wolfhece 1.8.7__py3-none-any.whl → 1.8.9__py3-none-any.whl

wolfhece/hydrology/PyWatershed.py

@@ -0,0 +1,1951 @@
+import os
+from scipy.interpolate import interpolate
+from scipy.spatial import KDTree
+from matplotlib import figure as mplfig
+from matplotlib.axes import Axes
+import matplotlib.pyplot as plt
+import logging
+from tqdm import tqdm
+
+from ..PyTranslate import _
+from ..PyVertex import wolfvertex, cloud_vertices
+from ..wolf_array import *
+from ..PyCrosssections import crosssections as CrossSections
+from ..GraphNotebook import PlotNotebook
+from .read import *
+
+LISTDEM=['dem_before_corr','dem_after_corr','dem_10m','dem_20m','crosssection']
+#LISTDEM=['dem_after_corr']
+
+class Node_Watershed:
+    """Noeud du modèle hydrologique maillé"""
+    i:int   #indice i dans la matrice
+    j:int   #indice j dans la matrice
+
+    x:float # coordonnée X
+    y:float # coordonnée Y
+
+    index:int  # Numérotation de la maille dans la liste de l'objet Watershed qui l'a initialisé
+
+    dem:dict[str,float] # dictionnaire des valeurs d'altitudes
+    demdelta:float      # correction apportée dans la phase de prépro
+
+    crosssections:list  # sections en travers sontenues dans la maille
+
+    time:float      # temps de propagation - cf Drainage_basin.time
+    slope:float     # pente calculée ne prépro
+    sloped8:float   # pente selon les 8 voisins
+
+    slopecorr:dict  # pente corrigée
+    demcorr:dict    # dictionnaire d'alrtitude corrigées
+
+    river:bool      # maille rivière => True
+    reach:int       # index du bief
+    sub:int         # inde du sous-bassin
+    forced:bool     # maille d'échange forcé
+    uparea:float    # surface drainée - cf Drainage_basin.cnv
+
+    strahler:int    # indice de Strahler - cf "create_index"
+    reachlevel:int  # Niveau du bief - cf "create_index"
+
+    cums:float      # longueur curviligne cumulée **depuis l'aval** -- cf "incr_curvi"
+    incrs:float     # incrément de longueur curvi - dx ou sqrt(2)*dx si voisin en crois ou en diagonale
+
+    down:"Node_Watershed"           # pointeur vers le noeud aval
+    up:list["Node_Watershed"]       # pointeurs vers le(s) noeud(s) amont
+    upriver:list["Node_Watershed"]  # pointeurs vers le(s) noeud(s) **rivière** amont
+
+    flatindex:int = -1  # index de la zone de plat
+
+    def incr_curvi(self):
+        """Incrémentation de la longueur curviligne"""
+
+        if self.down is None:
+            self.cums=0.
+        else:
+            self.cums = self.down.cums+self.incrs
+
+        for curup in self.up:
+            curup.incr_curvi()
+
+    def mean_slope_up(self, threshold:float)-> float:
+        """Pente moyenne sur depuis les mailles amont"""
+        curnode: Node_Watershed
+        meanslope=0.
+        nbmean=0
+        for curnode in self.up:
+            if curnode.slope>threshold:
+                nbmean+=1.
+                meanslope+=curnode.slope
+        if nbmean>0:
+            meanslope=meanslope/nbmean
+
+        return meanslope
+
+    def slope_down(self, threshold:float)->float:
+        """
+        Recherche d'une pente supérieure à un seuil
+        Parcours vers l'aval
+        """
+        slopedown=0.
+        curnode=self
+        while curnode.slope < threshold:
+            if curnode.down is None:
+                break
+            curnode=curnode.down
+
+        slopedown = curnode.slope
+        return slopedown
+
+    def slope_upriver(self,threshold:float)->float:
+        """
+        Recherche d'une pente supérieure à un seuil
+        Parcours vers l'amont uniquement selon les rivières
+        """
+        slopeup=0.
+        if self.slope<threshold:
+            if len(self.upriver)>0:
+                slopeup=self.upriver[0].slope_upriver(threshold)
+            else:
+                slopeup=-1.
+        else:
+            slopeup = self.slope
+
+        return slopeup
+
+    def set_strahler(self, strahler:int):
+        """
+
+        """
+        self.strahler = strahler
+
+
+class RiverSystem:
+    """
+    Classe du réseau de rivières d'un modèle hydrologique WOLF
+    """
+    nbreaches:int   # nombre de biefs
+
+    #  reaches
+    #  |__['reaches']
+    #  |  |__[idx]
+    #  |     |__['upstream']
+    #  |     |__['baselist']
+    #  |__['indexed']
+    #  |__['strahler']
+
+    reaches:dict    # dictionnaire des biefs
+
+    kdtree:KDTree   # structure de recherche de voisinage
+
+    upmin:dict      # cf slope_correctionmin
+    upmax:dict      # cf slope_correctionmax
+
+    parent:"Watershed"  # objet Watershed parent
+    upstreams:dict      # dictionnaire des noeuds en amont
+
+    maxlevels:int       # nombre total de niveaux
+    maxstrahler:int     # indice de Strahler max
+
+    tslopemin:float =None   # seuil de pente minimale
+    tslopemax:float =None   # seuil de pente maximale
+
+    plotter:PlotNotebook = None # gestionnaire de graphiques
+    savedir:str         # répertoire de sauvegarde
+
+    def __init__(self,
+                 rivers:list[Node_Watershed],
+                 parent:"Watershed",
+                 thslopemin:float,
+                 thslopemax:float,
+                 savedir:str='',
+                 computecorr:bool=False,
+                 *args,
+                 **kwargs):
+
+        self.savedir = savedir
+        self.parent  = parent
+
+        self.all_nodes = rivers
+
+        self.init_kdtree(self.all_nodes)
+
+        self.nbreaches = max([x.reach for x in rivers])
+        self.reaches   = {}
+        self.reaches['reaches'] = {}
+
+        self.upstreams = {}
+        self.upstreams['list'] = []
+
+        for curreach in range(1,self.nbreaches+1):
+            # attention numérotation 1-based
+            listreach, curup = parent.find_rivers(whichreach=curreach)
+
+            if len(curup.upriver) == 0:
+                # on est en tête de réseau
+                self.upstreams['list'].append(curup)
+
+            self.reaches['reaches'][curreach]={}
+            curdict=self.reaches['reaches'][curreach]
+            curdict['upstream']=curup
+            curdict['baselist']=listreach
+
+        self.create_index() # index et Strahler
+
+        if computecorr:
+            self.tslopemin=thslopemin
+            self.tslopemax=thslopemax
+            self.slope_correctionmin()
+            self.slope_correctionmax()
+
+        return super().__init__(*args, **kwargs)
+
+    def init_kdtree(self, nodes:list[Node_Watershed]):
+        """Create a KDTree structure from coordinates"""
+        xy = [[curnode.x, curnode.y]  for curnode in nodes]
+        self.kdtree = KDTree(xy)
+
+    def get_nearest_nodes(self, xy:np.ndarray, nb=int) -> tuple[np.ndarray, list[Node_Watershed]]:
+        """
+        Return the distance and the  nearest Node_Watershed
+
+        :param xy = np.ndarray - shape (n,2)
+        :param nb = number of neighbors
+
+        return
+        """
+        dd, ii = self.kdtree.query(xy, nb)
+
+        return dd, [self.all_nodes[curi] for curi in ii]
+
+    def get_cums(self, whichreach:int=None, whichup:int=None):
+        """
+        Récupération de la position curvi
+        """
+        curnode:Node_Watershed
+        if whichreach is not None:
+            nodeslist=self.reaches['reaches'][whichreach]['baselist']
+            x=[curnode.cums for curnode in nodeslist]
+        elif whichup is not None:
+            x=[]
+            curnode=self.upstreams['list'][whichup]
+            while curnode is not None:
+                x.append(curnode.cums)
+                curnode=curnode.down
+        else:
+            x=[]
+
+        return x
+
+    def get_dem(self, whichdem:str, whichreach:int=None, whichup:int=None):
+        """
+        Récupération de l'altitude pour une matrice spécifique
+        """
+        if whichreach is not None:
+            nodeslist=self.reaches['reaches'][whichreach]['baselist']
+            dem=[curnode.dem[whichdem] for curnode in nodeslist]
+        elif whichup is not None:
+            curnode:Node_Watershed
+            dem=[]
+            curnode=self.upstreams['list'][whichup]
+            while curnode is not None:
+                dem.append(curnode.dem[whichdem])
+                curnode=curnode.down
+        return dem
+
+    def get_dem_corr(self, whichdem:str, whichreach:int=None, whichup:int=None):
+        """
+        Récupération de l'altitude corrigée pour une matrice spécifique
+        """
+        if whichreach is not None:
+            nodeslist=self.reaches['reaches'][whichreach]['baselist']
+            dem=[curnode.demcorr[whichdem] for curnode in nodeslist]
+        elif whichup is not None:
+            curnode:Node_Watershed
+            dem=[]
+            curnode=self.upstreams['list'][whichup]
+            while curnode is not None:
+                dem.append(curnode.dem[whichdem])
+                curnode=curnode.down
+        return dem
+
+    def get_slope(self, whichslope:str=None, whichreach:int=None, whichup:int=None):
+        """
+        Récupération de la pente
+        """
+        if whichslope is None:
