openforis-whisp 3.0.0a7__py3-none-any.whl → 3.0.0a9__py3-none-any.whl

openforis_whisp/datasets.py

@@ -373,15 +373,25 @@ def g_esri_2020_2023_crop_prep():
 #### disturbances by year
 
 # RADD_year_2019 to RADD_year_< current year >
+# Coverage: Primary humid tropical forest areas of South America, sub-Saharan Africa,
+#           and insular Southeast Asia at 10m spatial resolution.
+#           Available from January 2019 to present for Africa,
+#           and from January 2020 to present for South America and Southeast Asia.
 def g_radd_year_prep():
-    from datetime import datetime
-
+    """
+    RADD alerts per year as multiband image.
+    Each band is binary (1 = alert detected).
+    Coverage: Primary humid tropical forest areas of South America, sub-Saharan Africa,
+              and insular Southeast Asia at 10m spatial resolution.
+              Available from January 2019 to present for Africa,
+              and from January 2020 to present for South America and Southeast Asia.
+    """
     radd = ee.ImageCollection("projects/radar-wur/raddalert/v1")
     radd_date = (
         radd.filterMetadata("layer", "contains", "alert").select("Date").mosaic()
     )
     start_year = 19
-    current_year = datetime.now().year % 100
+    current_year = CURRENT_YEAR_2DIGIT
 
     def make_band(year, img_stack):
         start = year * 1000
@@ -584,27 +594,412 @@ def g_radd_before_2020_prep():
     ).selfMask()
 
