add some comments for pbm_norm config

Author: OuyangWenyu

torchhydro/configs/config.py

@@ -204,6 +204,7 @@ def default_config_file():
                     "ssma",
                     "susma",
                 ],
+                # NOTE: pbm_norm is True means norm and denorm for differentiable models; if you use pure data-driven models, you should set it as False
                 "pbm_norm": False,
             },
             "stat_dict_file": None,

torchhydro/datasets/data_scalers.py

@@ -320,7 +320,7 @@ def inverse_transform(self, target_values):
         stat_dict = self.stat_dict
         target_cols = self.data_cfgs["target_cols"]
         if self.pbm_norm:
-            # for pbm's output, its unit is mm/day, so we don't need to recover its unit
+            # for (differentiable models) pbm's output, its unit is mm/day, so we don't need to recover its unit
             pred = target_values
         else:
             pred = _trans_norm(