Fix pandas-compatible scalar assignment in tdating

pysteps/io/exporters.py

@@ -100,6 +100,10 @@
 try:
     from osgeo import gdal, osr
 
+    # Preserve current behavior explicitly (no GDAL exceptions) and avoid the
+    # GDAL 4.0 future-warning emitted when neither mode is selected.
+    if hasattr(gdal, "DontUseExceptions"):
+        gdal.DontUseExceptions()
     GDAL_IMPORTED = True
 except ImportError:
     GDAL_IMPORTED = False

pysteps/io/importers.py

@@ -104,6 +104,10 @@
 try:
     from osgeo import gdal, gdalconst, osr
 
+    # Preserve current behavior explicitly (no GDAL exceptions) and avoid the
+    # GDAL 4.0 future-warning emitted when neither mode is selected.
+    if hasattr(gdal, "DontUseExceptions"):
+        gdal.DontUseExceptions()
     GDAL_IMPORTED = True
 except ImportError:
     GDAL_IMPORTED = False
@@ -1615,11 +1619,11 @@ def _import_saf_crri_data(filename, idx_x=None, idx_y=None):
     ds_rainfall = netCDF4.Dataset(filename)
     if "crr_intensity" in ds_rainfall.variables.keys():
         if idx_x is not None:
-            data = np.array(ds_rainfall.variables["crr_intensity"][idx_y, idx_x])
-            quality = np.array(ds_rainfall.variables["crr_quality"][idx_y, idx_x])
+            data = np.asarray(ds_rainfall.variables["crr_intensity"][idx_y, idx_x])
+            quality = np.asarray(ds_rainfall.variables["crr_quality"][idx_y, idx_x])
         else:
-            data = np.array(ds_rainfall.variables["crr_intensity"])
-            quality = np.array(ds_rainfall.variables["crr_quality"])
+            data = np.asarray(ds_rainfall.variables["crr_intensity"][:])
+            quality = np.asarray(ds_rainfall.variables["crr_quality"][:])
         precipitation = np.where(data == 65535, np.nan, data)
     else:
         precipitation = None

pysteps/nowcasts/interface.py

@@ -1,4 +1,4 @@
-"""
+r"""
 pysteps.nowcasts.interface
 ==========================
 
@@ -55,7 +55,7 @@
 
 
 def get_method(name):
-    """
+    r"""
     Return a callable function for computing nowcasts.
 
     Description:

pysteps/scripts/fit_vel_pert_params.py

@@ -74,8 +74,8 @@
         (l1,) = pyplot.plot(t, f(t, *p_par), "r-")
         (l2,) = pyplot.plot(t, f(t, *p_perp), "g-")
 
-    p_str_1 = lambda p: "%.2f\cdot t^{%.2f}+%.2f" % (p[0], p[1], p[2])
-    p_str_2 = lambda p: "%.2f\cdot t^{%.2f}%.2f" % (p[0], p[1], p[2])
+    p_str_1 = lambda p: "%.2f\\cdot t^{%.2f}+%.2f" % (p[0], p[1], p[2])
+    p_str_2 = lambda p: "%.2f\\cdot t^{%.2f}%.2f" % (p[0], p[1], p[2])
     if fit_succeeded:
         lbl = lambda p: p_str_1(p) if p[2] > 0.0 else p_str_2(p)
         pyplot.legend(

pysteps/timeseries/correlation.py

@@ -1,5 +1,5 @@
 # -*- coding: utf-8 -*-
-"""
+r"""
 pysteps.timeseries.correlation
 ==============================
 