 
-# # DIST_after_2020
-# # alerts only for after 2020 currently so need to use date
-# def glad_dist_after_2020_prep():
+# DIST_after_2020
+
+# DIST alerts are for all veg types so masked by EUFO forest 2020
+# NB alerts only for 2024 onwards (in GEE at least, available for 2023 ofrom the GLAD site)
+# for conistency using "...after_2020..." terminology.
+def g_glad_dist_after_2020_prep():
+
+    # no need to filter by date as all dates are later than 2023
+
+    # Load the vegetation disturbance collections
+    VEGDISTSTATUS = ee.ImageCollection(
+        "projects/glad/HLSDIST/current/VEG-DIST-STATUS"
+    ).mosaic()
+
+    # Key for high-confidence alerts (values 3, 6, 7, 8)
+    high_conf_values = [3, 6, 7, 8]
+
+    # Create high-confidence mask
+    dist_high_conf = VEGDISTSTATUS.remap(
+        high_conf_values, [1] * len(high_conf_values), 0
+    )
+
+    return dist_high_conf.updateMask(g_jrc_gfc_2020_prep()).rename(
+        "DIST_after_2020"
+    )  # Mask alerts to forest and rename band
+
+
+# # DIST_alert_2024 to DIST_alert_< current year >
+# # Notes:
+# # 1) so far only available for 2024 onwards in GEE
+# # 2) masked alerts (as dist alerts are for all vegetation) to JRC EUFO 2020 layer, as close to EUDR definition
+
+
+def g_glad_dist_year_prep():
+    """
+    GLAD DIST alerts per year as multiband image.
+    Each band is binary (1 = high-confidence disturbance alert).
+    Uses VEG-DIST-DATE to filter by year, VEG-DIST-STATUS for confidence.
+    Masked to EUFO 2020 forest.
+    Note: Only available from 2024 onwards.
+    Fully server-side using ee.List.iterate (no Python for loop).
+    """
+    # Load the vegetation disturbance collections
+    #  Vegetation disturbance status (0-8, class flag, 8-bit)
+    VEGDISTSTATUS = ee.ImageCollection(
+        "projects/glad/HLSDIST/current/VEG-DIST-STATUS"
+    ).mosaic()
+    # Initial vegetation disturbance date (>0: days since 2020-12-31, 16-bit)
+    VEGDISTDATE = ee.ImageCollection(
+        "projects/glad/HLSDIST/current/VEG-DIST-DATE"
+    ).mosaic()
+
+    # Key for high-confidence alerts (values 3, 6, 7, 8)
+    # 3 = <50% loss, high confidence, ongoing
+    # 6 = ≥50% loss, high confidence, ongoing
+    # 7 = <50% loss, high confidence, finished
+    # 8 = ≥50% loss, high confidence, finished
+    high_conf_values = [3, 6, 7, 8]
+    dist_high_conf = VEGDISTSTATUS.remap(
+        high_conf_values, [1] * len(high_conf_values), 0
+    )
+
+    # Year range: 2024 to current year
+    start_year = 2024
+    end_year = CURRENT_YEAR
+
+    # Reference date for day offset calculation (2020-12-31)
+    ref_date = ee.Date("2020-12-31")
+
+    # Create first band (2024)
+    first_year = ee.Number(start_year)
+    first_start_days = ee.Date.fromYMD(first_year, 1, 1).difference(ref_date, "day")
+    first_end_days = ee.Date.fromYMD(first_year.add(1), 1, 1).difference(
+        ref_date, "day"
+    )
+    first_year_mask = VEGDISTDATE.gte(first_start_days).And(
+        VEGDISTDATE.lt(first_end_days)
+    )
+    first_band_name = ee.String("DIST_year_").cat(first_year.format("%d"))
+    first_band = (
+        first_year_mask.updateMask(dist_high_conf).rename(first_band_name).selfMask()
+    )
+
+    # Server-side iteration to add remaining years
+    years = ee.List.sequence(start_year + 1, end_year)
+
+    def add_year_band(year, img_stack):
+        year_num = ee.Number(year)
+        start_days = ee.Date.fromYMD(year_num, 1, 1).difference(ref_date, "day")
+        end_days = ee.Date.fromYMD(year_num.add(1), 1, 1).difference(ref_date, "day")
+        year_mask = VEGDISTDATE.gte(start_days).And(VEGDISTDATE.lt(end_days))
+        band_name = ee.String("DIST_year_").cat(year_num.format("%d"))
+        year_band = year_mask.updateMask(dist_high_conf).rename(band_name).selfMask()
+        return ee.Image(img_stack).addBands(year_band)
+
+    img_stack = ee.Image(years.iterate(add_year_band, first_band))
+
+    # Mask to EUFO 2020 forest
+    return img_stack.updateMask(g_jrc_gfc_2020_prep())
+
+
+# GLAD-L (GLAD Landsat) Alerts
+# Coverage: Entire tropics (30°N to 30°S) from January 1, 2018 to present,
+#           and from 2015 to present for select countries in the Amazon, Congo Basin,
+#           and insular Southeast Asia.
+# Uses confidence bands per year where values >= 2 are confirmed alerts
+# Asset paths per year:
+#   2021: projects/glad/alert/2021final (conf21)
+#   2022: projects/glad/alert/2022final (conf22)
+#   2023: projects/glad/alert/2023final (conf23)
+#   2024: NOT AVAILABLE
+#   2025+: projects/glad/alert/UpdResult (conf25, conf26, etc.)
+# More info: https://glad.umd.edu/dataset/glad-forest-alerts
+
+
+# GLAD-L_after_2020 (combined alerts from 2021 to current year, excluding 2024)
+def g_glad_l_after_2020_prep():
+    """
+    GLAD Landsat alerts after 2020 (combined from 2021 onwards).
+    Uses confidence bands with threshold >= 2 for confirmed alerts.
+    Note: 2024 data is not available.
+    """
+    # Load yearly assets and combine into single multiband image
+    glad_combined = (
+        ee.ImageCollection("projects/glad/alert/2021final")
+        .mosaic()
+        .select("conf21")
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2022final")
+            .mosaic()
+            .select("conf22")
+        )
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2023final")
+            .mosaic()
+            .select("conf23")
+        )
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/UpdResult")
+            .mosaic()
+            .select(["conf25", "conf26"])
+        )
+    )
+
+    # Combine alerts from all available years (confidence >= 2)
+    # 2024 not available
+    combined_alerts = (
+        glad_combined.select("conf21")
+        .gte(2)
+        .Or(glad_combined.select("conf22").gte(2))
+        .Or(glad_combined.select("conf23").gte(2))
+        .Or(glad_combined.select("conf25").gte(2))
+        .Or(glad_combined.select("conf26").gte(2))
+    )
+
+    return combined_alerts.rename("GLAD-L_after_2020").selfMask()
+
 
-#     # Load the vegetation disturbance collections
-#     VEGDISTSTATUS = ee.ImageCollection(
-#         "projects/glad/HLSDIST/current/VEG-DIST-STATUS"
-#     ).mosaic()
+# GLAD-L_before_2020 (combined alerts from 2017 to 2020)
+def g_glad_l_before_2020_prep():
+    """
+    GLAD Landsat alerts before 2020 (combined from 2017-2020 inclusive).
+    Uses confidence bands with threshold >= 2 for confirmed alerts.
+    Note: 2015 and 2016 assets are not available in GEE.
+    Coverage: Tropics (30°N to 30°S).
+    """
+    # Load yearly assets and combine
+    glad_combined = (
+        ee.ImageCollection("projects/glad/alert/2017final")
+        .mosaic()
+        .select("conf17")
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2018final")
+            .mosaic()
+            .select("conf18")
+        )
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2019final")
+            .mosaic()
+            .select("conf19")
+        )
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2020final")
+            .mosaic()
+            .select("conf20")
+        )
+    )
 
