admin_by_time_range(time_range, admin="110000", data_code="SURF_CHN_MUL_HOR_N",
+                                       sta_levels=None, ranges=None, order=None, 
+                                       count=None, trans_type=True,
+                                       elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve observation records from CIMISS by provinces and time range.
+    
+    Args:
+        time_range (str): time for retrieve, "[YYYYMMDDHHMISS,YYYYMMDDHHMISS]"
+        admin (str, optional):  administration(or province code), sperated by ",",
+                      like "110000" is Beijing, "440000" is Guangdong. Defaults to "110000".
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> elements = ("Station_Id_C,Station_Id_d,Station_Name,"
+                    "Station_levl,Datetime,Lat,Lon,PRE_Time_0808")
+    >>> time_range = "[20160801000000,20160801000000]"
+    >>> data_code = "SURF_CHN_MUL_DAY"
+    >>> data = cmadaas_obs_in_admin_by_time_range(
+            time_range,admin="110000", data_code=data_code,
+            elements=elements)
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'timeRange': time_range,
+              'adminCodes': admin,
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleInRegionByTimeRange"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_basin_by_time(times, basin="CJLY", data_code="SURF_CHN_MUL_HOR_N",
+                                 sta_levels=None, ranges=None, order=None, 
+                                 count=None, trans_type=True,
+                                 elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve station records from CIMISS in basin by time.
+    
+    Args:
+        times (str): time for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        basin (str, optional):  basin codes, sperated by ",",  like "CJLY" is Yangzi River, 
+                                "sta_2480" is 2480 stations. Defaults to "CJLY".
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_in_basin_by_time('20200206000000', basin="CJLY")
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'times': times,
+              'basinCodes': basin,
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getSurfEleInBasinByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_basin_by_time_range(time_range, basin="CJLY", data_code="SURF_CHN_MUL_HOR_N",
+                                       sta_levels=None, ranges=None, order=None, 
+                                       count=None, trans_type=True,
+                                       elements="Station_Id_C,Datetime,Lat,Lon,TEM"):
+    """
+    Retrieve observation records from CIMISS by basin and time range.
+    
+    Args:
+        time_range (str): time for retrieve, "[YYYYMMDDHHMISS,YYYYMMDDHHMISS]"
+        basin (str, optional):  basin codes, sperated by ",",  like "CJLY" is Yangzi River, 
+                                "sta_2480" is 2480 stations. Defaults to "CJLY".
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        sta_levels (str, optional): station levels, seperated by ',',
+             like "011,012,013" for standard, base and general stations. Defaults to None.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+    >>> elements = ("Station_Id_C,Station_Id_d,Station_Name,"
+                    "Station_levl,Datetime,Lat,Lon,PRE_Time_0808")
+    >>> time_range = "[20160801000000,20160801000000]"
+    >>> data_code = "SURF_CHN_MUL_DAY"
+    >>> data = cmadaas_obs_in_basin_by_time_range(
+            time_range, basin="CJLY", data_code=data_code,
+            elements=elements)
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'timeRange': time_range,
+              'basinCodes': basin,
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if sta_levels is not None: params['staLevels'] = sta_levels
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    # http://10.20.76.55/cimissapiweb/apicustomapiclassdefine_list.action?ids=getNafpEleGridByTimeAndLevelAndValidtime&apiclass=NAFP_API
+    interface_id = "getSurfEleInBasinByTimeRange"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_period(minYear, maxYear, minMD, maxMD, data_code="SURF_CHN_MUL_HOR_N",
+                          ranges=None, order=None, count=None, trans_type=True,
+                          elements="Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM"):
+    """
+    Retrieve station records from CIMISS by same period in years.
+    
+    Args:
+        minYear (int): start year, like 2005
+        maxYear (int): end year, like 2005
+        minMD (str): start date, like "0125" is 01/25
+        maxMD (str): end date, like "0205" is 02/25
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+        obsdata = cmadaas_obs_by_period(2015, 2018, "0501", "0505")
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'minYear': str(int(minYear)),
+              'maxYear': str(int(maxYear)),
+              'minMD': str(minMD),
+              'maxMD': str(maxMD),
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # Interface, refer to
+    interface_id = "getSurfEleByInHistoryBySamePeriod"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_period_and_id(minYear, maxYear, minMD, maxMD, data_code="SURF_CHN_MUL_HOR_N",
+                                 ranges=None, order=None, count=None, trans_type=True,
+                                 elements="Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM",
+                                 sta_ids="54511"):
+    """
+    Retrieve station id records from CIMISS by same period in years.
+    
+    Args:
+        minYear (int): start year, like 2005
+        maxYear (int): end year, like 2005
+        minMD (str): start date, like "0125" is 01/25
+        maxMD (str): end date, like "0205" is 02/25
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+         sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. Defaults to "54511".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+        obsdata = cmadaas_obs_by_period_and_id(2015, 2018, "0501", "0505", sta_ids="54511")
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'minYear': str(int(minYear)),
+              'maxYear': str(int(maxYear)),
+              'minMD': str(minMD),
+              'maxMD': str(maxMD),
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements,
+              'staIds': sta_ids}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # Interface, refer to
+    interface_id = "getSurfEleInHistoryBySamePeriodAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_in_admin_by_period(minYear, maxYear, minMD, maxMD, admin="110000", data_code="SURF_CHN_MUL_HOR_N",
+                                   ranges=None, order=None, count=None, trans_type=True,
+                                   elements="Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM"):
+    """
+    Retrieve station records from CIMISS by same period in years and in provinces.
+    
+    Args:
+        minYear (int): start year, like 2005
+        maxYear (int): end year, like 2005
+        minMD (str): start date, like "0125" is 01/25
+        maxMD (str): end date, like "0205" is 02/25
+        admin (str, optional):  administration(or province code), sperated by ",",
+                      like "110000" is Beijing, "440000" is Guangdong. Defaults to "110000".
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MUL_HOR_N".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+            Defaults to "Station_Id_C,Station_Id_d,lat,lon,Datetime,TEM".
+    
+    Returns:
+        pandas data frame: observation records.
+    
+    Examples:
+        obsdata = cmadaas_obs_in_admin_by_period(2015, 2018, "0501", "0505", admin="110000")
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'minYear': str(int(minYear)),
+              'maxYear': str(int(maxYear)),
+              'minMD': str(minMD),
+              'maxMD': str(maxMD),
+              'adminCodes': admin,
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # Interface, refer to
+    interface_id = "getSurfEleInHistoryBySamePeriodAndRegion"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_grid_by_time(time, limit=None, data_code="SURF_CMPA_FAST_5KM",
+                             fcst_ele="PRE", zoom=None, varname='data', 
+                             units=None, scale_off=None,
+                             cache=True, cache_clear=True):
+    """
+    Retrieve surface analysis grid products, like CMPAS-V2.1融合降水分析实时数据产品(NC).
+    For SURF_CMPA_RT_NC, this function will retrieve the 0.01 resolution data and take long time.
+    该函数主要用于降水实况分析数据检索, 若需要温度、风的实况分析, 使用cmadaas_analysis_by_time.
+
+    :param time: analysis time, datetime object or string like "20171008000000", format: YYYYMMDDHHMISS
+    :param limit: [min_lat, min_lon, max_lat, max_lon]
+    :param data_code: MUSIC data code, default is "SURF_CMPA_FRT_5KM"
+        "SURF_CMPA_RT_NC": 实时产品滞后约50~60分钟,分两种空间分辨率: 1小时/0.05°、1小时/0.01°
+        "SURF_CMPA_FRT_5KM_10MIN"(PRE_10MIN): CMPAS-V2.1融合降水分析实时数据10分钟产品
+        "SURF_CMPA_FAST_5KM": CMPAS-V2.1融合降水分析快速数据小时产品, 时效在15分钟以内, 包含小时降水和3小时累计降水两种要素
+        "SURF_CMPA_FRT_5KM": CMPAS-V2.1融合降水分析实时数据小时产品(GRIB, 5km)
+        "SURF_CMPA_FAST_5KM_DAY": CMPAS-V2.1融合降水分析快速数据日产品(GRIB, 5km)
+        "SURF_CMPA_FRT_5KM_DAY": CMPAS-V2.1融合降水分析实时数据日产品(GRIB, 5km)
+    :param fcst_ele: elements
+    :param zoom: the zoom out integer > 1, like 2. 像元缩小倍数, 数值大于1, 1的整数倍.
+    :param varname: set variable name, default is 'data'
+    :param units: forecast element's units, defaults to retrieved units.
+    :param scale_off: [scale, offset], return values = values*scale + offset.
+    :param cache: cache retrieved data to local directory, default is True.
+    :return: data, xarray type
+
+    :Example:
+    >>> time_str = "20171106120000"
+    >>> data_code = "SURF_CMPA_FAST_5KM"
+    >>> data = cmadaas_obs_grid_by_time(time_str, data_code=data_code, fcst_ele="PRE")
+    """
+    # check time_str type
+    if isinstance(time, datetime):
+        time_str = time.strftime("%Y%m%d%H%M%S")
+    elif isinstance(time, str):
+        time_str = time
+    else:
+        warnings.warn('The type of time is not supportted')
+        return None
+
+    # retrieve data from cached file
+    if cache:
+        directory = os.path.join(data_code, fcst_ele)
+        filename = time_str
+        if limit is not None:
+            filename = filename + '.' + str(limit).replace(" ","")
+        cache_file = CONFIG.get_cache_file(directory, filename, name="CMADaaS", cache_clear=cache_clear)
+        if cache_file.is_file():
+            with open(cache_file, 'rb') as f:
+                data = pickle.load(f)
+                return data
+
+    # set retrieve parameters
+    if limit is None:
+        params = {'dataCode': data_code,
+                  'time': time_str,
+                  'fcstEle': fcst_ele}
+        if zoom is not None: params['zoomOut'] = str(zoom)
+        interface_id = "getSurfEleGridByTime"
+    else:
+        params = {'dataCode': data_code,
+                  'time': time_str,
+                  'minLat': '{:.10f}'.format(limit[0]),
+                  "minLon": '{:.10f}'.format(limit[1]),
+                  "maxLat": '{:.10f}'.format(limit[2]),
+                  "maxLon": '{:.10f}'.format(limit[3]),
+                  'fcstEle': fcst_ele}
+        if zoom is not None: params['zoomOut'] = str(zoom)
+        interface_id = "getSurfEleGridInRectByTime"
+    
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # get time information
+    time = datetime.strptime(time_str, '%Y%m%d%H%M%S')
+    time = np.array([time], dtype="datetime64[ms]")
+
+    # extract coordinates and data
+    start_lat = float(contents['startLat'])
+    start_lon = float(contents['startLon'])
+    nlon = int(contents['lonCount'])
+    nlat = int(contents['latCount'])
+    dlon = float(contents['lonStep'])
+    dlat = float(contents['latStep'])
+    lon = start_lon + np.arange(nlon) * dlon
+    lat = start_lat + np.arange(nlat) * dlat
+    name = contents['fieldNames']
+    if units is None:
+        units = contents['fieldUnits']
+    
+    # define coordinates and variables
+    time_coord = ('time', time)
+    lon_coord = ('lon', lon, {
+        'long_name':'longitude', 'units':'degrees_east',
+        '_CoordinateAxisType':'Lon', 'axis': "X"})
+    lat_coord = ('lat', lat, {
+        'long_name':'latitude', 'units':'degrees_north',
+        '_CoordinateAxisType':'Lat', 'axis': "Y"})
+    varattrs = {'long_name': name, 'units': units}
+
+    # construct xarray
+    data = np.array(contents['DS'], dtype=np.float32)
+    data[data == 9999.] = np.nan
+    if scale_off is not None:
+        data = data * scale_off[0] + scale_off[1]
+    data = data[np.newaxis, ...]