+            if whichreach is not None:
+                nodeslist=self.reaches['reaches'][whichreach]['baselist']
+                slope=[curnode.slope for curnode in nodeslist]
+            elif whichup is not None:
+                curnode:Node_Watershed
+                slope=[]
+                curnode=self.upstreams['list'][whichup]
+                while curnode is not None:
+                    slope.append(curnode.slope)
+                    curnode=curnode.down
+        else:
+            if whichreach is not None:
+                nodeslist=self.reaches['reaches'][whichreach]['baselist']
+                slope=[curnode.slopecorr[whichslope]['value'] for curnode in nodeslist]
+            elif whichup is not None:
+                curnode:Node_Watershed
+                slope=[]
+                curnode=self.upstreams['list'][whichup]
+                while curnode is not None:
+                    slope.append(curnode.slopecorr[whichslope]['value'])
+                    curnode=curnode.down
+
+        return slope
+
+    def create_index(self):
+        """
+        Incrément d'index depuis l'amont jusque l'exutoire final
+        Parcours des mailles rivières depuis tous les amonts et Incrémentation d'une unité
+        Résultat :
+         - tous les biefs en amont sont à 1
+         - Les autres biefs contiennent le nombre de biefs en amont
+
+        Indice de Strahler
+        """
+        for curup in self.upstreams['list']:
+            curnode:Node_Watershed
+            curnode=curup
+            while not curnode is None:
+                curnode.reachlevel +=1
+                curnode=curnode.down
+
+        #recherche de l'index max --> à l'exutoire
+        self.maxlevels = self.parent.outlet.reachlevel
+        self.maxstrahler=0
+        self.reaches['indexed']={}
+        for i in range(1,self.maxlevels+1):
+            self.reaches['indexed'][i]=[]
+
+        #création de listes pour chaque niveau
+        for curreach in self.reaches['reaches']:
+            curdict=self.reaches['reaches'][curreach]
+            listreach=curdict['baselist']
+            curlevel=listreach[0].reachlevel
+            self.reaches['indexed'][curlevel].append(curreach)
+
+        #création de listes pour chaque amont
+        #  on parcourt toutes les mailles depuis chaque amont et on ajoute les index de biefs qui sont différents
+        for idx,curup in enumerate(self.upstreams['list']):
+            curdict=self.upstreams[idx]={}
+            curdict['up']=curup
+            curdict['fromuptodown']=[]
+            curdict['fromuptodown'].append(curup.reach)
+            curnode=curup.down
+            while not curnode is None:
+                if curnode.reach!=curdict['fromuptodown'][-1]:
+                    curdict['fromuptodown'].append(curnode.reach)
+                curnode=curnode.down
+
+        #création de l'indice de Strahler
+        self.reaches['strahler']={}
+        #on commence par ajouter les biefs de 1er niveau qui sont à coup sûr d'indice 1
+        self.reaches['strahler'][1]=self.reaches['indexed'][1]
+        for curreach in self.reaches['strahler'][1]:
+            self.set_strahler_in_nodes(curreach,1)
+
+        #on parcourt les différents niveaux
+        for i in range(2,self.maxlevels+1):
+            listlevel=self.reaches['indexed'][i]
+            for curreach in listlevel:
+                curup:Node_Watershed
+                curup=self.reaches['reaches'][curreach]['upstream']
+                upidx=list(x.strahler for x in curup.upriver)
+                sameidx=upidx[0]==upidx[-1]
+                maxidx=max(upidx)
+
+                curidx=maxidx
+                if sameidx:
+                    curidx+=1
+                    if not curidx in self.reaches['strahler'].keys():
+                         #création de la liste du niveau supérieur
+                         self.reaches['strahler'][curidx]=[]
+                         self.maxstrahler=curidx
+
+                self.reaches['strahler'][curidx].append(curreach)
+                self.set_strahler_in_nodes(curreach,curidx)
+
+
+        myarray=WolfArray(mold=self.parent.subs_array)
+        myarray.reset()
+        curnode:Node_Watershed
+        for curreach in self.reaches['reaches']:
+            curdict=self.reaches['reaches'][curreach]
+            listreach=curdict['baselist']
+            for curnode in listreach:
+                i=curnode.i
+                j=curnode.j
+                myarray.array[i,j]=curnode.strahler
+        myarray.filename = self.parent.dir+'\\Characteristic_maps\\Drainage_basin.strahler'
+        myarray.write_all()
+        myarray.reset()
+        for curreach in self.reaches['reaches']:
+            curdict=self.reaches['reaches'][curreach]
+            listreach=curdict['baselist']
+            for curnode in listreach:
+                i=curnode.i
+                j=curnode.j
+                myarray.array[i,j]=curnode.reachlevel
+        myarray.filename = self.parent.dir+'\\Characteristic_maps\\Drainage_basin.reachlevel'
+        myarray.write_all()
+
+    def set_strahler_in_nodes(self, whichreach:int, strahler:int):
+        """
+        Mise à jour de la propriété dans chaque noeud du bief
+        """
+        listnodes = self.reaches['reaches'][whichreach]['baselist']
+
+        curnode:Node_Watershed
+        for curnode in listnodes:
+            curnode.set_strahler(strahler)
+
+    def plot_dem(self, which:int=-1):
+        """
+        Graphiques
+        """
+        mymarkers=['x','+','1','2','3','4']
+        if which==-1:
+            fig=self.plotter.add('All Reaches')
+
+            ax=fig.add_ax()
+
+            for curreach in self.reaches['reaches']:
+                x=np.array(self.get_cums(whichreach=curreach))
+                for idx,curdem in enumerate(LISTDEM):
+                    y=np.array(self.get_dem(curdem,whichreach=curreach))
+
+                    xmask=np.ma.masked_where(y==99999.,x)
+                    ymask=np.ma.masked_where(y==99999.,y)
+
+                    ax.scatter(xmask,ymask,marker=mymarkers[idx],label=curdem)
+            ax.legend()
+            fig.canvas.draw()
+
+        elif which==-99:
+            size=int(np.ceil(np.sqrt(self.nbreaches)))
+
+            fig=self.plotter.add('reaches')
+
+            for index,curreach in enumerate(self.reaches['reaches']):
+                #curax=ax[int(np.floor(index/size)),int(np.mod(index,size))]
+                curax=fig.add_ax()
+
+                curdict=self.reaches['reaches'][curreach]
+                x=np.array(self.get_cums(whichreach=curreach))
+
+                for idx,curdem in enumerate(LISTDEM):
+                    y=np.array(self.get_dem(curdem,whichreach=curreach))
+
+                    xmask=np.ma.masked_where(y==99999.,x)
+                    ymask=np.ma.masked_where(y==99999.,y)
+
+                    curax.scatter(xmask,ymask,marker=mymarkers[idx],label=curdem)
+            curax.legend()
+            fig.canvas.draw()
+
+        elif which==-98:
+            size=int(np.ceil(np.sqrt(len(self.upstreams['list']))))
+
+            fig=self.plotter.add('reaches')
+
+            for idxup,curup in enumerate(self.upstreams['list']):
+                curax=fig.add_ax()
+
+                x=np.array(self.get_cums(whichup=idxup))
+
+                for idx,curdem in enumerate(LISTDEM):
+                    y=np.array(self.get_dem(curdem,whichup=idxup))
+
+                    xmask=np.ma.masked_where(y==99999.,x)
+                    ymask=np.ma.masked_where(y==99999.,y)
+                    curax.scatter(xmask,ymask,marker=mymarkers[idx],label=curdem)
+
+            curax.legend()
+            fig.canvas.draw()
+
+        elif which>-1:
+            if which<len(self.upstreams['list']):
+                if not self.plotter is None:
+                    fig=self.plotter.add('Upstream n°'+str(which))
+                else:
+                    fig=plt.figure()
+
+                ax=fig.add_ax()
+
+                x=np.array(self.get_cums(whichup=which))
+                for idx,curdem in enumerate(LISTDEM):
+                    y=np.array(self.get_dem(curdem,whichup=which))
+
+                    xmask=np.ma.masked_where(y==99999.,x)
+                    ymask=np.ma.masked_where(y==99999.,y)
+                    ax.scatter(xmask,ymask,marker=mymarkers[idx],label=curdem)
+
+            ax.legend()
+            fig.canvas.draw()
+
+    def plot_dem_and_corr(self, which:int=-1, whichdem:str='dem_after_corr'):
+        """
+        Graphiques
+        """
+
+        if which<len(self.upstreams['list']):
+            if not self.plotter is None:
+                fig=self.plotter.add('Upstream n°'+str(which))
+            else:
+                fig=plt.figure()
+                fig.suptitle('Upstream n°'+str(which))
+
+            ax=fig.add_ax()
+
+            x=np.array(self.get_cums(whichup=which))
+            y=np.array(self.get_dem(whichdem,whichup=which))
+
+            xcorr=self.upmin[which][whichdem][0]
+            ycorr=self.upmin[which][whichdem][1]
+
+            xmask=np.ma.masked_where(y==99999.,x)