@@ -28,7 +28,7 @@ def temporal_autocorrelation(
     window="gaussian",
     window_radius=np.inf,
 ):
-    """
+    r"""
     Compute lag-l temporal autocorrelation coefficients
     :math:`\gamma_l=\mbox{corr}(x(t),x(t-l))`, :math:`l=1,2,\dots,n-1`,
     from a time series :math:`x_1,x_2,\dots,x_n`. If a multivariate time series
@@ -133,7 +133,7 @@ def temporal_autocorrelation(
 def temporal_autocorrelation_multivariate(
     x, d=0, mask=None, window="gaussian", window_radius=np.inf
 ):
-    """
+    r"""
     For a :math:`q`-variate time series
     :math:`\mathbf{x}_1,\mathbf{x}_2,\dots,\mathbf{x}_n`, compute the lag-l
     correlation matrices :math:`\mathbf{\Gamma}_l`, where

pysteps/tracking/tdating.py

@@ -381,17 +381,18 @@ def advect(cells_id, labels, V1, output_splits_merges=False):
         new_cen_x = cell.cen_x + ad_x
         new_cen_y = cell.cen_y + ad_y
 
-        cells_ad.loc[ID, "x"] = new_x
-        cells_ad.loc[ID, "y"] = new_y
-        cells_ad.loc[ID, "flowx"] = ad_x
-        cells_ad.loc[ID, "flowy"] = ad_y
-        cells_ad.loc[ID, "cen_x"] = new_cen_x
-        cells_ad.loc[ID, "cen_y"] = new_cen_y
-        cells_ad.loc[ID, "ID"] = cell.ID
+        # Use scalar cell assignment (.at) to store array-like payloads in object columns.
+        cells_ad.at[ID, "x"] = new_x
+        cells_ad.at[ID, "y"] = new_y
+        cells_ad.at[ID, "flowx"] = ad_x
+        cells_ad.at[ID, "flowy"] = ad_y
+        cells_ad.at[ID, "cen_x"] = new_cen_x
+        cells_ad.at[ID, "cen_y"] = new_cen_y
+        cells_ad.at[ID, "ID"] = cell.ID
 
         cell_unique = np.zeros(labels.shape)
         cell_unique[new_y, new_x] = 1
-        cells_ad.loc[ID, "cont"] = skime.find_contours(cell_unique, 0.8)
+        cells_ad.at[ID, "cont"] = skime.find_contours(cell_unique, 0.8)
 
     return cells_ad
 
@@ -411,7 +412,7 @@ def match(cells_ad, labels, match_frac=0.4, split_frac=0.1, output_splits_merges
         IDs = np.unique(ID_vec)
         n_IDs = len(IDs)
         if n_IDs == 1 and IDs[0] == 0:
-            cells_ov.loc[ID_a, "t_ID"] = 0
+            cells_ov.at[ID_a, "t_ID"] = 0
             continue
         IDs = IDs[IDs != 0]
         n_IDs = len(IDs)
@@ -429,18 +430,18 @@ def match(cells_ad, labels, match_frac=0.4, split_frac=0.1, output_splits_merges
             # splits here
             if sum(valid_split_ids) > 1:
                 # Save split information
-                cells_ov.loc[ID_a, "splitted"] = True
-                cells_ov.loc[ID_a, "split_IDs"] = IDs[valid_split_ids]
-                cells_ov.loc[ID_a, "split_fracs"] = N / len(ID_vec)
+                cells_ov.at[ID_a, "splitted"] = True
+                cells_ov.at[ID_a, "split_IDs"] = IDs[valid_split_ids].tolist()
+                cells_ov.at[ID_a, "split_fracs"] = (N / len(ID_vec)).tolist()
 
         m = np.argmax(N)
         ID_match = IDs[m]
         ID_coverage = N[m] / len(ID_vec)
         if ID_coverage >= match_frac:
-            cells_ov.loc[ID_a, "t_ID"] = ID_match
+            cells_ov.at[ID_a, "t_ID"] = ID_match
         else:
-            cells_ov.loc[ID_a, "t_ID"] = 0
-        cells_ov.loc[ID_a, "frac"] = ID_coverage
+            cells_ov.at[ID_a, "t_ID"] = 0
+        cells_ov.at[ID_a, "frac"] = ID_coverage
     return cells_ov, labels, possible_merge_ids