+    data = xr.Dataset({
+        varname:(['time', 'lat', 'lon'], data, varattrs)},
+        coords={'time':time_coord, 'lat':lat_coord, 'lon':lon_coord})
+
+    # add global attributes
+    data.attrs['Conventions'] = "CF-1.6"
+    data.attrs['Origin'] = 'CIMISS Server by MUSIC API'
+
+    # cache data
+    if cache:
+        with open(cache_file, 'wb') as f:
+            pickle.dump(data, f, protocol=pickle.HIGHEST_PROTOCOL)
+
+    # return data
+    return data
+
+
+def cmadaas_obs_grid_by_times(times_str, pbar=True, allExists=True, **kargs):
+    """
+    Retrieve multiple surface analysis grid products, like CMPAS-V2.1融合降水分析实时数据产品（NC）.
+
+    :param times_str: analysis time string list, like ["20200208000000", "20200209000000"], format: YYYYMMDDHHMISS
+    :param allExists (boolean): all files should exist, or return None.
+    :param pbar (boolean): Show progress bar, default to False.
+    :param **kargs: key arguments passed to cmadaas_model_by_time function.
+    :return: data, xarray type
+
+    Examples:
+    >>> times = ["20200208000000", "20200209000000"]
+    >>> data_code = "SURF_CMPA_FRT_5KM"
+    >>> data = cmadaas_obs_grid_by_times(times, data_code=data_code, fcst_ele="PRE")
+    """
+
+    dataset = []
+    tqdm_times_str = tqdm(times_str, desc=kargs['data_code'] + ": ") if pbar else times_str
+    for time_str in tqdm_times_str:
+        data = cmadaas_obs_grid_by_time(time_str, **kargs)
+        if data:
+            dataset.append(data)
+        else:
+            if allExists:
+                warnings.warn("{} doese not exists.".format(kargs['data_code']+'/'+time_str))
+                return None
+    
+    return xr.concat(dataset, dim='time')
+
+
+def cmadaas_obs_by_days_of_year(data_code="SURF_CHN_DAY_MMUT_19812010", 
+                                days_of_year=None, elements="Station_Name,Station_Id_C,Day_Seq,TEM_Avg",
+                                ranges=None, count=None, trans_type=True, sta_ids="54511"):
+    """
+    按日序检索地面累年日值数据要素, 为累年日值专用.
+
+    Args:
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_DAY_MMUT_19812010".
+        mons_of_year (str, optional): months of year. Defaults to "1,2,3,4,5,6,7,8,9,10,11,12".
+        elements (str, optional): . Defaults to "Station_Name,Station_Id_C,Day_Seq,TEM_Avg".
+        ranges (str, optional): elements value ranges, seperated by ';'
+                                range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                                       [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b. 
+                                list: a,b,c;
+                                e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.
+                                Defaults to None.
+        count (str, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric.
+                                     Defaults to True.
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. If None, return all stations. Defaults to None.
+
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_days_of_year(sta_ids="54511")
+    """
+
+    # check days_of_year
+    if days_of_year is None:
+        days_of_year = ",".join((np.arange(365)+1).astype("str"))
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'daysOfYear': days_of_year}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    if sta_ids is None:
+        interface_id = "getSurfMDayEleByDaysOfYear"
+    else:
+        params['staIds'] = sta_ids
+        interface_id = "getSurfMDayEleByDaysOfYearAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_pens_of_year(data_code="SURF_CHN_PEN_MMUT_19812010", 
+                                pens_of_year=None, elements="Station_Name,Station_Id_C,Pen_Seq,TEM_Avg",
+                                ranges=None, count=None, trans_type=True, sta_ids="54511"):
+    """
+    按候序检索地面累年候值数据要素, 为累年候值专用.
+
+    Args:
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_PEN_MMUT_19812010".
+        mons_of_year (str, optional): months of year. Defaults to "1,2,3,4,5,6,7,8,9,10,11,12".
+        elements (str, optional): . Defaults to "Station_Name,Station_Id_C,Pen_Seq,TEM_Avg".
+        ranges (str, optional): elements value ranges, seperated by ';'
+                                range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                                       [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b. 
+                                list: a,b,c;
+                                e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.
+                                Defaults to None.
+        count (str, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric.
+                                     Defaults to True.
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. If None, return all stations. Defaults to None.
+
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_pens_of_year(sta_ids="54511")
+    """
+
+    # check days_of_year
+    if pens_of_year is None:
+        pens_of_year = ",".join((np.arange(72)+1).astype("str"))
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'pensOfYear': pens_of_year}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    if sta_ids is None:
+        interface_id = "getSurfMPenEleByPensOfYear"
+    else:
+        params['staIds'] = sta_ids
+        interface_id = "getSurfMPenEleByPensOfYearAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_tens_of_year(data_code="SURF_CHN_TEN_MMUT_19812010", 
+                                tens_of_year=None, elements="Station_Name,Station_Id_C,Ten_Seq,TEM_Avg",
+                                ranges=None, count=None, trans_type=True, sta_ids="54511"):
+    """
+    按旬序检索地面累年旬值数据要素, 为累年旬值专用.
+
+    Args:
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_TEN_MMUT_19812010".
+        mons_of_year (str, optional): months of year. Defaults to "1,2,3,4,5,6,7,8,9,10,11,12".
+        elements (str, optional): . Defaults to "Station_Name,Station_Id_C,Ten_Seq,TEM_Avg".
+        ranges (str, optional): elements value ranges, seperated by ';'
+                                range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                                       [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b. 
+                                list: a,b,c;
+                                e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.
+                                Defaults to None.
+        count (str, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric.
+                                     Defaults to True.
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. If None, return all stations. Defaults to None.
+
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_tens_of_year(sta_ids="54511")
+    """
+
+    # check days_of_year
+    if tens_of_year is None:
+        tens_of_year = ",".join((np.arange(36)+1).astype("str"))
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'tensOfYear': tens_of_year}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    if sta_ids is None:
+        interface_id = "getSurfMTenEleByTensOfYear"
+    else:
+        params['staIds'] = sta_ids
+        interface_id = "getSurfMTenEleByTensOfYearAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_months_of_year(data_code="SURF_CHN_MON_MMUT_19812010", 
+                                  mons_of_year="1,2,3,4,5,6,7,8,9,10,11,12",
+                                  elements="Station_Name,Station_Id_C,Mon_Seq,RHU_Avg",
+                                  ranges=None, count=None, trans_type=True, sta_ids=None):
+    """
+    按月序检索地面累年月值数据要素, 为累年月值专用.
+
+    Args:
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_MON_MMUT_19812010".
+        mons_of_year (str, optional): months of year. Defaults to "1,2,3,4,5,6,7,8,9,10,11,12".
+        elements (str, optional): . Defaults to "Station_Name,Station_Id_C,Mon_Seq,RHU_Avg".
+        ranges (str, optional): elements value ranges, seperated by ';'
+                                range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                                       [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b. 
+                                list: a,b,c;
+                                e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.
+                                Defaults to None.
+        count (str, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric.
+                                     Defaults to True.
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. If None, return all stations. Defaults to None.
+
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_months_of_year()
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'monsOfYear': mons_of_year}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    if sta_ids is None:
+        interface_id = "getSurfMMonEleByMonthsOfYear"
+    else:
+        params['staIds'] = sta_ids
+        interface_id = "getSurfMMonEleByMonthsOfYearAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_obs_by_years(data_code="SURF_CHN_YER_MMUT_19812010",
+                         elements="Station_Name,Station_Id_C,RHU_Avg",
+                         ranges=None, count=None, trans_type=True, sta_ids=None):
+    """
+    检索地面累年年值数据要素, 为累年年值专用.
+
+    Args:
+        data_code (str, optional): dataset code. Defaults to "SURF_CHN_YER_MMUT_19812010".
+        elements (str, optional): . Defaults to "Station_Name,Station_Id_C,RHU_Avg".
+        ranges (str, optional): elements value ranges, seperated by ';'
+                                range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                                       [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b. 
+                                list: a,b,c;
+                                e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.
+                                Defaults to None.
+        count (str, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric.
+                                     Defaults to True.
+        sta_ids (str, optional): station ids, 'xxxxx,xxxxx,...'. If None, return all stations. Defaults to None.
+
+    Returns:
+       pandas data frame: observation records.
+    
+    Examples:
+    >>> data = cmadaas_obs_by_years()
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    if sta_ids is None:
+        interface_id = "getSurfMYerEle"
+    else:
+        params['staIds'] = sta_ids
+        interface_id = "getSurfMYerEleByStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_get_sounding_latest_time(data_code="UPAR_CHN_MUL_FTM", latestTime=24):
+    """
+    Get the latest time of the soundings.
+    
+    Args:
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        latestTime (int, optional): latestTime > 0, like 2 is return 
+                                    the latest time in 2 hours. Defaults to 12.