+            ymask=np.ma.masked_where(y==99999.,y)
+
+            ax.scatter(xmask,ymask,marker='x',label=whichdem)
+            ax.scatter(xcorr,ycorr,marker='+',label='selected points')
+
+            ax.legend()
+            fig.canvas.draw()
+
+            if not self.savedir=='':
+                plt.savefig(self.savedir+'\\Up'+str(which)+'_'+whichdem+'.png')
+
+    def plot_slope(self, which:int=-1):
+        """
+        Graphiques
+        """
+        mymarkers=['x','+','1','2','3','4']
+        if which==-1:
+            fig=self.plotter.add('reaches')
+            ax=fig.add_ax()
+
+            for curreach in self.reaches['reaches']:
+                x=self.get_cums(whichreach=curreach)
+                for idx,curdem in enumerate(LISTDEM):
+                    y=self.get_slope(curdem,whichreach=curreach)
+                    ax.scatter(x,y,marker=mymarkers[idx],label=curdem)
+            fig.canvas.draw()
+
+        elif which==-99:
+            size=int(np.ceil(np.sqrt(self.nbreaches)))
+            fig=self.plotter.add('reaches')
+
+            for index,curreach in enumerate(self.reaches['reaches']):
+                curax=fig.add_ax()
+
+                x=self.get_cums(whichreach=curreach)
+
+                for idx,curdem in enumerate(LISTDEM):
+                    y=self.get_slope(curdem,whichreach=curreach)
+                    curax.scatter(x,y,marker=mymarkers[idx],label=curdem)
+            curax.legend()
+            fig.canvas.draw()
+
+        elif which==-98:
+            size=int(np.ceil(np.sqrt(len(self.upstreams['list']))))
+
+            fig=self.plotter.add('reaches')
+
+            for idxup,curup in enumerate(self.upstreams['list']):
+                curax=fig.add_ax()
+                x=self.get_cums(whichup=idxup)
+
+                for idx,curdem in enumerate(LISTDEM):
+                    y=self.get_slope(curdem,whichup=idxup)
+                    curax.scatter(x,y,marker=mymarkers[idx],label=curdem)
+            curax.legend()
+            fig.canvas.draw()
+
+    def write_slopes(self):
+        """
+        Ecriture sur disque
+        """
+        #Uniquement les pentes rivières
+        for curlist in LISTDEM:
+            slopes= WolfArray(self.parent.dir+'\\Characteristic_maps\\Drainage_basin.slope')
+            slopes.reset()
+            for curreach in self.reaches['reaches']:
+                curdict=self.reaches['reaches'][curreach]
+                listreach=curdict['baselist']
+
+                curnode:Node_Watershed
+                ijval = np.asarray([[curnode.i, curnode.j, curnode.slopecorr[curlist]['value']]  for curnode in listreach])
+                slopes.array[np.int32(ijval[:,0]),np.int32(ijval[:,1])]=ijval[:,2]
+
+            slopes.filename = self.parent.dir+'\\Characteristic_maps\\Drainage_basin.slope_corr_riv_'+curlist
+            slopes.write_all()
+
+    def slope_correctionmin(self):
+        """
+        Correction pente minimale
+        """
+        if self.tslopemin is not None:
+            logging.info(_('select min - river'))
+            self.selectmin()
+            logging.info(_('slope correction min - river'))
+            self.compute_slopescorr(self.upmin)
+
+    def slope_correctionmax(self):
+        """
+        Correction pente maximale
+        """
+        if self.tslopemax is not None:
+            logging.info(_('select max - river'))
+            self.selectmax()
+            logging.info(_('slope correction max - river'))
+            self.compute_slopescorr(self.upmax)
+
+    def selectmin(self):
+        """
+        Sélection des valeurs minimales afin de conserver une topo décroissante vers l'aval --> une pente positive
+        """
+        self.upmin={}
+
+        #on initialise le dictionnaire de topo min pour chaque amont
+        for idx,curup in enumerate(self.upstreams['list']):
+            self.upmin[idx]={}
+
+        curnode:Node_Watershed
+        for curdem in LISTDEM:
+            logging.info(_('Current DEM : {}'.format(curdem)))
+            for idx,curup in enumerate(self.upstreams['list']):
+                #on part de l'amont
+                curnode=curup
+                x=[]
+                y=[]
+
+                x.append(curnode.cums)
+
+                if curdem=='crosssection':
+                    basey=min(curnode.dem[curdem],curnode.dem['dem_after_corr'])
+                else:
+                    basey=curnode.dem[curdem]
+
+                y.append(basey)
+                curnode=curnode.down
+
+                locs= self.parent.resolution
+                while not curnode is None:
+                    if curdem=='crosssection':
+                        yloc=min(curnode.dem[curdem],curnode.dem['dem_after_corr'])
+                    else:
+                        yloc=curnode.dem[curdem]
+
+                    #on ajoute la maille si la pente est suffisante, sinon cekla créera un trou dans le parcours
+                    if (basey-yloc)/locs>self.tslopemin:
+                        x.append(curnode.cums)
+                        y.append(yloc)
+                        basey=yloc
+                        locs= self.parent.resolution
+                    else:
+                        locs+= self.parent.resolution
+
+                    #if curnode.i==232 and curnode.j==226:
+                    #    a=1
+
+                    curnode=curnode.down
+
+                #on stocke les vecteurs de coordonnées curvi et d'altitudes pour les zones respectant les critères
+                self.upmin[idx][curdem]=[x,y]
+
+    def selectmax(self):
+        """
+        Sélection des valeurs maximales afin de conserver une topo décroissante vers l'aval --> une pente positive
+        """
+        # on travaille sur base de la topo corrigée min
+        self.upmax={}
+
+        #on initialise le dictionnaire de topo max pour chaque amont
+        for idx,curup in enumerate(self.upstreams['list']):
+            self.upmax[idx]={}
+
+        ds=self.parent.resolution
+        curnode:Node_Watershed
+        for curdem in LISTDEM:
+            logging.info(_('Current DEM : {}'.format(curdem)))
+            for idx,curup in enumerate(self.upstreams['list']):
+                curnode=curup
+                x=[]
+                y=[]
+
+                basey=curnode.demcorr[curdem]['value']
+
+                x.append(curnode.cums)
+                y.append(basey)
+                curnode=curnode.down
+
+                locs= ds
+                while not curnode is None:
+                    yloc=curnode.demcorr[curdem]['value']
+
+                    if (basey-yloc)/locs>self.tslopemax:
+                        while len(x)>1 and (basey-yloc)/locs>self.tslopemax:
+                            x.pop()
+                            y.pop()
+                            basey=y[-1]
+                            locs+=ds
+
+                    if yloc<y[-1]:
+                        x.append(curnode.cums)
+                        y.append(yloc)
+                        basey=yloc
+                        locs=ds
+
+                    curnode=curnode.down
+
+                self.upmax[idx][curdem]=[x,y]
+
+    def compute_slopescorr(self, whichdict:dict):
+        """
+        Calcul des pents corrigées
+        """
+        curnode:Node_Watershed
+        for curdem in LISTDEM:
+            logging.info(_('Current DEM : {}'.format(curdem)))
+            for idx,curup in enumerate(self.upstreams['list']):
+                curdict=whichdict[idx][curdem]
+                xmin=curdict[0]
+                if len(xmin)>1:
+                    ymin=curdict[1]
+                    x=self.get_cums(whichup=idx)
+
+                    #on cale une fonction d'interpolation sur la sélection dans lequalle on a oublié les pentes faibles --> à trou
+                    f=interpolate.interp1d(xmin,ymin, fill_value='extrapolate')
+                    #on interpole sur tous les x --> on remplit les trous
+                    y=f(x)
+                    #calcul des pentes sur base des noeuds aval
+                    slopes=self.compute_slope_down(x,y)
+
+                    #on remplit le dictionnaire de résultat
+                    curnode=curup
+                    i=0
+                    while not curnode is None:
+                        #if curnode.i==232 and curnode.j==226:
+                        #    a=1
+                        curnode.demcorr[curdem]['parts'].append(y[i])
+                        curnode.slopecorr[curdem]['parts'].append(slopes[i])
+                        i+=1
+                        curnode=curnode.down
+
+        #calcul de la moyenne sur base des valeurs partielles
+        for curdem in LISTDEM:
+            for curreach in self.reaches['reaches']:
+                nodeslist=self.reaches['reaches'][curreach]['baselist']
+                for curnode in nodeslist:
+                    #if curnode.i==232 and curnode.j==226:
+                    #    a=1
+                    if len(nodeslist)<2:
+                        if not self.tslopemin is None:
+                            curnode.slopecorr[curdem]['value']=max(self.tslopemin,curnode.slope)
+                        else:
+                            curnode.slopecorr[curdem]['value']=self.tslopemin=1.e-4
+
+                        if not self.tslopemax is None:
+                            curnode.slopecorr[curdem]['value']=min(self.tslopemax,curnode.slope)
+                    else:
+                        curnode.demcorr[curdem]['value']=np.mean(curnode.demcorr[curdem]['parts'])
+                        curnode.slopecorr[curdem]['value']=np.mean(curnode.slopecorr[curdem]['parts'])
+
+                    #on vide les parts
+                    curnode.demcorr[curdem]['parts']=[]
+                    curnode.slopecorr[curdem]['parts']=[]
+
+    def compute_slope_down(self, x, y):
+        """
+        Calcul de pente sur base de x et y
+        """
+        slope=[]
+        for i in range(len(x)-1):
+            slope.append((y[i+1]-y[i])/(x[i+1]-x[i]))
+        slope.append(slope[-1])
+        return slope
+
+    def plot_all_in_notebook(self):
+        """
+        Graphiques
+        """
+        self.plotter = PlotNotebook()
+
+        for i in range(self.nbreaches):
+            self.plot_dem_and_corr(i,whichdem='crosssection')
+        self.plot_dem()
+        self.plot_slope(-98)
+        self.plot_dem(-98)
+
+class RunoffSystem:
+    """Classe de l'ensemble des mailles de ruissellement d'un modèle hydrologique WOLF"""
+    nodes:list[Node_Watershed]  # liste de noeuds
+
+    parent:"Watershed"
+    upstreams:dict
+
+    tslopemin:float
+    tslopemax:float
+
+    upmin:dict
+    upmax:dict
+
+    def __init__(self,
+                 runoff:list[Node_Watershed],
+                 parent:"Watershed",
+                 thslopemin:float = None,
+                 thslopemax:float = None,
+                 computecorr:bool=False,
+                 *args,
+                 **kwargs):
+
+        self.parent  = parent
+        self.nodes = runoff
+        self.upstreams={}
+
+        #sélection des mailles qui ont une surface unitaire comme surface drainée
+        areaup = pow(parent.resolution,2)/1.e6
+        self.upstreams['list']=list(filter(lambda x: (x.uparea-areaup)<1.e-6 ,runoff))
+
+        if computecorr:
+            self.tslopemin = thslopemin
+            self.tslopemax = thslopemax
+
+            self.slope_correctionmin()
+            self.slope_correctionmax()
+
+        return super().__init__(*args, **kwargs)
+
+    def get_oneup(self, idx:int) -> Node_Watershed:
+        """
+        Récupération d'un amont sur base de l'index
+        """
+        return self.upstreams['list'][idx]
+
+    def get_cums(self,whichup:int=None):
+
+        if not whichup is None:
+            curnode:Node_Watershed
+            x=[]
+            curnode=self.get_oneup(whichup)
+            while not curnode.river:
+                x.append(curnode.cums)
+                curnode=curnode.down
+            if len(x)==1:
+                x.append(curnode.cums)
+        else:
+            x=[]
+
+        return x
+
+    def get_dem(self, whichdem:str, whichup:int=None):
+        if not whichdem in LISTDEM:
+            return
+
+        if not whichup is None:
+            curnode:Node_Watershed
+            dem=[]
+            curnode=self.get_oneup(whichup)
+            while not curnode.river:
+                dem.append(curnode.dem[whichdem])
+                curnode=curnode.down
+        return dem
+
+    def get_dem_corr(self, whichdem:str, whichup:int=None):
+        if not whichdem in LISTDEM:
+            return
+
+        if not whichup is None:
+            curnode:Node_Watershed
+            dem=[]
+            curnode=self.get_oneup(whichup)
+            while not curnode.river:
+                dem.append(curnode.dem[whichdem])
+                curnode=curnode.down
+        return dem
+
+    def get_slope(self, whichslope:str=None, whichup:int=None):
+
+        if whichslope is None:
+            if not whichup is None:
+                curnode:Node_Watershed
+                slope=[]
+                curnode=self.get_oneup(whichup)
+                while not curnode.river:
+                    slope.append(curnode.slope)
+                    curnode=curnode.down
+        else:
+            if not whichup is None:
+                curnode:Node_Watershed
+                slope=[]
+                curnode=self.get_oneup(whichup)
+                while not curnode.river:
+                    slope.append(curnode.slopecorr[whichslope]['value'])
+                    curnode=curnode.down
+
+        return slope
+
+    def plot_dem(self, which:int=-1):
+
+        mymarkers=['x','+','1','2','3','4']
+        if which>-1:
+            if which<len(self.upstreams['list']):
+                fig=plt.figure()
+                fig.suptitle('Upstream n°'+str(which))
+
+                x=np.array(self.get_cums(whichup=which))
+                for idx,curdem in enumerate(LISTDEM):
+                    y=np.array(self.get_dem(curdem,whichup=which))
+
+                    xmask=np.ma.masked_where(y==99999.,x)
+                    ymask=np.ma.masked_where(y==99999.,y)
+                    plt.scatter(xmask,ymask,marker=mymarkers[idx],label=curdem)
+
+            plt.legend()
+        plt.show()
+
+    def plot_dem_and_corr(self, which:int=-1, whichdem:str='dem_after_corr'):
+
+        if which<len(self.upstreams['list']):
+            fig=plt.figure()
+            fig.suptitle('Upstream n°'+str(which))
+
+            x=np.array(self.get_cums(whichup=which))
+            y=np.array(self.get_dem(whichdem,whichup=which))
+
+            xcorr=self.upmin[which][whichdem][0]
+            ycorr=self.upmin[which][whichdem][1]
+
+            xmask=np.ma.masked_where(y==99999.,x)
+            ymask=np.ma.masked_where(y==99999.,y)
+
+            plt.scatter(xmask,ymask,marker='x',label=whichdem)
+            plt.scatter(xcorr,ycorr,marker='+',label='selected points')
+
+            plt.legend()
+            plt.savefig(r'D:\OneDrive\OneDrive - Universite de Liege\Crues\2021-07 Vesdre\Simulations\Hydrologie\Up'+str(which)+'_'+whichdem+'.png')
+            #plt.show()
+
+    def write_slopes(self):
+        #Uniquement les pentes runoff
+        for curlist in LISTDEM:
+            slopes= WolfArray(self.parent.dir+'\\Characteristic_maps\\Drainage_basin.slope')
+            slopes.reset()
+            curnode:Node_Watershed
+            for curnode in self.nodes:
+                i=curnode.i
+                j=curnode.j
+                slopes.array[i,j]=curnode.slopecorr[curlist]['value']
+
+            slopes.filename = self.parent.dir+'\\Characteristic_maps\\Drainage_basin.slope_corr_run_'+curlist
+            slopes.write_all()
+
+    def slope_correctionmin(self):
+        if not self.tslopemin is None:
+            logging.info(_('select min - runoff'))
+            self.selectmin()
+            logging.info(_('slope correction min - runoff'))
+            self.compute_slopescorr(self.upmin)
+
+    def slope_correctionmax(self):
+        if not self.tslopemax is None:
+            logging.info(_('select max - runoff'))
+            self.selectmax()
+            logging.info(_('slope correction max - runoff'))
+            self.compute_slopescorr(self.upmax)
+
+    def selectmin(self):
+        #Sélection des valeurs minimales afin de conserver une topo décroissante vers l'aval --> une pente positive
+        self.upmin={}
+
+        #on initialise le dictionnaire de topo min pour chaque amont
+        for idx,curup in enumerate(self.upstreams['list']):
+            self.upmin[idx]={}
+
+        ds=self.parent.resolution
+        curnode:Node_Watershed
+        for curdem in LISTDEM:
+            logging.info(_('Current DEM : {}'.format(curdem)))
+            for idx,curup in enumerate(self.upstreams['list']):
+                curnode=curup
+                x=[]
+                y=[]
+
+                if curdem=='crosssection':
+                    basey=min(curnode.dem[curdem],curnode.dem['dem_after_corr'])
+                else:
+                    basey=curnode.dem[curdem]
+
+                x.append(curnode.cums)
+                y.append(basey)
+                curnode=curnode.down
+
+                locs=ds
+                while not curnode is None:
+                    if curdem=='crosssection':
+                        yloc=min(curnode.dem[curdem],curnode.dem['dem_after_corr'])
+                    else:
+                        yloc=curnode.dem[curdem]
+
+                    if (basey-yloc)/locs>self.tslopemin:
+                        x.append(curnode.cums)
+                        y.append(yloc)
+                        basey=yloc
+                        locs=ds
+                        if curnode.river:
+                            break
+                    else:
+                        locs+=ds
+                    curnode=curnode.down
+
+                self.upmin[idx][curdem]=[x,y]
+
+    def selectmax(self):
+        #Sélection des valeurs minimales afin de conserver une topo décroissante vers l'aval --> une pente positive
+        self.upmax={}