-#     # Key for high-confidence alerts (values 3, 6, 7, 8)
-#     high_conf_values = [3, 6, 7, 8]
+    # Combine alerts from all available years (confidence >= 2)
+    combined_alerts = (
+        glad_combined.select("conf17")
+        .gte(2)
+        .Or(glad_combined.select("conf18").gte(2))
+        .Or(glad_combined.select("conf19").gte(2))
+        .Or(glad_combined.select("conf20").gte(2))
+    )
 
-#     # Create high-confidence mask
-#     dist_high_conf = VEGDISTSTATUS.remap(
-#         high_conf_values, [1] * len(high_conf_values), 0
-#     )
+    return combined_alerts.rename("GLAD-L_before_2020").selfMask()
 
-#     return dist_high_conf.updateMask(jrc_gfc_2020_prep()).rename(
-#         "DIST_after_2020"
-#     )  # Mask alerts to forest and rename band
 
+# GLAD-L timeseries - multiband image with one band per year
+def g_glad_l_year_prep():
+    """
+    GLAD Landsat alerts per year as multiband image.
+    Each band is binary (1 = alert with confidence >= 2).
+    Coverage: Entire tropics (30°N to 30°S) from January 1, 2018 to present,
+              and from 2017 to present for select countries in the Amazon,
+              Congo Basin, and insular Southeast Asia.
+    Note: 2015 and 2016 assets are not available in GEE.
+    Note: 2024 data is not available.
+    Includes years from 2017 onwards.
+    """
+    # Build multiband image with all available years
+    # Years 2017-2023 use YYYYfinal assets, 2025+ use UpdResult
+    # Note: 2015final and 2016final assets do not exist in GEE
+    img_stack = (
+        # 2017
+        ee.ImageCollection("projects/glad/alert/2017final")
+        .mosaic()
+        .select("conf17")
+        .gte(2)
+        .rename("GLAD-L_year_2017")
+        .selfMask()
+        # 2018
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2018final")
+            .mosaic()
+            .select("conf18")
+            .gte(2)
+            .rename("GLAD-L_year_2018")
+            .selfMask()
+        )
+        # 2019
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2019final")
+            .mosaic()
+            .select("conf19")
+            .gte(2)
+            .rename("GLAD-L_year_2019")
+            .selfMask()
+        )
+        # 2020
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2020final")
+            .mosaic()
+            .select("conf20")
+            .gte(2)
+            .rename("GLAD-L_year_2020")
+            .selfMask()
+        )
+        # 2021
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2021final")
+            .mosaic()
+            .select("conf21")
+            .gte(2)
+            .rename("GLAD-L_year_2021")
+            .selfMask()
+        )
+        # 2022
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2022final")
+            .mosaic()
+            .select("conf22")
+            .gte(2)
+            .rename("GLAD-L_year_2022")
+            .selfMask()
+        )
+        # 2023
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/2023final")
+            .mosaic()
+            .select("conf23")
+            .gte(2)
+            .rename("GLAD-L_year_2023")
+            .selfMask()
+        )
+        # 2024 NOT AVAILABLE
+        # 2025
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/UpdResult")
+            .mosaic()
+            .select("conf25")
+            .gte(2)
+            .rename("GLAD-L_year_2025")
+            .selfMask()
+        )
+        # 2026
+        .addBands(
+            ee.ImageCollection("projects/glad/alert/UpdResult")
+            .mosaic()
+            .select("conf26")
+            .gte(2)
+            .rename("GLAD-L_year_2026")
+            .selfMask()
+        )
+    )
+
+    return img_stack
+
+
+# GLAD-S2 (GLAD Sentinel-2) Alerts
+# GLAD-S2_after_2020 (combined alerts from 2021 to current year)
+def g_glad_s2_after_2020_prep():
+    """
+    GLAD Sentinel-2 alerts after 2020 (filtered and combined from 2021 onwards - original data starts 2019).