+    Returns:
+        datetime object, the latest time
+
+    Examples:
+    >>> print(cmadaas_get_sounding_latest_time())
+        2020-03-12 03:00:00
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'latestTime': str(latestTime+8)}    # 由于服务器是世界时, 因此需要+8才能与当前时间比较.
+
+    # interface id
+    interface_id = "getUparLatestTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    time = pd.to_datetime(data['Datetime'], format="%Y%m%d%H%M%S")
+
+    return time[0]
+
+
+def cmadaas_sounding_by_time(times, data_code="UPAR_CHN_MUL_FTM", ranges=None, 
+                             order=None, count=None, distinct=False, trans_type=True, 
+                             elements="Station_Name,Station_Id_C,Lat,Lon,Alti,Year,Mon,Day,Hour,Min,Second,PRS_HWC,GPH,TEM,DPT,DTD,WIN_D"):
+    """
+    Retrieve sounding records from CMADaaS by times.
+    
+    Args:
+        times (str): times for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        distinct (bool, optional): return unique records. Defaults to False.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+    
+    Returns:
+        pandas data frame: sounding records.
+    
+    Examples:
+        uparData = cmadaas_sounding_by_time('20220531000000')
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'times': times,
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+    if distinct: params['distinct'] = "true"
+
+    # Interface, refer to
+    interface_id = "getUparEleByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_sounding_by_time_and_press(
+        times, data_code="UPAR_CHN_MUL_FTM", press=500, ranges=None,
+        order=None, count=None, distinct=False, trans_type=True, 
+        elements="Station_Name,Station_Id_C,Lat,Lon,Alti,Year,Mon,Day,Hour,Min,Second,PRS_HWC,GPH,TEM,DPT,DTD,WIN_D"):
+    """
+    Retrieve sounding records from CMADaaS by times.
+    
+    Args:
+        times (str): times for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        press (str, optional): pressure level. Defaults to 500.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        distinct (bool, optional): return unique records. Defaults to False.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+    
+    Returns:
+        pandas data frame: sounding records.
+    
+    Examples:
+        uparData = cmadaas_sounding_by_time_and_press('20220531000000', press=500)
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'times': times,
+              'pLayers': str(press).strip(),
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+    if distinct: params['distinct'] = "true"
+
+    # Interface, refer to
+    interface_id = "getUparEleByTimeAndPress"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_sounding_by_time_and_height(
+        times, data_code="UPAR_CHN_MUL_FTM", height=5000, ranges=None,
+        order=None, count=None, distinct=False, trans_type=True, 
+        elements="Station_Name,Station_Id_C,Lat,Lon,Alti,Year,Mon,Day,Hour,Min,Second,PRS_HWC,GPH,TEM,DPT,DTD,WIN_D"):
+    """
+    Retrieve sounding records from CMADaaS by times.
+    
+    Args:
+        times (str): times for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        height (str, optional): height level. Defaults to 5000.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        distinct (bool, optional): return unique records. Defaults to False.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+    
+    Returns:
+        pandas data frame: sounding records.
+    
+    Examples:
+        uparData = cmadaas_sounding_by_time_and_height('20220531000000', height=5000)
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'times': times,
+              'hLayers': str(height).strip(),
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+    if distinct: params['distinct'] = "true"
+
+    # Interface, refer to
+    interface_id = "getUparEleByTimeAndHeight"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_sounding_in_rect_by_time(
+        times, limit=[3.51, 73.33, 53.33, 135.05], data_code="UPAR_CHN_MUL_FTM",
+        ranges=None, order=None, count=None, trans_type=True,
+        elements="Station_Name,Station_Id_C,Lat,Lon,Alti,Year,Mon,Day,Hour,Min,Second,PRS_HWC,GPH,TEM,DPT,DTD,WIN_D,WIN_S"):
+    """
+    Retrieve sounding records from CIMISS in region by times.
+    
+    Args:
+        times (str): time for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        limit (list, optional):  map limits, [min_lat, min_lon, max_lat, max_lon]
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+    
+    Returns:
+        pandas data frame: sounding records.
+    
+    Examples:
+    >>> data = cmadaas_sounding_in_rect_by_time('20220531000000', limit=[3.51, 73.33, 53.33, 135.05])
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'times': times,
+              'minLat': '{:.10f}'.format(limit[0]),
+              'minLon': '{:.10f}'.format(limit[1]),
+              'maxLat': '{:.10f}'.format(limit[2]),
+              'maxLon': '{:.10f}'.format(limit[3]),
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getUparEleInRectByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_sounding_by_time_and_id(
+        times, data_code="UPAR_CHN_MUL_FTM", sta_ids='54511', ranges=None,
+        order=None, count=None, distinct=False, trans_type=True, 
+        elements="Station_Name,Station_Id_C,Lat,Lon,Alti,Year,Mon,Day,Hour,Min,Second,PRS_HWC,GPH,TEM,DPT,DTD,WIN_D"):
+    """
+    Retrieve sounding records from CMADaaS by times.
+    
+    Args:
+        times (str): times for retrieve, 'YYYYMMDDHHMISS,YYYYMMDDHHMISS,...'
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        sta_ids (str, optional): station id, 多个以逗号(,)分隔，区域站以字母开头. Defaults to 54511.
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        distinct (bool, optional): return unique records. Defaults to False.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+    
+    Returns:
+        pandas data frame: sounding records.
+    
+    Examples:
+        uparData = cmadaas_sounding_by_time_and_id('20220531000000', sta_ids='54511')
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'times': times,
+              'staIds': str(sta_ids).strip(),
+              'orderby': order if order is not None else "Datetime:ASC",
+              'elements': elements}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+    if distinct: params['distinct'] = "true"
+
+    # Interface, refer to
+    interface_id = "getUparEleByTimeAndStaID"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_sounding_in_rect_by_time_range(
+        time_range, limit=[3.51, 73.33, 53.33, 135.05], data_code="UPAR_CHN_MUL_FTM",
+        ranges=None, order=None, count=None, trans_type=True,
+        elements="Station_Name,Station_Id_C,Lat,Lon,Alti,Year,Mon,Day,Hour,Min,Second,PRS_HWC,GPH,TEM,DPT,DTD,WIN_D,WIN_S"):
+    """
+    Retrieve sounding records from CIMISS in region by times.
+    
+    Args:
+        time_range (str): time for retrieve, "[YYYYMMDDHHMISS,YYYYMMDDHHMISS]"
+        limit (list, optional):  map limits, [min_lat, min_lon, max_lat, max_lon]
+        data_code (str, optional): dataset code. Defaults to "UPAR_CHN_MUL_FTM".
+        ranges (str, optional): elements value ranges, seperated by ';'
+            range: (a,) is >a, [a,) is >=a, (,a) is <a, (,a] is <=a, (a,b) is >a & <b, 
+                   [a,b) is >=a & <b, (a,b] is >a & <=b, [a,b] is >=a & <=b
+            list: a,b,c;
+            e.g., "VIS:(,1000);RHU:(70,)", "Q_PRE_1h:0,3,4" is PRE quantity is credible.. Defaults to None.
+        order (str, optional): elements order, seperated by ',', like
+            "TEM:asc,SUM_PRE_1h:desc" is ascending order temperature first and descending PRE_1h. Defaults to None.
+        count (int, optional): the number of maximum returned records. Defaults to None.
+        trans_type (bool, optional): transform the return data frame's column type to datetime, numeric. Defaults to True.
+        elements (str, optional): elements for retrieve, 'ele1,ele2,...'. 
+    
+    Returns:
+        pandas data frame: sounding records.
+    
+    Examples:
+    >>> data = cmadaas_sounding_in_rect_by_time_range('[20220531000000,20220531010000]', limit=[3.51, 73.33, 53.33, 135.05])
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'elements': elements,
+              'timeRange': time_range,
+              'minLat': '{:.10f}'.format(limit[0]),
+              'minLon': '{:.10f}'.format(limit[1]),
+              'maxLat': '{:.10f}'.format(limit[2]),
+              'maxLon': '{:.10f}'.format(limit[3]),
+              'orderby': order if order is not None else "Datetime:ASC"}
+    if ranges is not None: params['eleValueRanges'] = ranges
+    if count is not None: params['limitCnt'] = str(count)
+
+    # interface id
+    interface_id = "getUparEleInRectByTimeRange"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    if trans_type: data = cmadaas_obs_convert_type(data)
+
+    # return
+    return data
+
+
+def cmadaas_analysis_by_time(time, limit=None, data_code='NAFP_CLDAS2.0_NRT_ASI_NC',
+                             levattrs={'long_name':'Height above Ground', 'units':'m'}, level_type='-',
+                             fcst_level=None, fcst_ele="TMP", zoom=None, varname='data', units=None, 
+                             scale_off=None, cache=True, cache_clear=True):
+    """
+    Retrieve CLDAS analysis data from CMADaaS service.
+
+    :param time: analysis time, like "20160817120000", format: YYYYMMDDHHMISS
+    :param limit: [min_lat, min_lon, max_lat, max_lon]
+    :param data_code: MUSIC data code, default is "NAFP_CLDAS2.0_NRT_ASI_NC"
+                      CLDAS2.0近实时数据产品(NC)-亚洲区域. Others like:
+                      NAFP_HRCLDAS_CHN_1KM_RT, HRCLDAS中国0.01°x0.01°逐小时实时融合实况分析产品
+    :param fcst_level: vertical level, default is None.
+    :param level_type: vertical level type, default is -, 表示没有层次类型
+    :param fcst_ele: forecast element, default is 2m temperature "TAIR"
+    :param zoom: the zoom out integer > 1, like 2. 像元缩小倍数, 数值大于1, 1的整数倍.
+    :param varname: set variable name, default is 'data'
+    :param units: forecast element's units, defaults to retrieved units.
+    :param scale_off: [scale, offset], return values = values*scale + offset.
+    :param cache: cache retrieved data to local directory, default is True.
+    :return: xarray dataset.