+
+        #on initialise le dictionnaire de topo min pour chaque amont
+        for idx,curup in enumerate(self.upstreams['list']):
+            self.upmax[idx]={}
+
+        ds=self.parent.resolution
+        curnode:Node_Watershed
+        for curdem in LISTDEM:
+            logging.info(_('Current DEM : {}'.format(curdem)))
+            for idx,curup in enumerate(self.upstreams['list']):
+                curnode=curup
+                x=[]
+                y=[]
+
+                """
+                if curdem=='crosssection':
+                    basey=min(curnode.demcorr[curdem]['value'],curnode.demcorr['dem_after_corr']['value'])
+                else:
+                    basey=curnode.demcorr[curdem]['value']
+                """
+                basey=curnode.demcorr[curdem]['value']
+
+                x.append(curnode.cums)
+                y.append(basey)
+                curnode=curnode.down
+
+                locs= ds
+                while not curnode is None:
+                    """
+                    if curdem=='crosssection':
+                        yloc=min(curnode.demcorr[curdem]['value'],curnode.demcorr['dem_after_corr']['value'])
+                    else:
+                        yloc=curnode.demcorr[curdem]['value']
+                    """
+                    yloc=curnode.demcorr[curdem]['value']
+
+                    if (basey-yloc)/locs>self.tslopemax:
+                        while len(x)>1 and (basey-yloc)/locs>self.tslopemax:
+                            x.pop()
+                            y.pop()
+                            basey=y[-1]
+                            locs+=ds
+
+                    if yloc<y[-1]:
+                        x.append(curnode.cums)
+                        y.append(yloc)
+                        basey=yloc
+                        locs=ds
+                        if curnode.river:
+                            break
+
+                    curnode=curnode.down
+                    #if curnode.i==187 and curnode.j==207:
+                    #    a=1
+
+                self.upmax[idx][curdem]=[x,y]
+
+    def compute_slopescorr(self, whichdict:dict):
+        curnode:Node_Watershed
+        for curdem in LISTDEM:
+            logging.info(_('Current DEM : {}'.format(curdem)))
+            for idx,curup in enumerate(self.upstreams['list']):
+                curdict=whichdict[idx][curdem]
+                xmin=curdict[0]
+                if len(xmin)>1:
+                    ymin=curdict[1]
+                    x=self.get_cums(whichup=idx)
+
+                    f=interpolate.interp1d(xmin,ymin, fill_value='extrapolate')
+                    y=f(x)
+                    slopes=self.compute_slope_down(x,y)
+
+                    curnode=curup
+                    i=0
+                    while not curnode.river:
+                        #if curnode.i==187 and curnode.j==207:
+                        #    a=1
+                        curnode.demcorr[curdem]['parts'].append(y[i])
+                        curnode.slopecorr[curdem]['parts'].append(slopes[i])
+                        i+=1
+                        curnode=curnode.down
+        #calcul de la moyenne sur base des valeurs partielles
+        for curdem in LISTDEM:
+            for curnode in self.nodes:
+                #if curnode.i==187 and curnode.j==207:
+                #    a=1
+                if len(curnode.slopecorr[curdem]['parts'])<2:
+                    #Ce cas particulier peut arriver si des mailles BV sont remplies par une zone plate qui s'étend en rivière
+                    # Comme on ne recherche de mailles plus basses que dans la partie BV, il n'est pas possible de corriger les pentes
+                    if not self.tslopemin is None:
+                        curnode.slopecorr[curdem]['value']=max(self.tslopemin,curnode.slope)
+                    else:
+                        curnode.slopecorr[curdem]['value']=1.e-4
+
+                    if not self.tslopemax is None:
+                        curnode.slopecorr[curdem]['value']=min(self.tslopemax,curnode.slope)
+                else:
+                    curnode.demcorr[curdem]['value']=np.mean(curnode.demcorr[curdem]['parts'])
+                    curnode.slopecorr[curdem]['value']=np.mean(curnode.slopecorr[curdem]['parts'])
+
+                curnode.demcorr[curdem]['parts']=[]
+                curnode.slopecorr[curdem]['parts']=[]
+
+    def compute_slope_down(self, x, y):
+        """
+        Calcul de la pente sur base de listes X et Y
+        """
+        slope=[]
+        for i in range(len(x)-1):
+            slope.append((y[i+1]-y[i])/(x[i+1]-x[i]))
+        slope.append(slope[-1])
+        return slope
+
+class SubWatershed:
+    """ Classe sous-bassin versant """
+    def __init__(self,
+                 parent:"Watershed",
+                 name:str,
+                 idx:int,
+                 mask:WolfArray,
+                 nodes:list[Node_Watershed],
+                 runoff:list[Node_Watershed],
+                 rivers:list[Node_Watershed]) -> None:
+
+        self.parent = parent
+        self.index  = idx
+        self.name   = name
+        self.mask   = mask
+        self.nodes  = nodes
+        self.rivers = rivers
+        self.runoff = runoff
+        self.idx_reaches = np.unique(np.asarray([x.reach for x in rivers]))
+
+    @property
+    def surface(self) -> float:
+        return self.mask.nbnotnull * self.mask.dx * self.mask.dy
+
+class Watershed:
+    """Classe bassin versant"""
+
+    header:header_wolf  # header_wolf of "Drainage_basin.sub" wolf_array
+
+    dir: str            # Répertoire de modélisation
+
+    outlet:Node_Watershed   # exutoire
+
+    subs_array: WolfArray # "Drainage_basin.sub" wolf_array
+
+    nodes:list[Node_Watershed] # all nodes
+    nodesindex:np.ndarray # indirect access to mynodes, contains index of instance in list
+    rivers:list[Node_Watershed] # all river nodes
+    runoff:list[Node_Watershed] # all runoff nodes
+
+    couplednodes:list # forced exchanges
+
+    subcatchments: list[SubWatershed]
+    statisticss: dict
+
+    couplednodesxy:list[float,float,float,float]
+    couplednodesij:list[tuple[int,int],tuple[int,int]]
+
+    riversystem:RiverSystem     # réseau de rivières
+    runoffsystem:RunoffSystem   # écoulement diffus/hydrologique
+
+    def __init__(self,
+                 dir:str,
+                 thzmin:float=None,
+                 thslopemin:float=None,
+                 thzmax:float=None,
+                 thslopemax:float=None,
+                 crosssections:CrossSections=None,
+                 computestats:bool=False,
+                 computecorr:bool=False,
+                 plotstats:bool=False,
+                 plotriversystem=False,
+                 *args, **kwargs):
+
+        logging.info(_('Read files...'))
+        self.dir=os.path.normpath(dir)
+
+        self.subs_array   = WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.sub')
+
+        self.header = self.subs_array.get_header()
+
+        #FIXME
+        isOk, fe_file = check_path(os.path.join(self.dir, "Coupled_pairs.txt"), prefix=self.dir)
+        self.couplednodesxy=[]
+        self.couplednodesij=[]
+        
+        if isOk>=0:
+            f = open(fe_file, 'r')
+            lines = f.read().splitlines()
+            f.close()
+
+            if lines[0]=='COORDINATES':
+                for xy in enumerate(lines[1:]):
+                    xup,yup,xdown,ydown=xy[1].split('\t')
+                    self.couplednodesxy.append([float(xup),float(yup),float(xdown),float(ydown)])
+                    self.couplednodesij.append([self.subs_array.get_ij_from_xy(float(xup),float(yup)),self.subs_array.get_ij_from_xy(float(xdown),float(ydown))])
+
+        logging.info(_('Initialization of nodes...'))
+        self.nodesindex = np.zeros([self.subs_array.nbx,self.subs_array.nby], dtype=int)
+        self.outlet = None
+        self.up = None
+        self.init_nodes()
+
+        logging.info(_('Initialization of subwatersheds...'))
+        self.init_subs()
+
+        if not crosssections is None:
+            logging.info(_('Cross sections...'))
+            self.crosssections = crosssections
+            self.attrib_cs_to_nodes()
+        else:
+            self.crosssections = None
+
+        logging.info(_('Slopes corrections...'))
+        self.riversystem  = RiverSystem(self.rivers , self,thslopemin=thslopemin, thslopemax=thslopemax, computecorr=computecorr)
+        self.runoffsystem = RunoffSystem(self.runoff, self,thslopemin=thslopemin, thslopemax=thslopemax, computecorr=computecorr)
+
+        if computestats or plotstats:
+            logging.info(_('Statistics...'))
+            self.compute_stats(plotstats)
+
+        #Ecriture des résultats de correction des pentes
+        if computecorr:
+            logging.info(_('Writing data to disk'))
+            self.write_dem()
+            self.write_slopes()
+
+        if plotriversystem:
+            logging.info(_('Plot rivers'))
+            self.riversystem.plot_all_in_notebook()
+
+        logging.info(_('Done!'))