+    Uses alert band with threshold >= 2 for confirmed alerts.
+    Coverage: Primary humid tropical forest within Amazon basin region.
+    https://glad.umd.edu/dataset/glad-forest-alerts
+    """
+    col = "projects/glad/S2alert"
+    s2alert = ee.Image(col + "/alert")
+    alert_date = ee.Image(col + "/alertDate")
+
+    # Date encoding: days since 2019-01-01
+    # 2020-12-31 = 730 days (end of 2020)
+    days_end_2020 = 730
+
+    # Filter alerts after 2020 with confidence >= 2
+    #   - alert: confidence band (0-4, where 4 is highest confidence)
+    alerts_after_2020 = s2alert.gte(2).And(alert_date.gt(days_end_2020))
+
+    return alerts_after_2020.rename("GLAD-S2_after_2020").selfMask()
+
+
+# GLAD-S2_before_2020 (combined alerts from 2019 to 2020)
+def g_glad_s2_before_2020_prep():
+    """
+    GLAD Sentinel-2 alerts before 2020 (from 2019-01-01 to 2020-12-31 inclusive).
+    Uses alert band with threshold >= 2 for confirmed alerts.
+    Coverage: Primary humid tropical forest within Amazon basin region.
+    Note: Data starts from 2019.
+    """
+    col = "projects/glad/S2alert"
+    s2alert = ee.Image(col + "/alert")
+    alert_date = ee.Image(col + "/alertDate")
+
+    # Date encoding: days since 2019-01-01
+    # 2019-01-01 = 0, 2020-12-31 = 730 days
+    days_end_2020 = 730
+
+    # Filter alerts up to end of 2020 with confidence >= 2
+    alerts_before_2020 = s2alert.gte(2).And(alert_date.lte(days_end_2020))
+
+    return alerts_before_2020.rename("GLAD-S2_before_2020").selfMask()
+
+
+# GLAD-S2 timeseries - multiband image with one band per year
+def g_glad_s2_year_prep():
+    """
+    GLAD Sentinel-2 alerts per year as multiband image.
+    Each band is binary (1 = alert with confidence >= 2).
+    Coverage: Primary humid tropical forest areas of South America
+              from January 2019 to present.
+    Date encoding: days since 2019-01-01.
+    Includes years from 2019 onwards.
+    """
+    col = "projects/glad/S2alert"
+    s2alert = ee.Image(col + "/alert")
+    alert_date = ee.Image(col + "/alertDate")
+
+    # Confidence threshold for confirmed alerts
+    confirmed = s2alert.gte(2)
+
+    # Date encoding: days since 2019-01-01
+    # Year boundaries (days since 2019-01-01):
+    # 2019-01-01 = 0, 2020-01-01 = 365, 2021-01-01 = 731, 2022-01-01 = 1096
+    # 2023-01-01 = 1461, 2024-01-01 = 1827, 2025-01-01 = 2192, 2026-01-01 = 2557
+    ref_date = ee.Date("2019-01-01")
+
+    # Build multiband image with available years (data starts 2019)
+    start_year = 2019
+    end_year = CURRENT_YEAR
+
+    # Create first band (2019)
+    first_year = ee.Number(start_year)
+    first_start_days = ee.Date.fromYMD(first_year, 1, 1).difference(ref_date, "day")
+    first_end_days = ee.Date.fromYMD(first_year.add(1), 1, 1).difference(
+        ref_date, "day"
+    )
+    first_year_mask = alert_date.gte(first_start_days).And(
+        alert_date.lt(first_end_days)
+    )
+    first_band_name = ee.String("GLAD-S2_year_").cat(first_year.format("%d"))
+    first_band = (
+        confirmed.updateMask(first_year_mask).rename(first_band_name).selfMask()
+    )
+
+    # Server-side iteration to add remaining years
+    years = ee.List.sequence(start_year + 1, end_year)
+
+    def add_year_band(year, img_stack):
+        year_num = ee.Number(year)
+        start_days = ee.Date.fromYMD(year_num, 1, 1).difference(ref_date, "day")
+        end_days = ee.Date.fromYMD(year_num.add(1), 1, 1).difference(ref_date, "day")
+        year_mask = alert_date.gte(start_days).And(alert_date.lt(end_days))
+        band_name = ee.String("GLAD-S2_year_").cat(year_num.format("%d"))
+        year_band = confirmed.updateMask(year_mask).rename(band_name).selfMask()
+        return ee.Image(img_stack).addBands(year_band)
+
+    img_stack = ee.Image(years.iterate(add_year_band, first_band))
+
+    return img_stack
+
+
+#### disturbances combined (split into before and after 2020)
 
 # TMF_deg_before_2020
 def g_tmf_deg_before_2020_prep():