+
+    Examples:
+    >>> data = cmadaas_analysis_by_time(
+            "20210123000000", data_code="NAFP_CLDAS2.0_NRT_ASI_NC", 
+            fcst_ele='TMP', units="C", scale_off=[1.0, -273.15],
+            cache=False)
+    >>> data = cmadaas_analysis_by_time(
+            "20210821020000", data_code="NAFP_HRCLDAS_CHN_1KM_RT", 
+            fcst_ele='TAIR', units="C",
+            scale_off=[1.0, -273.15], cache=False)
+    """
+    # check time_str type
+    if isinstance(time, datetime):
+        time_str = time.strftime("%Y%m%d%H%M%S")
+    elif isinstance(time, str):
+        time_str = time
+    else:
+        warnings.warn('The type of time is not supportted')
+        return None
+
+    # retrieve data from cached file
+    if cache:
+        directory = os.path.join(data_code, fcst_ele, str(fcst_level))
+        filename = time_str
+        if limit is not None:
+            filename = filename + '.' + str(limit).replace(" ","")
+        cache_file = CONFIG.get_cache_file(directory, filename, name="CMADaaS", cache_clear=cache_clear)
+        if cache_file.is_file():
+            with open(cache_file, 'rb') as f:
+                data = pickle.load(f)
+                return data
+
+    # set retrieve parameters
+    if limit is None:
+        params = {'dataCode': data_code,
+                  'time': time_str,
+                  'levelType': str(level_type).strip(),
+                  'fcstEle': fcst_ele}
+        if zoom is not None: params['zoomOut'] = str(zoom)
+        interface_id = 'getNafpAnaEleGridByTimeAndLevel'
+    else:
+        params = {'dataCode': data_code,
+                  'time': time_str,
+                  'minLat': '{:.10f}'.format(limit[0]),
+                  "minLon": '{:.10f}'.format(limit[1]),
+                  "maxLat": '{:.10f}'.format(limit[2]),
+                  "maxLon": '{:.10f}'.format(limit[3]),
+                  'levelType': str(level_type).strip(),
+                  'fcstEle': fcst_ele}
+        interface_id = 'getNafpAnaEleGridInRectByTimeAndLevel'
+    if fcst_level is not None:
+        params['fcstLevel'] = '{:d}'.format(fcst_level)
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # get time information
+    time = datetime.strptime(time_str, '%Y%m%d%H%M%S')
+    time = np.array([time], dtype='datetime64[ms]')
+
+    # extract coordinates and data
+    start_lat = float(contents['startLat'])
+    start_lon = float(contents['startLon'])
+    nlon = int(contents['lonCount'])
+    nlat = int(contents['latCount'])
+    dlon = float(contents['lonStep'])
+    dlat = float(contents['latStep'])
+    lon = start_lon + np.arange(nlon)*dlon
+    lat = start_lat + np.arange(nlat)*dlat
+    name = contents['fieldNames']
+    if units is None:
+        units = contents['fieldUnits']
+
+    # define coordinates and variables
+    time_coord = ('time', time)
+    lon_coord = ('lon', lon, {
+        'long_name':'longitude', 'units':'degrees_east',
+        '_CoordinateAxisType':'Lon', "axis": 'X'})
+    lat_coord = ('lat', lat, {
+        'long_name':'latitude', 'units':'degrees_north',
+        '_CoordinateAxisType':'Lat', 'axis': 'Y'})
+    if fcst_level is not None:
+        level_coord = ('level', np.array([fcst_level]), levattrs)
+    varattrs = {'long_name': name, 'units': units}
+
+    # construct xarray
+    data = np.array(contents['DS'], dtype=np.float32)
+    if scale_off is not None:
+        data = data * scale_off[0] + scale_off[1]
+    if fcst_level is None:
+        data = data[np.newaxis, ...]
+        data = xr.Dataset({
+            varname:(['time', 'lat', 'lon'], data, varattrs)},
+            coords={'time':time_coord, 'lat':lat_coord, 'lon':lon_coord})
+    else:
+        data = data[np.newaxis, np.newaxis, ...]
+        data = xr.Dataset({
+            varname:(['time', 'level', 'lat', 'lon'], data, varattrs)},
+            coords={'time':time_coord, 'level':level_coord, 'lat':lat_coord, 'lon':lon_coord})
+
+    # add attributes
+    data.attrs['Conventions'] = "CF-1.6"
+    data.attrs['Origin'] = 'CIMISS Server by MUSIC API'
+
+    # cache data
+    if cache:
+        with open(cache_file, 'wb') as f:
+            pickle.dump(data, f, protocol=pickle.HIGHEST_PROTOCOL)
+
+    # return data
+    return data
+
+
+def cmadaas_analysis_by_times(times_str, pbar=True, allExists=True, **kargs):
+    """
+    Retrieve multiple CLDAS analysis data from CMADaaS service.
+
+    :param times_str: analysis time string list, like ["20200208000000", "20200209000000"], format: YYYYMMDDHHMISS
+    :param allExists (boolean): all files should exist, or return None.
+    :param pbar (boolean): Show progress bar, default to False.
+    :param **kargs: key arguments passed to cmadaas_model_by_time function.
+    :return: data, xarray type
+
+    Examples:
+    >>> times = ["20200208000000", "20200209000000"]
+    >>> data_code = "NAFP_CLDAS2.0_RT_GRB"
+    >>> data = cmadaas_analysis_by_times(times, data_code=data_code)
+    """
+
+    dataset = []
+    if pbar:
+        tqdm_times_str = tqdm(times_str, desc=kargs['data_code'] + ": ")
+    else:
+        tqdm_times_str = times_str
+    for time_str in tqdm_times_str:
+        data = cmadaas_analysis_by_time(time_str, **kargs)
+        if data:
+            dataset.append(data)
+        else:
+            if allExists:
+                warnings.warn("{} doese not exists.".format(kargs['data_code']+'/'+time_str))
+                return None
+    
+    return xr.concat(dataset, dim='time')
+
+
+def cmadaas_get_model_latest_time(data_code="NAFP_FOR_FTM_HIGH_EC_ANEA", latestTime=24):
+    """
+    Get the latest run time of the model.
+    注意不是所有的模式都有获取最新起步时间的接口函数.
+    
+    Args:
+        data_code (str, optional): dataset code, like "NAFP_FOR_FTM_HIGH_EC_ANEA".
+        latestTime (int, optional): latestTime > 0, like 2 is return 
+                                    the latest time in 2 hours. Defaults to 24.
+    Returns:
+        datetime object, the latest time
+
+    Examples:
+    >>> print(cmadaas_get_model_latest_time(data_code="NAFP_FOR_FTM_HIGH_EC_ANEA", latestTime=24))
+        2020-03-11 12:00:00
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'latestTime': str(latestTime+8)}    # 由于服务器是世界时, 因此需要+8才能与当前时间比较.
+
+    # interface id
+    interface_id = "getNafpLatestTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # construct pandas DataFrame
+    data = pd.DataFrame(contents['DS'])
+    time = pd.to_datetime(data['Datetime'], format="%Y%m%d%H%M%S")
+
+    return time[0]
+
+
+def cmadaas_model_grid(data_code, init_time, valid_time, fcst_ele, fcst_level, level_type, 
+                       fcst_member=None, limit=None, varname='data', units=None, 
+                       scale_off=None, cache=True, cache_clear=True,
+                       levattrs={'long_name':'height_above_ground', 'units':'m', '_CoordinateAxisType':'Height'}):
+    """
+    Retrieve model grid data from CMADaaS service.
+    refer to: http://10.20.76.55/cimissapiweb/apidataclassdefine_list.action
+
+    :param data_code: MUSIC data code, 
+                      "NAFP_ECMF_C1D_GLB_FOR"(default): 欧洲中心数值预报产品-高分辨率C1D-全球
+                      "NAFP_ECMF_FTM_HIGH_ANEA_FOR": 欧洲中心数值预报产品-高分辨率C1D-亚洲地区
+                      ......
+    :param init_time: model run time, like "2016081712", or datetime object.
+    :param valid_time: forecast hour, like 0 or 6
+    :param fcst_ele: forecast element, like 2m temperature "TEM"
+    :param fcst_level: vertical level, like 0
+    :param level_type: forecast level type, 表示Grib数据中的层次类型, 可在云平台上查询.
+    :param fcst_member: ensemble forecast member, 集合预报成员标识. 如果数据是集合预报, 该变量必须设置.
+    :param limit: [min_lat, min_lon, max_lat, max_lon]
+    :param varname: set variable name, default is 'data'
+    :param units: forecast element's units, defaults to retrieved units.
+    :param scale_off: [scale, offset], return values = values*scale + offset.
+    :param cache: cache retrieved data to local directory, default is True.
+    :param levattrs: level attributes, like:
+                     {'long_name':'height_above_ground', 'units':'m', '_CoordinateAxisType':'Height'}, default
+                     {'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'}
+                     {'long_name':'geopotential_height', 'units':'gpm', '_CoordinateAxisType':'GeoZ'}
+                     refer to https://www.unidata.ucar.edu/software/netcdf-java/current/reference/CoordinateAttributes.html
+    :return: xarray dataset.