+
+    @property
+    def nb_subs(self):
+        return int(ma.max(self.subs_array.array))
+
+    @property
+    def resolution(self):
+        return self.header.dx
+
+    def get_node_from_ij(self, i:int,j:int):
+        """
+        Récupération d'un objet Node_Watershed sur base des indices (i,j)
+        """
+        shape = self.nodesindex.shape
+        if i<0 or i>=shape[0]:
+            return None
+        if j<0 or j>=shape[1]:
+            return None
+        idx = self.nodesindex[i,j]
+        if idx<0 or idx >= len(self.nodes):
+            return None
+
+        return self.nodes[idx]
+
+    def get_node_from_xy(self, x:float, y:float):
+        """
+        Récupération d'un objet Node_Watershed sur base des coordonnées (x,y)
+        """
+        i,j = self.header.get_ij_from_xy(x,y)
+        return self.get_node_from_ij(i,j)
+
+    def write_slopes(self):
+        """
+        Ecriture sur disque
+        """
+        for curlist in LISTDEM:
+            curpath=self.dir+'\\Characteristic_maps\\corrslopes\\'+curlist
+            os.makedirs(curpath,exist_ok=True)
+            slopes= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.slope')
+
+            ijval = np.asarray([[curnode.i, curnode.j, curnode.slopecorr[curlist]['value']]  for curnode in self.nodes])
+            slopes.array[np.int32(ijval[:,0]),np.int32(ijval[:,1])]=ijval[:,2]
+
+            slopes.filename = curpath +'\\Drainage_basin.slope_corr'
+            slopes.write_all()
+
+    def write_dem(self):
+        """
+        Ecriture sur disque
+        """
+        for curlist in LISTDEM:
+            curpath=self.dir+'\\Characteristic_maps\\corrdem\\'+curlist
+            os.makedirs(curpath,exist_ok=True)
+            dem= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basincorr.b')
+
+            ijval = np.asarray([[curnode.i, curnode.j, curnode.demcorr[curlist]['value']]  for curnode in self.nodes])
+            dem.array[np.int32(ijval[:,0]),np.int32(ijval[:,1])]=ijval[:,2]
+
+            dem.filename = curpath +'\\Drainage_basincorr.b'
+            dem.write_all()
+
+    @property
+    def crosssections(self):
+        return self._cs
+
+    @crosssections.setter
+    def crosssections(self, value:CrossSections):
+        self._cs = value
+
+    def attrib_cs_to_nodes(self):
+        """
+        Attribution des sections en travers aux noeuds
+        """
+        if self.crosssections is not None:
+            for curlist in self.crosssections:
+                for namecs in curlist.myprofiles:
+                    curvert:wolfvertex
+                    curcs=curlist.myprofiles[namecs]
+
+                    try:
+                        curvert=curcs['bed']
+                    except:
+                        curvert=curlist.get_min(whichprofile=curcs)
+
+                    i,j=self.subs_array.get_ij_from_xy(curvert.x,curvert.y)
+                    curnode:Node_Watershed
+                    curnode =self.nodes[self.nodesindex[i,j]]
+
+                    if curnode.river:
+                        if curnode.crosssections is None:
+                            curnode.crosssections=[]
+                        curnode.crosssections.append(curcs)
+                        curnode.dem['crosssection']=min(curnode.dem['crosssection'],curvert.z)
+
+    def init_nodes(self):
+        """
+        Initialisation des noeuds
+        """
+
+        self.nodes=[Node_Watershed() for i in range(self.subs_array.nbnotnull)]
+
+        dem_before_corr= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.b')
+        dem_after_corr= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basincorr.b')
+        demdelta= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basindiff.b')
+        slopes= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.slope',masknull=False)
+        reaches= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.reachs')
+        cnv= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.cnv')
+        times= WolfArray(self.dir+'\\Characteristic_maps\\Drainage_basin.time')
+
+        dem_after_corr.array.mask = self.subs_array.array.mask
+
+        nb=0
+        for index,x in tqdm(np.ndenumerate(self.subs_array.array), 'Numbering'):
+            if(x>0):
+                i=int(index[0])
+                j=int(index[1])
+                self.nodesindex[i,j]=nb
+                nb+=1
+
+        curnode:Node_Watershed
+        nb=0
+        for index, i_sub in tqdm(np.ndenumerate(self.subs_array.array), 'Initialization'):
+            if(i_sub>0):
+                i=int(index[0])
+                j=int(index[1])
+                x, y = self.header.get_xy_from_ij(i,j)
+                curnode =self.nodes[self.nodesindex[i,j]]
+
+                curnode.i = i
+                curnode.j = j
+
+                curnode.x = x
+                curnode.y = y
+
+                curnode.crosssections = None
+                curnode.down = None
+
+                curnode.index=nb
+                curnode.dem={}
+                curnode.dem['dem_before_corr']=dem_before_corr.array[i,j]
+                curnode.dem['dem_after_corr']=dem_after_corr.array[i,j]
+                curnode.dem['crosssection']=99999.
+                curnode.demdelta=demdelta.array[i,j]
+                curnode.slope=slopes.array[i,j]
+
+                curnode.slopecorr={}
+                for curlist in LISTDEM:
+                    curnode.slopecorr[curlist]={}
+                    curnode.slopecorr[curlist]['parts']=[]
+                    curnode.slopecorr[curlist]['value']=curnode.slope
+
+                curnode.demcorr={}
+                for curlist in LISTDEM:
+                    curnode.demcorr[curlist]={}
+                    curnode.demcorr[curlist]['parts']=[]
+                    curnode.demcorr[curlist]['value']=curnode.dem['dem_after_corr']
+
+                curnode.sub=int(i_sub)
+                curnode.time=times.array[i,j]
+                curnode.uparea=cnv.array[i,j]
+                curnode.river=not reaches.array.mask[i,j]
+                if curnode.river:
+                    curnode.reach=int(reaches.array[i,j])
+                curnode.forced=False
+                curnode.up=[]
+                curnode.upriver=[]
+                curnode.strahler=0
+                curnode.reachlevel=0
+                nb+=1
+
+        curdown:Node_Watershed
+        #Liaison échanges forcés
+        incr=slopes.dx
+        for curexch in self.couplednodesij:
+            i=int(curexch[0][0])
+            j=int(curexch[0][1])
+            curnode=self.nodes[self.nodesindex[i,j]]
+            curnode.forced=True
+            idown = int(curexch[1][0])
+            jdown = int(curexch[1][1])
+            curdown = self.nodes[self.nodesindex[idown,jdown]]
+            curnode.down = curdown
+            curdown.up.append(curnode)
+            if curnode.river:
+                curdown.upriver.append(curnode)
+            curnode.incrs = incr * np.sqrt(pow(curdown.i-i,2)+pow(curdown.j-j,2))
+
+        #Liaison hors échanges forcés
+        for curnode in tqdm(self.nodes, 'Linking'):
+            if not curnode.forced:
+                i=curnode.i
+                j=curnode.j
+
+                curtop=curnode.dem['dem_after_corr']
+
+                neigh = [[i-1,j],[i+1,j],[i,j-1], [i,j+1]]
+                diff = [dem_after_corr.array[curi,curj]-curtop if not dem_after_corr.array.mask[curi,curj] else 100000. for curi,curj in neigh]
+                mindiff = np.min(diff)
+                if mindiff<0:
+                    index = diff.index(mindiff)
+                    if index==0:
+                        curdown = self.nodes[self.nodesindex[i-1,j]]
+                    elif index==1:
+                        curdown = self.nodes[self.nodesindex[i+1,j]]
+                    elif index==2:
+                        curdown = self.nodes[self.nodesindex[i,j-1]]
+                    else:
+                        curdown = self.nodes[self.nodesindex[i,j+1]]
+
+                    curnode.down = curdown
+                    curdown.up.append(curnode)
+                    if curnode.river:
+                        curdown.upriver.append(curnode)
+                    curnode.incrs=incr
+                else:
+                    self.outlet = curnode
+
+        #Rechreche de la pente dans les voisins en croix dans la topo non remaniée
+        for curnode in tqdm(self.nodes, 'Finding slope'):
+            if not curnode.forced:
+                i=curnode.i
+                j=curnode.j
+
+                curtop = curnode.dem['dem_before_corr']
+
+                neigh = [[i-1,j], [i+1,j], [i,j-1], [i,j+1], [i-1,j-1], [i+1,j+1], [i+1,j-1], [i-1,j+1]]
+                diff = [dem_before_corr.array[curi,curj]-curtop if not dem_before_corr.array.mask[curi,curj] else 100000. for curi,curj in neigh ]
+                mindiff = np.min(diff)
+
+                fact=1.