+
+    Examples:
+    >>> data = cmadaas_model_grid("NAFP_FOR_FTM_HIGH_EC_ANEA", "2021010512", 24, 'TEM', 850, 100, 
+                                  units="C", scale_off=[1.0, -273.15], limit=[10.,100,30,120.],
+                                  levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'},
+                                  cache=True)
+    >>> data = cmadaas_model_grid("NAFP_GRAPESREPS_FOR_FTM_DIS_CHN", "2022092400", 24, 'TEM', 2, 103, 
+                                  fcst_member=1, units="C", scale_off=[1.0, -273.15], 
+                                  limit=[10., 100, 30, 120.], cache=True)
+    """
+
+    # check initial time
+    if isinstance(init_time, datetime):
+        init_time_str = init_time.strftime("%Y%m%d%H")
+    elif isinstance(init_time, str):
+        init_time_str = init_time
+    else:
+        warnings.warn('The type of init_time is not supportted')
+        return None
+
+    # retrieve data from cached file
+    if cache:
+        directory = os.path.join(data_code, fcst_ele, str(fcst_level))
+        filename = init_time_str
+        if limit is not None:
+            filename = filename + '_' + str(limit).replace(" ","")
+        if fcst_member is not None:
+            filename = filename + '_' + str(fcst_member).replace(" ","")
+        filename = filename + '.' + str(valid_time).zfill(3)
+        cache_file = CONFIG.get_cache_file(directory, filename, name="CMADaaS", cache_clear=cache_clear)
+        if cache_file.is_file():
+            with open(cache_file, 'rb') as f:
+                data = pickle.load(f)
+                return data
+
+    # set retrieve parameters
+    if fcst_member is None:
+        if limit is None:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele}
+            interface_id = 'getNafpEleGridByTimeAndLevelAndValidtime'
+        else:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'minLat': '{:.10f}'.format(limit[0]),
+                    "minLon": '{:.10f}'.format(limit[1]),
+                    "maxLat": '{:.10f}'.format(limit[2]),
+                    "maxLon": '{:.10f}'.format(limit[3]),
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele}
+            interface_id = 'getNafpEleGridInRectByTimeAndLevelAndValidtime'
+    else:
+        if limit is None:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele,
+                    'fcstMember': '{:d}'.format(fcst_member)}
+            interface_id = 'getNafpEleGridByTimeAndLevelAndValidtimeAndFcstMember'
+        else:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'minLat': '{:.10f}'.format(limit[0]),
+                    "minLon": '{:.10f}'.format(limit[1]),
+                    "maxLat": '{:.10f}'.format(limit[2]),
+                    "maxLon": '{:.10f}'.format(limit[3]),
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele,
+                    'fcstMember': '{:d}'.format(fcst_member)}
+            interface_id = 'getNafpEleGridInRectByTimeAndLevelAndValidtimeAndFcstMember'
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # get time information
+    init_time = datetime.strptime(init_time_str, '%Y%m%d%H')
+    fhour = np.array([valid_time], dtype=np.float)
+    time = init_time + timedelta(hours=fhour[0])
+    init_time = np.array([init_time], dtype='datetime64[ms]')
+    time = np.array([time], dtype='datetime64[ms]')
+
+    # extract coordinates and data
+    start_lat = float(contents['startLat'])
+    start_lon = float(contents['startLon'])
+    nlon = int(contents['lonCount'])
+    nlat = int(contents['latCount'])
+    dlon = float(contents['lonStep'])
+    dlat = float(contents['latStep'])
+    lon = start_lon + np.arange(nlon)*dlon
+    lat = start_lat + np.arange(nlat)*dlat
+    name = contents['fieldNames']
+    if units is None:
+        units = contents['fieldUnits']
+        
+    # set missing fcst_level for like fcst_leve='-'
+    if type(fcst_level) == str:
+        fcst_level = 0
+
+    # define coordinates and variables
+    time_coord = ('time', time)
+    lon_coord = ('lon', lon, {
+        'long_name':'longitude', 'units':'degrees_east', '_CoordinateAxisType':'Lon'})
+    lat_coord = ('lat', lat, {
+        'long_name':'latitude', 'units':'degrees_north', '_CoordinateAxisType':'Lat'})
+    if fcst_level != 0:
+        level_coord = ('level', np.array([fcst_level]), levattrs)
+    if fcst_member is not None:
+        member_coord = ('realization', np.array([fcst_member]), {
+            'long_name':'realization', units:'1'})
+    varattrs = {'short_name': fcst_ele, 'long_name': name, 'units': units}
+
+    # construct xarray
+    data = np.array(contents['DS'], dtype=np.float32)
+    if scale_off is not None:
+        data = data * scale_off[0] + scale_off[1]
+    if fcst_member is None:
+        if fcst_level == 0:
+            data = data[np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['time', 'lat', 'lon'], data, varattrs)},
+                coords={'time':time_coord, 'lat':lat_coord, 'lon':lon_coord})
+        else:
+            data = data[np.newaxis, np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['time', 'level', 'lat', 'lon'], data, varattrs)},
+                coords={'time':time_coord, 'level':level_coord, 'lat':lat_coord, 'lon':lon_coord})
+    else:
+        if fcst_level == 0:
+            data = data[np.newaxis, np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['realization', 'time', 'lat', 'lon'], data, varattrs)},
+                coords={'realization':member_coord, 'time':time_coord,
+                        'lat':lat_coord, 'lon':lon_coord})
+        else:
+            data = data[np.newaxis, np.newaxis, np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['realization', 'time', 'level', 'lat', 'lon'], data, varattrs)},
+                coords={'realization':member_coord, 'time':time_coord, 'level':level_coord,
+                        'lat':lat_coord, 'lon':lon_coord})
+
+    # add time coordinates
+    data.coords['forecast_reference_time'] = init_time[0]
+    data.coords['forecast_period'] = ('time', fhour, {
+        'long_name':'forecast_period', 'units':'hour'})
+
+    # add attributes
+    data.attrs['Conventions'] = "CF-1.6"
+    data.attrs['Origin'] = 'CIMISS Server by MUSIC API'
+
+    # cache data
+    if cache:
+        with open(cache_file, 'wb') as f:
+            pickle.dump(data, f, protocol=pickle.HIGHEST_PROTOCOL)
+
+    # return data
+    return data
+
+
+def cmadaas_model_grids(data_code, init_time, valid_times, fcst_ele, fcst_level, level_type,
+                        allExists=True, pbar=False, **kargs):
+    """
+    Retrieve multiple valid time grids at the same initial time from CMADaaS service.
+    
+    :param data_code: MUSIC data code, 
+                      "NAFP_FOR_FTM_HIGH_EC_GLB"(default): 欧洲中心数值预报产品-高分辨率C1D-全球
+                      "NAFP_FOR_FTM_HIGH_EC_ASI": 欧洲中心数值预报产品-高分辨率C1D-亚洲地区
+                      "NAFP_FOR_FTM_HIGH_EC_ANEA": 欧洲中心数值预报产品-高分辨率C1D-东北亚地区
+                      ......
+    :param init_time: model run time, like "2016081712", or datetime object.
+    :param valid_times: forecast hours, like [0, 6, 12, 15, 18, ...]
+    :param fcst_ele: forecast element, like 2m temperature "TEF2"
+    :param fcst_level: vertical level, like 0
+    :param level_type: vertical level type, like 0
+    :param allExists (boolean): all files should exist, or return None.
+    :param pbar (boolean): Show progress bar, default to False.
+    :param **kargs: key arguments passed to cmadaas_model_grid function.
+
+    Examples:
+    >>> valid_times = [6*i for i in 0:13]
+    >>> data = cmadaas_model_grids("NAFP_FOR_FTM_HIGH_EC_ANEA", "2020021512", valid_times, 'TEM', 850, 100, units="C", scale_off=[1.0, -273.15], 
+                                   levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'})
+    """
+
+    dataset = []
+    tqdm_valid_times = tqdm(valid_times, desc=data_code + ": ") if pbar else valid_times
+    for valid_time in tqdm_valid_times:
+        data = cmadaas_model_grid(data_code, init_time, valid_time, fcst_ele, fcst_level, level_type, **kargs)
+        if data:
+            dataset.append(data)
+        else:
+            if allExists:
+                if isinstance(init_time, datetime):
+                    init_time_str = init_time.strftime("%Y%m%d%H")
+                else:
+                    init_time_str = init_time
+                warnings.warn("{} doese not exists.".format(data_code+'/'+init_time_str+'.'+str(valid_time).zfill(3)))
+                return None
+    
+    return xr.concat(dataset, dim='time')
+
+
+def cmadaas_model_points(data_code, init_time, valid_times, fcst_ele, fcst_level, level_type, points, **kargs):
+    """
+    Retrieve model point time series at the same initial time from CMADaaS service.
+    
+    :param data_code: MUSIC data code, 
+                      "NAFP_FOR_FTM_HIGH_EC_GLB"(default): 欧洲中心数值预报产品-高分辨率C1D-全球
+                      "NAFP_FOR_FTM_HIGH_EC_ASI": 欧洲中心数值预报产品-高分辨率C1D-亚洲地区
+                      "NAFP_FOR_FTM_HIGH_EC_ANEA": 欧洲中心数值预报产品-高分辨率C1D-东北亚地区
+                      ......
+    :param init_time: model run time, like "2016081712", or datetime object
+    :param valid_times: forecast hours, like [0, 6, 12, 15, 18, ...]
+    :param fcst_ele: forecast element, like 2m temperature "TEF2", temperature "TEM"
+    :param fcst_level: vertical level, like 0
+    :param level_type: vertical level type, like 100
+    :param points: dictionary, {'lon':[...], 'lat':[...]}.
+    :param **kargs: key arguments passed to cmadaas_model_grids function.
+
+    Examples:
+    >>> valid_times = [6*i for i in 0:13]
+    >>> points = {'lon':[116.3833, 110.0], 'lat':[39.9, 32]}
+    >>> data = cmadaas_model_points("NAFP_FOR_FTM_HIGH_EC_ANEA", "2020021512", valid_times, 'TEM', 850, 100, points,
+                                    units="C", scale_off=[1.0, -273.15], 
+                                    levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'})
+    """
+
+    data = cmadaas_model_grids(data_code, init_time, valid_times, fcst_ele, fcst_level, level_type, **kargs)
+    if data:
+        return data.interp(lon=('points', points['lon']), lat=('points', points['lat']))
+    else:
+        return None
+
+
+def cmadaas_model_3D_grid(data_code, init_time, valid_time, fcst_ele, fcst_levels, level_type, allExists=True, pbar=False, **kargs):
+    """
+    Retrieve multiple level grids at the same initial time from CMADaaS service.
+    
+    :param data_code: MUSIC data code, 
+                      "NAFP_FOR_FTM_HIGH_EC_GLB"(default): 欧洲中心数值预报产品-高分辨率C1D-全球
+                      "NAFP_FOR_FTM_HIGH_EC_ASI": 欧洲中心数值预报产品-高分辨率C1D-亚洲地区
+                      "NAFP_FOR_FTM_HIGH_EC_ANEA": 欧洲中心数值预报产品-高分辨率C1D-东北亚地区
+                      ......
+    :param init_time: model run time, like "2016081712"
+    :param valid_time: forecast hour, like 0
+    :param fcst_ele: forecast element, like 2m temperature "TEF2", temperature "TEM"
+    :param fcst_levels: vertical levels, like [1000, 950, 925, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 100]
+    :param level_type: vertical level type, like 100
+    :param allExists (boolean): all files should exist, or return None.
+    :param pbar (boolean): Show progress bar, default to False.
+    :param **kargs: key arguments passed to cmadaas_model_grid function.
+
+    Examples:
+    >>> levels = [1000, 950, 925, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 100]
+    >>> data = cmadaas_model_3D_grid("NAFP_FOR_FTM_HIGH_EC_ANEA", "2020021512", 24, 'TEM', levels, 100, units="C", scale_off=[1.0, -273.15], 
+                                    levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'})
+    """
+
+    dataset = []
+    tqdm_fcst_levels = tqdm(fcst_levels, desc=data_code + ": ") if pbar else fcst_levels
+    for fcst_level in tqdm_fcst_levels:
+        data = cmadaas_model_grid(data_code, init_time, valid_time, fcst_ele, fcst_level, level_type, **kargs)
+        if data:
+            dataset.append(data)
+        else:
+            if allExists:
+                if isinstance(init_time, datetime):
+                    init_time_str = init_time.strftime("%Y%m%d%H")
+                else:
+                    init_time_str = init_time
+                warnings.warn("{} doese not exists.".format(data_code+'/'+init_time_str+'.'+str(valid_time).zfill(3)))
+                return None
+    
+    return xr.concat(dataset, dim='level')
+
+
+def cmadaas_model_3D_grids(data_code, init_time, valid_times, fcst_ele, fcst_levels, level_type, allExists=True, pbar=False, **kargs):
+    """
+    Retrieve multiple time and level grids at the same initial time from CMADaaS service.
+    
+    :param data_code: MUSIC data code, 
+                      "NAFP_FOR_FTM_HIGH_EC_GLB"(default): 欧洲中心数值预报产品-高分辨率C1D-全球
+                      "NAFP_FOR_FTM_HIGH_EC_ASI": 欧洲中心数值预报产品-高分辨率C1D-亚洲地区
+                      "NAFP_FOR_FTM_HIGH_EC_ANEA": 欧洲中心数值预报产品-高分辨率C1D-东北亚地区
+                      ......
+    :param init_time: model run time, like "2016081712", or datetime object
+    :param valid_times: forecast hour, like  [0, 6, 12, 15, 18, ...]
+    :param fcst_ele: forecast element, like 2m temperature "TEF2", temperature "TEM"
+    :param fcst_levels: vertical levels, like [1000, 950, 925, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 100]
+    :param level_type: vertical level type, like 100
+    :param allExists (boolean): all files should exist, or return None.
+    :param pbar (boolean): Show progress bar, default to False.
+    :param **kargs: key arguments passed to cmadaas_model_grid function.
+
+    Examples:
+    >>> valid_times = [6*i for i in range(13)]
+    >>> levels = [1000, 950, 925, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 100]
+    >>> data = cmadaas_model_3D_grids("NAFP_FOR_FTM_HIGH_EC_ANEA", "2020021512", 24, 'TEM', levels, 100, units="C", scale_off=[1.0, -273.15], 
+                                     levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'})
+    """
+
+    dataset = []
+    tqdm_valid_times = tqdm(valid_times, desc=data_code + ": ") if pbar else valid_times
+    for valid_time in tqdm_valid_times:
+        dataset_temp = []
+        for fcst_level in fcst_levels:
+            data = cmadaas_model_grid(data_code, init_time, valid_time, fcst_ele, fcst_level, level_type, **kargs)
+            if data:
+                dataset_temp.append(data)
+            else:
+                if allExists:
+                    if isinstance(init_time, datetime):
+                        init_time_str = init_time.strftime("%Y%m%d%H")
+                    else:
+                        init_time_str = init_time
+                    warnings.warn("{} doese not exists.".format(data_code+'/'+init_time_str+'.'+str(valid_time).zfill(3)))
+                    return None
+        dataset.append(xr.concat(dataset_temp, dim='level'))
+    
+    return xr.concat(dataset, dim='time')
+
+
+def cmadaas_model_profiles(data_code, init_time, valid_times, fcst_ele, fcst_levels, level_type, points, **kargs):
+    """
+    Retrieve time series of vertical profile from 3D [time, level, lat, lon] grids
+    at the same initial time from CMADaaS service.
+    
+    :param data_code: MUSIC data code, 
+                      "NAFP_FOR_FTM_HIGH_EC_GLB"(default): 欧洲中心数值预报产品-高分辨率C1D-全球
+                      "NAFP_FOR_FTM_HIGH_EC_ASI": 欧洲中心数值预报产品-高分辨率C1D-亚洲地区
+                      "NAFP_FOR_FTM_HIGH_EC_ANEA": 欧洲中心数值预报产品-高分辨率C1D-东北亚地区
+                      ......
+    :param init_time: model run time, like "2016081712", or datetime object
+    :param valid_times: forecast hour, like  [0, 6, 12, 15, 18, ...]
+    :param fcst_ele: forecast element, like 2m temperature "TEF2", temperature "TEM"
+    :param fcst_levels: vertical levels, like [1000, 950, 925, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 100]
+    :param level_type: vertical level type, like 100
+    :param point: dictionary, {'lon':[...], 'lat':[...]}.
+    :param **kargs: key arguments passed to cmadaas_model_grid function.
+
+    Examples:
+    >>> valid_times = [6*i for i in range(13)]
+    >>> levels = [1000, 950, 925, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 100]
+    >>> points = {'lon':[116.3833, 110.0], 'lat':[39.9, 32]}
+    >>> data = cmadaas_model_profiles("NAFP_FOR_FTM_HIGH_EC_ANEA", "2020021512", 24, 'TEM', levels, 100, points, units="C", scale_off=[1.0, -273.15], 
+                                     levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'})
+    """
+
+    data = cmadaas_model_3D_grids(data_code, init_time, valid_times, fcst_ele, fcst_levels, level_type, **kargs)
+    if data:
+        return data.interp(lon=('points', points['lon']), lat=('points', points['lat']))
+    else:
+        return None
+
+
+def cmadaas_model_by_time(init_time, valid_time=0, limit=None, fcst_member=None,
+                          data_code='NAFP_FOR_FTM_HIGH_EC_GLB', fcst_level=0, level_type=1, 
+                          levattrs={'long_name':'pressure_level', 'units':'hPa', '_CoordinateAxisType':'Pressure'},
+                          fcst_ele="TEM", varname='data', units=None, scale_off=None, cache=True, cache_clear=True):
+    """
+    Retrieve grid data from CMADaaS service.
+    与cmadass_model_grid功能相似, 只是接口方式不同(多了一些默认参数的设置), 
+    建议使用cmadass_model_grid为主.
+
+    :param init_time: model run time, like "2016081712", or datetime object.
+    :param valid_time: forecast hour, default is 0
+    :param limit: [min_lat, min_lon, max_lat, max_lon]
+    :param fcst_member: ensemble forecast member, 集合预报成员标识. 如果数据是集合预报, 该变量必须设置.
+    :param varname: set variable name, default is 'data'
+    :param data_code: MUSIC data code, default is "NAFP_FOR_FTM_HIGH_EC_GLB", 即EC高分辨率全球地面预报数据.
+    :param fcst_level: vertical level, default is 0, 表示地面数据, 可在云平台上查询.
+    :param level_type: forecast level type, default is 1, 表示Grib数据中的层次类型, 可在云平台上查询.
+    :param fcst_ele: forecast element, default is 2m temperature "TEF2"
+    :param units: forecast element's units, defaults to retrieved units.
+    :param scale_off: [scale, offset], return values = values*scale + offset.
+    :param cache: cache retrieved data to local directory, default is True.
+    :return: xarray dataset.
+
+    Examples:
+    >>> data = cmadaas_model_by_time("2021011512", data_code="NAFP_FOR_FTM_HIGH_EC_ANEA", 
+                                     fcst_level=850, level_type=100, fcst_ele='TEM', units="C",
+                                     scale_off=[1.0, -273.15], cache=False)
+    """
+
+    # check initial time
+    if isinstance(init_time, datetime):
+        init_time_str = init_time.strftime("%Y%m%d%H")
+    else:
+        init_time_str = init_time
+
+    # retrieve data from cached file
+    if cache:
+        directory = os.path.join(data_code, fcst_ele, str(fcst_level))
+        filename = init_time_str
+        if limit is not None:
+            filename = filename + '_' + str(limit).replace(" ","")
+        if fcst_member is not None:
+            filename = filename + '_' + str(fcst_member).replace(" ","")
+        filename = filename + '.' + str(valid_time).zfill(3)
+        cache_file = CONFIG.get_cache_file(directory, filename, name="CMADaaS", cache_clear=cache_clear)
+        if cache_file.is_file():
+            with open(cache_file, 'rb') as f:
+                data = pickle.load(f)
+                return data
+
+    # set retrieve parameters
+    if fcst_member is None:
+        if limit is None:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele}
+            interface_id = 'getNafpEleGridByTimeAndLevelAndValidtime'
+        else:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'minLat': '{:.10f}'.format(limit[0]),
+                    "minLon": '{:.10f}'.format(limit[1]),
+                    "maxLat": '{:.10f}'.format(limit[2]),
+                    "maxLon": '{:.10f}'.format(limit[3]),
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele}
+            interface_id = 'getNafpEleGridInRectByTimeAndLevelAndValidtime'
+    else:
+        if limit is None:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele,
+                    'fcstMember': '{:d}'.format(fcst_member)}
+            interface_id = 'getNafpEleGridByTimeAndLevelAndValidtimeAndFcstMember'
+        else:
+            params = {'dataCode': data_code,
+                    'time': init_time_str + '0000',
+                    'minLat': '{:.10f}'.format(limit[0]),
+                    "minLon": '{:.10f}'.format(limit[1]),
+                    "maxLat": '{:.10f}'.format(limit[2]),
+                    "maxLon": '{:.10f}'.format(limit[3]),
+                    'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+                    'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+                    'validTime': '{:d}'.format(valid_time),
+                    'fcstEle': fcst_ele,
+                    'fcstMember': '{:d}'.format(fcst_member)}
+            interface_id = 'getNafpEleGridInRectByTimeAndLevelAndValidtimeAndFcstMember'
+
+    
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # get time information
+    init_time = datetime.strptime(init_time_str, '%Y%m%d%H')
+    fhour = np.array([valid_time], dtype=np.float)
+    time = init_time + timedelta(hours=fhour[0])
+    init_time = np.array([init_time], dtype='datetime64[ms]')
+    time = np.array([time], dtype='datetime64[ms]')
+
+    # extract coordinates and data
+    start_lat = float(contents['startLat'])
+    start_lon = float(contents['startLon'])
+    nlon = int(contents['lonCount'])
+    nlat = int(contents['latCount'])
+    dlon = float(contents['lonStep'])
+    dlat = float(contents['latStep'])
+    lon = start_lon + np.arange(nlon)*dlon
+    lat = start_lat + np.arange(nlat)*dlat
+    name = contents['fieldNames']
+    if units is None:
+        units = contents['fieldUnits']
+        
+    # set missing fcst_level for fcst_leve='-'
+    if type(fcst_level) == str:
+        fcst_level = 0
+
+    # define coordinates and variables
+    time_coord = ('time', time)
+    lon_coord = ('lon', lon, {
+        'long_name':'longitude', 'units':'degrees_east',
+        '_CoordinateAxisType':'Lon', 'axis':'X'})
+    lat_coord = ('lat', lat, {
+        'long_name':'latitude', 'units':'degrees_north', '_CoordinateAxisType':'Lat'})
+    if fcst_level != 0:
+        level_coord = ('level', np.array([fcst_level]), levattrs)
+    if fcst_member is not None:
+        member_coord = ('realization', np.array([fcst_member]), {
+            'long_name':'realization', units:'1'})
+    varattrs = {'long_name': name, 'units': units}
+
+    # construct xarray
+    data = np.array(contents['DS'], dtype=np.float32)
+    if scale_off is not None:
+        data = data * scale_off[0] + scale_off[1]
+    if fcst_member is None:
+        if fcst_level == 0:
+            data = data[np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['time', 'lat', 'lon'], data, varattrs)},
+                coords={'time':time_coord, 'lat':lat_coord, 'lon':lon_coord})
+        else:
+            data = data[np.newaxis, np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['time', 'level', 'lat', 'lon'], data, varattrs)},
+                coords={'time':time_coord, 'level':level_coord, 'lat':lat_coord, 'lon':lon_coord})
+    else:
+        if fcst_level == 0:
+            data = data[np.newaxis, np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['realization', 'time', 'lat', 'lon'], data, varattrs)},
+                coords={'realization':member_coord, 'time':time_coord,
+                        'lat':lat_coord, 'lon':lon_coord})
+        else:
+            data = data[np.newaxis, np.newaxis, np.newaxis, ...]