+                if mindiff<0:
+                    index = diff.index(mindiff)
+                    if index>3:
+                        fact=np.sqrt(2)
+
+                curnode.sloped8 = -mindiff/(self.resolution*fact)
+
+        self.rivers=list(filter(lambda x: x.river,self.nodes))
+        self.rivers.sort(key=lambda x: x.dem['dem_after_corr'])
+        sys.setrecursionlimit(len(self.nodes))
+        self.outlet.incr_curvi()
+
+        self.find_dem_subpixels()
+
+        self.runoff=self.find_runoffnodes()
+
+    def find_rivers(self, whichsub:int=0, whichreach:int=0) -> tuple[list[Node_Watershed], Node_Watershed]:
+        """
+        Recherche des mailles rivières
+        @param whichsub : numéro du sous-bassin à traiter
+        @param whicreach : numéro du tronçon à identifier
+        """
+        if whichsub>0 and whichsub<=self.nb_subs:
+            if whichreach>0:
+                myrivers=list(filter(lambda x: x.river and x.sub==whichsub and x.reach==whichreach,self.rivers))
+            else:
+                myrivers=list(filter(lambda x: x.river and x.sub==whichsub,self.rivers))
+        else:
+            if whichreach>0:
+                myrivers=list(filter(lambda x: x.river and x.reach==whichreach,self.rivers))
+            else:
+                myrivers=list(filter(lambda x: x.river,self.rivers))
+
+        myrivers.sort(key=lambda x: x.dem['dem_after_corr'])
+
+        up=None
+        if len(myrivers)>0:
+            up=myrivers[-1]
+
+        return myrivers,up
+
+    def find_sub(self, whichsub:int=0) -> list[Node_Watershed]:
+        """
+        Recherche des mailles du sous-bassin versant
+        @param whichsub : numéro du sous-bassin à traiter
+        """
+        if whichsub>0 and whichsub<=self.nb_subs:
+            mysub=list(filter(lambda x: x.sub==whichsub, self.nodes))
+        else:
+            mysub=self.nodes.copy()
+
+        mysub.sort(key=lambda x: x.dem['dem_after_corr'])
+
+        return mysub
+
+    def init_subs(self):
+        """
+        Initialize Sub-Catchments
+        """
+        self.subcatchments=[]
+
+        #Initialisation de la matrice de mask (d'une extension et d'une résolution similaire aux données radar)
+        for i in tqdm(range(1,self.nb_subs+1), 'Subwatershed'):
+            curmask = WolfArray(mold=self.subs_array)
+            curmask.mask_allexceptdata(float(i))
+            all_river_nodes, _ = self.find_rivers(i)
+
+            cursub = SubWatershed(self,
+                                  name = 'sub n'+str(i),
+                                  idx=i-1,
+                                  mask = curmask,
+                                  nodes = self.find_sub(i),
+                                  runoff=self.find_runoffnodes(i),
+                                  rivers=all_river_nodes,
+                                  )
+
+            self.subcatchments.append(cursub)
+
+    def find_runoffnodes(self, whichsub:int=0) -> list[Node_Watershed]:
+        """
+        Recherche des mailles du bassin versant seul (sans les rivières)
+        @param whichsub : numéro du sous-bassin à traiter
+        """
+        if whichsub>0 and whichsub<=self.nb_subs:
+            myrunoff=list(filter(lambda x: not x.river and x.sub==whichsub,self.nodes))
+        else:
+            myrunoff=list(filter(lambda x: not x.river,self.nodes))
+
+        myrunoff.sort(key=lambda x: x.dem['dem_after_corr'])
+
+        return myrunoff
+
+    def index_flatzone(self, listofsortednodes:list, threshold:float):
+        """
+        Indexation des zones de plat
+        """
+
+        curnode:Node_Watershed
+        curflat:Node_Watershed
+
+        curindex=0
+        for curnode in listofsortednodes[-1:1:-1]:
+            addone=False
+            while curnode.slope<threshold and curnode.flatindex==-1:
+                addone=True
+                curnode.flatindex=curindex
+                if curnode.down is None:
+                    break
+                curnode=curnode.down
+            if addone:
+                curindex+=1
+
+        return curindex
+
+    def find_flatnodes(self, listofsortednodes:list):
+        """
+        Recherche des mailles dans des zones de faibles pentes
+        @param listofsortednodes : liste triée de mailles
+        """
+        myflatnodes=list(filter(lambda x: x.flatindex>-1,listofsortednodes))
+
+        return myflatnodes
+
+    def find_flatzones(self, listofsortednodes:list, maxindex:int):
+        """
+        Recherche des mailles dans des zones de faibles pentes
+        @param listofsortednodes : liste triée de mailles
+        """
+        myflatzones=[[]] * maxindex
+        for i in range(maxindex):
+            myflatzones[i]=list(filter(lambda x: x.flatindex==i,listofsortednodes))
+
+        return myflatzones
+
+    def find_dem_subpixels(self):
+        """
+        Recherche des altitudes dans un mnt plus dense
+        """
+        demsubs = {}
+        file_10m = os.path.join(self.dir,'mnt10m.bin')
+        isOk, file_10m = check_path(file_10m, prefix=self.dir)
+        if isOk>=0:
+            dem_10m=WolfArray(file_10m)
+            demsubs["dem_10m"] = dem_10m
+
+        file_20m = os.path.join(self.dir,'mnt20m.bin')
+        isOk, file_20m = check_path(file_20m, prefix=self.dir)
+        if isOk>=0:
+            dem_20m=WolfArray(file_20m)
+            demsubs["dem_20m"] = dem_20m
+
+
+        # demsubs={'dem_10m':dem_10m,'dem_20m':dem_20m}
+        if len(demsubs)==0:
+            return
+
+        curnode:Node_Watershed
+        for curdem in tqdm(demsubs, 'Sub-pixeling'):
+            locdem=demsubs[curdem]
+            dx=locdem.dx
+            dy=locdem.dy
+
+            for curnode in tqdm(self.nodes):
+                curi=curnode.i
+                curj=curnode.j
+
+                curx,cury=self.subs_array.get_xy_from_ij(curi,curj)
+
+                decalx=(self.resolution-dx)/2.
+                decaly=(self.resolution-dy)/2.
+                x1=curx-decalx
+                y1=cury-decaly
+                x2=curx+decalx
+                y2=cury+decaly
+
+                i1,j1=locdem.get_ij_from_xy(x1,y1)
+                i2,j2=locdem.get_ij_from_xy(x2,y2)
+
+                curnode.dem[curdem]=np.min(locdem.array[i1:i2+1,j1:j2+1])
+
+    def compute_stats(self, plot:bool=False):
+        """
+        Calcul des statistiques de pente
+        """
+        self.statisticss={}
+
+        slopes=np.array(list(x.slope for x in self.nodes))
+        slopesrunoff=np.array(list(x.slope for x in list(filter(lambda x: not x.river,self.nodes))))
+        slopesriver=np.array(list(x.slope for x in list(filter(lambda x: x.river,self.nodes))))
+
+        curdict=self.statisticss
+        curdict['slopemin'] = np.min(slopes)
+        curdict['slopemax'] = np.max(slopes)
+        curdict['slopemedian'] = np.median(slopes)
+        curdict['slopemean'] = np.mean(slopes)
+        curdict['hist'] = slopes
+        curdict['hist_watershed'] = slopesrunoff
+        curdict['hist_reaches'] = slopesriver
+        curdict['count_neg'] = np.count_nonzero(slopes < 0.)
+
+        logging.info(_('Min : {}'.format(curdict['slopemin'])))
+        logging.info(_('Max : {}'.format(curdict['slopemax'])))
+        logging.info(_('Median : {}'.format(curdict['slopemedian'])))
+        logging.info(_('Mean : {}'.format(curdict['slopemean'])))
+        logging.info(_('Non Zero : {}'.format(curdict['count_neg'])))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]={}
+
+            slopes=np.array(list(x.slopecorr[curlist]['value'] for x in self.nodes))
+            slopesrunoff=np.array(list(x.slopecorr[curlist]['value'] for x in list(filter(lambda x: not x.river,self.nodes))))
+            slopesriver=np.array(list(x.slopecorr[curlist]['value'] for x in list(filter(lambda x: x.river,self.nodes))))
+
+            curdict['slopemin'] = np.min(slopes)
+            curdict['slopemax'] = np.max(slopes)
+            curdict['slopemedian'] = np.median(slopes)
+            curdict['slopemean'] = np.mean(slopes)
+            curdict['hist'] = slopes
+            curdict['hist_watershed'] = slopesrunoff
+            curdict['hist_reaches'] = slopesriver
+            curdict['count_neg'] = np.count_nonzero(slopes < 0.)
+
+            logging.info(_('Current list : '.format(curlist)))
+            logging.info(_('Min : {}'.format(curdict['slopemin'])))
+            logging.info(_('Max : {}'.format(curdict['slopemax'])))
+            logging.info(_('Median : {}'.format(curdict['slopemedian'])))
+            logging.info(_('Mean : {}'.format(curdict['slopemean'])))
+            logging.info(_('Non Zero : {}'.format(curdict['count_neg'])))
+
+        if plot:
+            self.plot_stats()
+
+    def plot_stats(self):
+
+        self.myplotterstats = PlotNotebook()
+
+        bin1=np.array([1.e-8,1.e-7,1.e-6,5.e-6])
+        bin2=np.linspace(1.e-5,1e-3,num=20)
+        bin3=np.linspace(2.e-3,1e-1,num=20)
+        bin4=np.linspace(.11,1,num=100)
+        bins=np.concatenate((bin1,bin2,bin3,bin4))
+
+        fig=self.myplotterstats.add(_('Slope distribution - log'))
+
+        ax = fig.add_ax()
+        ax.hist(self.statisticss['hist'],bins,cumulative=True,density=True,histtype=u'step',label='base')
+        ax.set_xscale('log')
+        ax.set_xlabel(_('All meshes'))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]
+            ax.hist(curdict['hist'],bins,cumulative=True,density=True,histtype=u'step',label=curlist)
+
+        ax = fig.add_ax()
+        ax.hist(self.statisticss['hist_watershed'],bins,cumulative=True,density=True,histtype=u'step',label='base')
+        ax.set_xscale('log')
+        ax.set_xlabel(_('Watershed'))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]
+            ax.hist(curdict['hist_watershed'],bins,cumulative=True,density=True,histtype=u'step',label=curlist)
+
+        ax = fig.add_ax()
+        ax.hist(self.statisticss['hist_reaches'],bins,cumulative=True,density=True,histtype=u'step',label='base')
+        ax.set_xscale('log')
+        ax.set_xlabel(_('River'))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]
+            ax.hist(curdict['hist_reaches'],bins,cumulative=True,density=True,histtype=u'step',label=curlist)
+
+        ax.legend()
+        fig.canvas.draw()
+
+        fig=self.myplotterstats.add(_('Slope distribution'))
+        ax:plt.axis
+
+        ax = fig.add_ax()
+        ax.hist(self.statisticss['hist'],bins,cumulative=True,density=True,histtype=u'step',label='base')
+        ax.set_xlabel(_('All meshes'))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]
+            ax.hist(curdict['hist'],bins,cumulative=True,density=True,histtype=u'step',label=curlist)
+
+        ax = fig.add_ax()
+        ax.hist(self.statisticss['hist_watershed'],bins,cumulative=True,density=True,histtype=u'step',label='base')
+        ax.set_xlabel(_('Watershed'))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]
+            ax.hist(curdict['hist_watershed'],bins,cumulative=True,density=True,histtype=u'step',label=curlist)
+
+        ax = fig.add_ax()
+        ax.hist(self.statisticss['hist_reaches'],bins,cumulative=True,density=True,histtype=u'step',label='base')
+        ax.set_xlabel(_('River'))
+
+        for curlist in LISTDEM:
+            curdict=self.statisticss[curlist]
+            ax.hist(curdict['hist_reaches'],bins,cumulative=True,density=True,histtype=u'step',label=curlist)
+
+        ax.legend()
+        fig.canvas.draw()
+
+    def analyze_flatzones(self):
+        """
+        Analyse des zones de plat
+        """
+        self.myplotterflat = PlotNotebook()
+
+        ### Flat zones
+        eps=1e-7
+        #indexation des zones "indépendantes" de plats - ruissellement
+        maxindex=self.index_flatzone(self.runoff,eps)
+        #identification des mailles dans les zones
+        myflatnodes=self.find_flatnodes(self.runoff)
+        #création de listes avec les noeuds dans chaque zone
+        myflats=self.find_flatzones(myflatnodes,maxindex)
+
+        #calcul de la longueur de la zone de plat --> sommation du nombre de mailles
+        lenflats=np.zeros((maxindex),dtype=np.int32)
+        for i in range(maxindex):
+            lenflats[i]=len(myflats[i])
+
+        #indexation des zones "indépendantes" de plats - rivières
+        maxindexrivers=self.index_flatzone(self.rivers,eps)
+        #création de listes avec les noeuds dans chaque zone - rivières
+        myflatsrivers=self.find_flatzones(self.rivers,maxindexrivers)
+
+        #calcul de la longueur de la zone de plat --> sommation du nombre de mailles
+        lenflatsrivers=np.zeros((maxindexrivers),dtype=np.int32)
+        for i in range(maxindexrivers):
+            lenflatsrivers[i]=len(myflatsrivers[i])
+
+        fig:mplfig.Figure
+        fig=self.myplotterflat.add("Nb nodes in flat area")
+        ax=fig.add_ax()
+        mybins=np.arange(0.5,np.max(lenflats),1.)
+        myticks=np.arange(1,np.ceil(np.max(lenflats)),1)
+        ax.hist(lenflats,bins=mybins)
+        ax.set_xlabel(_('Nb nodes in flat area - runoff'))
+        ax.set_xticks(myticks)
+        ax.set_xbound(.5,np.max(lenflats))
+        ax.set_ylabel('Nb flat areas')
+        ax.set_yscale('log')
+
+        ax=fig.add_ax()
+        mybinsrivers=np.arange(0.5,np.max(lenflatsrivers),1.)
+        myticksrivers=np.arange(1,np.ceil(np.max(lenflatsrivers)),1)
+        ax.hist(lenflatsrivers,bins=mybinsrivers)
+        ax.set_xlabel(_('Nb nodes in flat area - rivers'))
+        ax.set_xticks(myticksrivers)
+        ax.set_xbound(.5,np.max(lenflatsrivers))
+        ax.set_ylabel('Nb flat areas')
+        ax.set_yscale('log')
+
+        fig=self.myplotterflat.add("Nb nodes in flat area")
+        ax=fig.add_ax()
+        ax.hist(lenflats,bins=mybins,cumulative=True,density=True)
+        ax.set_xlabel(_('Nb nodes in flat area - runoff'))
+        ax.set_xticks(myticks)
+        ax.set_xbound(.5,np.max(lenflats))
+        ax.set_ylabel('Cumulative flat areas')
+        #ax.set_yscale('log')
+
+        ax=fig.add_ax()
+        ax.hist(lenflatsrivers,bins=mybinsrivers,cumulative=True,density=True)
+        ax.set_xlabel(_('Nb nodes in flat area - rivers'))
+        ax.set_xticks(myticksrivers)
+        ax.set_xbound(.5,np.max(lenflatsrivers))
+        ax.set_ylabel('Cumulative flat areas')
+        #ax.set_yscale('log')
+        fig.canvas.draw()
+
+        #Tri des pentes dans différentes listes
+
+        #toutes les mailles
+        sdown=[]
+        sup=[]
+        for curflat in myflats:
+            for curnode in curflat:
+                #recherche de la pente aval plus grande que le seuil
+                sdown.append(curnode.slope_down(eps))
+                #recherche de la pente amont moyenne - uniquement pour les mailles qui ont une pente supérieure au seuil
+                sup.append(curnode.mean_slope_up(eps))
+
+        sflat=[]
+        sdownraw=[]
+        for curflat in myflats:
+            for curnode in curflat:
+                #pente de la maille aval
+                sdownraw.append(curnode.down.slope)
+                #pente courante
+                sflat.append(curnode.slope)
+
+        #mailles rivières
+        sdownriv=[]
+        supriv=[]
+        suponlyriv=[]
+        for curflat in myflatsrivers:
+            for curnode in curflat:
+                #recherche de la pente aval plus grande que le seuil
+                sdownriv.append(curnode.slope_down(eps))
+                #recherche de la pente amont moyenne - uniquement pour les mailles qui ont une pente supérieure au seuil
+                supriv.append(curnode.mean_slope_up(eps))
+                #recherche de la pente amont > seuil
+                suponlyriv.append(curnode.slope_upriver(eps))
+
+        sdownd8=[]
+        suponlyriv1=[]
+        for curflat in myflatsrivers:
+            for curnode in curflat:
+                #pente aval selon voisines D8
+                sdownd8.append(curnode.sloped8)
+                #recherche de la pente amont > seuil
+                suponlyriv1.append(curnode.slope_upriver(eps))
+
+        sflatriver=[]
+        sdownrawriver=[]
+        sd8rawriver=[]
+        for curflat in myflatsrivers:
+            if len(curflat)==1:
+                for curnode in curflat:
+                    if not curnode.down is None:
+                        sd8rawriver.append(curnode.sloped8)
+                        sdownrawriver.append(curnode.down.slope)
+                        sflatriver.append(curnode.slope)
+
+
+        #tracage des graphiques
+        fig=self.myplotterflat.add("Scatter plots")
+        ax=fig.add_ax()
+        ax.scatter(sdownrawriver,sflatriver,marker='o',label='slope down vs flat slope')
+        ax.scatter(sdownriv,suponlyriv,marker='+',label='slope down vs slope d8')
+        ax=fig.add_ax()
+        ax.scatter(sdownraw,sflat,marker='0',label='slope down vs flat slope')
+        ax.scatter(sdown,sup,marker='+',label='slope down vs slope up')
+        fig.canvas.draw()
+
+        fig=self.myplotterflat.add("Scatter plots 2")
+        curax=fig.add_ax()
+        curax.scatter(sdown,sup,marker='+')
+        curax.set_xlabel(_('Slope down [-]'))
+        curax.set_ylabel(_('Mean slope up [-]'))
+        curax.set_aspect('equal','box')
+        curax.set_xbound(0,.55)
+        curax.set_ybound(0,.55)
+        curax.set_title('Runoff')
+
+        curax=fig.add_ax()
+        curax.scatter(sdownriv,supriv,marker='+')
+        curax.set_xlabel(_('Slope down [-]'))
+        curax.set_ylabel(_('Mean slope up [-]'))
+        curax.set_aspect('equal','box')
+        curax.set_xbound(0,.55)
+        curax.set_ybound(0,.55)
+        curax.set_title('River')
+
+        curax=fig.add_ax()
+        curax.scatter(sdownriv,suponlyriv,marker='+')
+        curax.set_xlabel(_('Slope down [-]'))
+        curax.set_ylabel(_('Slope up only river [-]'))
+        curax.set_aspect('equal','box')
+        curax.set_xbound(0,.55)
+        curax.set_ybound(0,.55)
+        curax.set_title('River')
+
+        curax=fig.add_ax()
+        curax.scatter(sdownd8,suponlyriv1,marker='+')
+        curax.set_xlabel(_('Slope D8 [-]'))
+        curax.set_ylabel(_('Slope up only river [-]'))
+        curax.set_aspect('equal','box')
+        curax.set_xbound(0,.3)
+        curax.set_ybound(0,.3)
+        curax.set_title('River')
+        fig.canvas.draw()