+            data = xr.Dataset({
+                varname:(['realization', 'time', 'level', 'lat', 'lon'], data, varattrs)},
+                coords={'realization':member_coord, 'time':time_coord, 'level':level_coord,
+                        'lat':lat_coord, 'lon':lon_coord})
+
+    # add time coordinates
+    data.coords['forecast_reference_time'] = init_time[0]
+    data.coords['forecast_period'] = ('time', fhour, {
+        'long_name':'forecast_period', 'units':'hour'})
+
+    # add attributes
+    data.attrs['Conventions'] = "CF-1.6"
+    data.attrs['Origin'] = 'CIMISS Server by MUSIC API'
+
+    # cache data
+    if cache:
+        with open(cache_file, 'wb') as f:
+            pickle.dump(data, f, protocol=pickle.HIGHEST_PROTOCOL)
+
+    # return data
+    return data
+
+
+def cmadaas_model_by_times(init_time, valid_times=np.arange(0, 75, 6), pbar=True, allExists=True, **kargs):
+    """
+    Retrieve multiple model grids from CMADaaS service.
+
+    :param init_time: model run time, like "2016081712", or datetime object.
+    :param limit: [min_lat, min_lon, max_lat, max_lon]
+    :param valid_times: forecast hours, default is [0, 6, 12, ..., 72]
+    :param allExists (boolean): all files should exist, or return None.
+    :param pbar (boolean): Show progress bar, default to False.
+    :param **kargs: key arguments passed to cmadaas_model_by_time function.
+    :return: xarray dataset.
+
+    Examples:
+    >>> data = cmadaas_model_by_times("2020021512", data_code="NAFP_FOR_FTM_HIGH_EC_ANEA", time_range=[0, 72], 
+                                     fcst_level=850, fcst_ele='TEM', units="C", scale_off=[1.0, -273.15])
+    """
+
+    # check time
+    if isinstance(init_time, datetime):
+        init_time_str = init_time.strftime("%Y%m%d%H")
+    else:
+        init_time_str = init_time
+
+    dataset = []
+    tqdm_valid_times = tqdm(valid_times, desc=kargs['data_code'] + ": ") if pbar else valid_times
+    for valid_time in tqdm_valid_times:
+        data = cmadaas_model_by_time(init_time_str, valid_time=valid_time, **kargs)
+        if data:
+            dataset.append(data)
+        else:
+            if allExists:
+                warnings.warn("{} doese not exists.".format(kargs['data_code']+'/'+init_time_str))
+                return None
+    
+    return xr.concat(dataset, dim='time')
+
+
+def cmadaas_model_by_pionts(init_time_str, data_code='NAFP_FOR_FTM_HIGH_EC_ANEA',
+                            fcst_level=850, level_type=100, time_range=[0, 72], 
+                            points="39.90/116.40", fcst_ele="TEM"):
+    """
+    Retrieve grid point data from CMADaaS service.
+
+    :param init_time_str: model run time, like "2020020600"
+    :param data_code: MUSIC data code, default is "NAFP_FOR_FTM_HIGH_EC_ANEA"
+    :param fcst_level: vertical level, default is 850.
+    :param level_type: forecast level type, default=100, 表示Grib数据中的层次类型, 可在云平台上查询.
+    :param time_range: [minimum, maximum] forecast hour, default is [0, 72]
+    :param points: point location "latitude/longitude", also support
+                   multiple points like "39.90/116.40,32.90/112.40"
+    :param fcst_ele: forecast element, default is temperature "TEM"
+    :return: pandas dataframe
+    """
+
+    # set retrieve parameters
+    params = {'dataCode': data_code,
+              'time': init_time_str + '0000',
+              'fcstLevel': fcst_level if type(fcst_level) == str else '{:d}'.format(fcst_level),
+              'levelType': level_type if type(level_type) == str else '{:d}'.format(level_type),
+              'minVT': '{:d}'.format(time_range[0]),
+              'maxVT': '{:d}'.format(time_range[1]),
+              'latLons': points,
+              'fcstEle': fcst_ele}
+    interface_id = 'getNafpEleAtPointByTimeAndLevelAndValidtimeRange'
+
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # convert to numeric
+    data = pd.DataFrame(contents['DS'])
+    data['Lat'] = pd.to_numeric(data['Lat'])
+    data['Lon'] = pd.to_numeric(data['Lon'])
+    data['Validtime'] = pd.to_datetime(data['Validtime'], format="%Y%m%d%H%M%S")
+    data[fcst_ele] = pd.to_numeric(data[fcst_ele])
+
+    return data
+
+
+def cmadaas_model_by_piont_levels(init_time, fcst_levels=[1000, 925, 850, 700, 500],
+                                  pbar=True, **kargs):
+    """
+    Retrieve grid point data from CMADaaS service.
+
+    :param init_time: model run time, like "2020020600", or datetime object
+    :param data_code: MUSIC data code, default is "NAFP_FOR_FTM_HIGH_EC_ANEA"
+    :param fcst_levels: vertical levels, list like [1000, 950, 925, ...]
+    :param time_range: [minimum, maximum] forecast hour, default is [0, 72]
+    :param point: point location "latitude/longitude"
+    :param fcst_ele: forecast element, default is temperature "TEM"
+    :return: pandas dataframe
+    """
+
+    if isinstance(init_time, datetime):
+        init_time_str = init_time.strftime("%Y%m%d%H")
+    else:
+        init_time_str = init_time
+
+    # loop every level
+    tqdm_fcst_levels = tqdm(fcst_levels, desc=kargs['data_code'] + ": ") if pbar else fcst_levels
+    data = None
+    for fcst_level in tqdm_fcst_levels:
+        temp = cmadaas_model_by_pionts(init_time_str, fcst_level=fcst_level, **kargs)
+        if temp is None:
+            return None
+        
+        temp['level'] = fcst_level
+        if data is None:
+            data = temp
+        else:
+            data = pd.concat([data, temp])
+
+    data = data.pivot(index='Validtime', columns='level',values=kargs['fcst_ele'])
+    data = xr.DataArray(data, coords=[data.index.values, data.columns.values],
+                        dims=['time', 'level'], name=kargs['fcst_ele'])
+    data = data.loc[{'level':sorted(data.coords['level'].values, reverse=True)}]
+    data = data.loc[{'time':sorted(data.coords['time'].values)}]
+
+    return data
+
+
+def cmadaas_get_model_file(time, data_code="NAFP_FOR_FTM_HIGH_EC_ANEA", fcst_ele=None,
+                           level_type='100', out_dir=None, pbar=False, just_url=False):
+    """
+    Download numeric weather predication model file.
+    
+    注意cmadass设置了ip访问次数, 如果下载的文件数量过多, 则会返回"-5004 Reach the hor access line."
+    表示单位小时内检索过于频繁，请降低检索频次. 另外, 也要注意下载的文件大小, 如果文件只有几K，表示下载不成功, 删除重新下载.
+    
+    Args:
+        times (str): model initial time for retrieve, 
+                     single time 'YYYYMMDDHHMISS'; or
+                     time range, '[YYYYMMDDHHMISS,YYYYMMDDHHMISS]'
+        data_code (str, optional): dataset code. Defaults to "SURF_CMPA_RT_NC".
+        fcst_ele (str, optional): focast element. Defaults to None.
+        out_dir (str, optional): download files to out_dir. if out_dir is None,
+                                 the cached directory will be used. Defaults to None.
+        pbar (bool, optional): Show progress bar, default to True.
+        just_url (bool, optional): if just_url = True, return url string array, no files are downloaded.
+
+    Returns:
+        the list of download files path.
+
+    Examples:
+    >>> out_files = cmadaas_get_model_file('20210113000000', just_url=True)
+    >>> out_files = cmadaas_get_model_file('20210113000000', fcst_ele='TEM', out_dir=".")
+    >>> out_files = cmadaas_get_model_file('[20210111000000,20210130000000]', fcst_ele='TEM', out_dir=".")
+    """
+    
+    # check initial time
+    if isinstance(time, datetime):
+        time_str = time.strftime("%Y%m%d%H%M%S")
+    else:
+        time_str = time
+
+    if out_dir is None:
+        out_dir = CONFIG.get_cache_file(data_code, "", name="CMADaaS")
+    
+    time_str = time_str.strip()
+    if fcst_ele is None:
+        params = {'dataCode': data_code}
+        if time_str[0] == '[':
+            # set retrieve parameters
+            params['timeRange'] = time_str 
+            interface_id = "getNafpFileByTimeRange"
+        else:
+            # set retrieve parameters
+            params['time'] = time_str
+            interface_id = "getNafpFileByTime"
+    else:
+        params = {'dataCode': data_code,
+                  'fcstEle': fcst_ele.strip(),
+                  'levelType': str(level_type).strip()} 
+        if time_str[0] == '[':
+            # set retrieve parameters
+            params['timeRange'] = time_str
+            interface_id = "getNafpFileByElementAndTimeRange"
+        else:
+            # set retrieve parameters
+            params['time'] = time_str
+            interface_id = "getNafpFileByElementAndTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+
+    # just return the url
+    if just_url:
+        return contents['DS']
+
+    # loop every file and download
+    out_files = []
+    files = tqdm(contents['DS']) if pbar else contents['DS']
+    for file in files:
+        out_file = Path(out_dir) / file['FILE_NAME']
+        if not out_file.is_file():
+            urllib.request.urlretrieve(file['FILE_URL'], out_file)
+        out_files.append(out_file)
+
+    return out_files
+
+
+def cmadaas_get_model_file_with_filter(
+    time, data_code="NAFP_FOR_FTM_HIGH_EC_ANEA", 
+    filter=None, out_dir=None, pbar=False, just_url=False):
+    """
+    Download numeric weather predication model file.
+    与cmadaas_get_model_file函数不同, 本程序增加了对检索文件的过滤.
+    
+    Args:
+        times (str): model initial time for retrieve, 
+                     single time 'YYYYMMDDHHMISS'; or
+                     time range, '[YYYYMMDDHHMISS,YYYYMMDDHHMISS]'
+        data_code (str, optional): dataset code. Defaults to "SURF_CMPA_RT_NC".
+        filter (str, optional): filename's filter string. Defaults to None.
+        out_dir (str, optional): download files to out_dir. if out_dir is None,
+                                 the cached directory will be used. Defaults to None.
+        pbar (bool, optional): Show progress bar, default to True.
+        just_url (bool, optional): if just_url = True, return url string array, no files are downloaded.
+
+    Returns:
+        the list of download files path.
+
+    Examples:
+    >>> out_files = cmadaas_get_model_file_with_filter(
+        '20220920000000', data_code="NAFP_GRAPESREPS_FOR_FTM_DIS_CHN", 
+        filter="*_TEM_103_*_4_4.*", just_url=True)
+    """
+    
+    # check initial time
+    if isinstance(time, datetime):
+        time_str = time.strftime("%Y%m%d%H%M%S")
+    else:
+        time_str = time
+    time_str = time_str.strip()
+
+    if out_dir is None:
+        out_dir = CONFIG.get_cache_file(data_code, "", name="CMADaaS")
+    
+    params = {'dataCode': data_code}
+    if time_str[0] == '[':
+        # set retrieve parameters
+        params['timeRange'] = time_str 
+        interface_id = "getNafpFileByTimeRange"
+    else:
+        # set retrieve parameters
+        params['time'] = time_str
+        interface_id = "getNafpFileByTime"
+
+    # retrieve data contents
+    contents = get_rest_result(interface_id, params)
+    contents = _load_contents(contents)
+    if contents is None:
+        return None
+    
+    # filter contents
+    files = []
+    if filter is not None:
+        for file in contents['DS']:
+            if fnmatch.fnmatch(file['FILE_NAME'], filter):
+                files.append(file)
+    else:
+        files = contents['DS']
+    if len(files) < 1:
+        return None
+
+    # just return the url
+    if just_url:
+        return files
+
+    # loop every file and download
+    out_files = []
+    files = tqdm(files) if pbar else files
+    for file in files:
+        out_file = Path(out_dir) / file['FILE_NAME']
+        if not out_file.is_file():
+            urllib.request.urlretrieve(file['FILE_URL'], out_file)
+        out_files.append(out_file)
+
+    return out_files
+
+def cmadaas_get_radar_vol(starttime,endtime,sta_ids,data_code,outdir):
+    interface_id = 'getRadaFileByTimeRangeAndStaId'
+    starttime = datetime.strptime(starttime, '%Y%m%d%H%M%S')
+    endtime = datetime.strptime(endtime, '%Y%m%d%H%M%S')
+    curtime = starttime
+    allfilepaths=[]
+    allfilenames=[]
+    while curtime <= endtime:
+            
+        times = curtime.strftime('%Y%m%d%H0000')
+        tmpt = curtime + timedelta(hours=1) 
+        tmptstr = tmpt.strftime('%Y%m%d%H0000')
+        time_range = '[' + times + ',' + tmptstr + ']'
+        # time_range='[20180301000000,20201101000000]'
+        print(time_range)
+        for sta in sta_ids:
+            params = {'timeRange': time_range,
+                    'dataCode': data_code,
+                    'staIds': sta,
+                    'elements': 'FILE_NAME',
+                    }
+            contents = get_rest_result(interface_id, params, data_format='json')
+            if contents == None:
+                print('No Valid data: ' + sta)
+                continue
+            
+            json_dic1 = json.loads(contents, strict=False)
+            if json_dic1['returnCode'] == '0':
+                        
+                dd = json_dic1['DS']
+                http = urllib3.PoolManager()
+                for idx in range(len(dd)):
+        
+                    url = dd[idx]["FILE_URL"]
+                    filename = dd[idx]["FILE_NAME"]
+
+                    newoutpath = outdir + os.sep  + filename[15:19] +\
+                        os.sep + filename[15:23] + os.sep+ sta
+                    if not os.path.exists(newoutpath):
+                        os.makedirs(newoutpath)
+                    if not os.path.exists( newoutpath + os.sep + filename):
+                        # 如果不存在，才下载
+                        req = http.request("GET",url)
+                        if req.status != 200:
+                            print('Can not access the url: ' + url)
+                            
+                        else:
+                            fout = open(newoutpath + os.sep + filename, 'wb')
+                            fout.write(req.data)
+                            fout.close()
+                            print('Download succeed: ' + filename)
+                            allfilepaths.append(newoutpath)
+                            allfilenames.append(filename)
+                    else:
+                        print(filename + ' already exists!')
+                        allfilepaths.append(newoutpath)
+                        allfilenames.append(filename)
+            else:
+                print('No Valid Record!  ' + sta)    
+                print(json_dic1)
+        curtime += timedelta(hours=1) 
+    return allfilepaths,allfilenames
+
+#从天擎下载探测中心制作的三维反射率因子数据
+def cmadaas_get_radar_3dref(starttime,endtime,data_code,outdir):
+    '''
+    根据时间段检索雷达数据 interface_id = getRadaFileByTimeRange
+    data_code = RADA_L3_MST_PRE_3REF_QC
+    '''
+    interface_id = 'getRadaFileByTimeRange'
+    starttime = datetime.strptime(starttime, '%Y%m%d%H%M%S')
+    endtime = datetime.strptime(endtime, '%Y%m%d%H%M%S')
+    curtime = starttime
+    allfilepaths=[]
+    allfilenames=[]
+    while curtime < endtime:
+            
+        times = curtime.strftime('%Y%m%d%H0000')
+        tmpt = curtime + timedelta(hours=1) 
+        tmptstr = tmpt.strftime('%Y%m%d%H0000')
+        time_range = '[' + times + ',' + tmptstr + ']'
+        # time_range='[20180301000000,20201101000000]'
+        print(time_range)
+        # 逐6分钟进行检测，如果已经存在了就不再查询
+        sst = curtime
+        ett = tmpt
+        cst = sst
+        bneed_download=False
+        while cst < ett:
+            # cpath = outdir + os.sep + cst.strftime('%Y') + os.sep + cst.strftime('%Y%m%d')
+            cpath = outdir
+            if not os.path.exists(cpath):
+                bneed_download = True
+                break
+            else:
+                cfile = cpath + os.sep + cst.strftime('ACHN_CAP_%Y%m%d_%H%M00') + '_3d.nc'
+                if not os.path.exists(cfile):   
+                    cfile2 = cpath + os.sep + cst.strftime('ACHN_CAP_%Y%m%d_%H%M00') + '_23.bin'
+                    if not os.path.exists(cfile2):
+                        cfile3 = cpath + os.sep + cst.strftime('ACHN_CAP_%Y%m%d_%H%M00') + '_23.nc'
+                        if not os.path.exists(cfile3):
+                            bneed_download = True
+                            break
+                        else:
+                            allfilepaths.append(cpath)
+                    else:
+                        allfilepaths.append(cpath)
+                        
+                
+                    
+            cst += timedelta(minutes=6)
+        if bneed_download:
+            params = {'timeRange': time_range,
+                    'dataCode': data_code,
+                    'elements': 'FILE_NAME',
+                    }
+            contents = get_rest_result(interface_id, params, data_format='json')
+            if contents == None:
+                print('No Valid data ')
+                curtime += timedelta(hours=1) 
+                continue
+            
+            json_dic1 = json.loads(contents, strict=False)
+            if json_dic1['returnCode'] == '0':
+                        
+                dd = json_dic1['DS']
+                http = urllib3.PoolManager()
+                for idx in range(len(dd)):
+        
+                    url = dd[idx]["FILE_URL"]
+                    filename = dd[idx]["FILE_NAME"]
+
+                    newoutpath = outdir + os.sep  + filename[44:48] +\
+                        os.sep + filename[44:52]
+                    if not os.path.exists(newoutpath):
+                        os.makedirs(newoutpath)
+                    outname_3d = filename[35:60] + '3d.nc'
+                    outname_single = filename[35::]
+                    if  os.path.exists( newoutpath + os.sep + outname_3d):
+                        print(outname_3d + ' already exists!')
+                        # allfilepaths.append(newoutpath)
+                        # allfilenames.append(filename)
+                    elif  os.path.exists( newoutpath + os.sep + outname_single):
+                        print(outname_single + ' already exists!')
+                        allfilepaths.append(newoutpath)
+                        allfilenames.append(outname_single)
+                    else:
+                        # 如果不存在，才下载
+                        req = http.request("GET",url)
+                        if req.status != 200:
+                            print('Can not access the url: ' + url)
+                        else:
+                            fout = open(newoutpath + os.sep + outname_single, 'wb')
+                            fout.write(req.data)
+                            fout.close()
+                            print('Download succeed: ' + outname_single)
+                            allfilepaths.append(newoutpath)
+                            allfilenames.append(outname_single)
+                
+        else: 
+            # print(json_dic1)
+            print('no need to download')
+        curtime += timedelta(hours=1) 
+    return allfilepaths,allfilenames