diff --git a/src/plotting/postprocessing_plotting.py b/src/plotting/postprocessing_plotting.py
index b39de8e957a110121c0e8812608d32aad3431570..854182613cdb63456dc8f62d2421560d829ee629 100644
--- a/src/plotting/postprocessing_plotting.py
+++ b/src/plotting/postprocessing_plotting.py
@@ -639,15 +639,16 @@ class PlotTimeSeries(RunEnvironment):
     def _plot_ahead(self, ax, data):
         color = sns.color_palette("Blues_d", self._window_lead_time).as_hex()
         for ahead in data.coords["ahead"].values:
-            plot_data = data.sel(type="CNN", ahead=ahead).drop(["type", "ahead"]).squeeze()
-            index = plot_data.index + np.timedelta64(int(ahead), self._sampling)
+            plot_data = data.sel(type="CNN", ahead=ahead).drop(["type", "ahead"]).squeeze().shift(index=ahead)
             label = f"{ahead}{self._sampling}"
-            ax.plot(index, plot_data.values, color=color[ahead-1], label=label)
+            ax.plot(plot_data, color=color[ahead-1], label=label)
 
     def _plot_obs(self, ax, data):
-        obs_data = data.sel(type="obs", ahead=1)
-        index = data.index + np.timedelta64(1, self._sampling)
-        ax.plot(index, obs_data.values, color=matplotlib.colors.cnames["green"], label="obs")
+        ahead = 1
+        obs_data = data.sel(type="obs", ahead=ahead).shift(index=ahead)
+        # index = data.index + np.timedelta64(1, self._sampling)
+        # ax.plot(index, obs_data.values, color=matplotlib.colors.cnames["green"], label="obs")
+        ax.plot(obs_data, color=matplotlib.colors.cnames["green"], label="obs")
 
     @staticmethod
     def _get_time_range(data):
diff --git a/src/run_modules/post_processing.py b/src/run_modules/post_processing.py
index 962c9f52065729381ce11e8a8adcbeed45a4c011..00d3d2d83409bccf478024b553ac889a4517424b 100644
--- a/src/run_modules/post_processing.py
+++ b/src/run_modules/post_processing.py
@@ -162,7 +162,7 @@ class PostProcessing(RunEnvironment):
 
             for normalised in [True, False]:
                 # create empty arrays
-                nn_prediction, persistence_prediction, ols_prediction = self._create_empty_prediction_arrays(data, count=3)
+                nn_prediction, persistence_prediction, ols_prediction, observation = self._create_empty_prediction_arrays(data, count=4)
 
                 # nn forecast
                 nn_prediction = self._create_nn_forecast(input_data, nn_prediction, mean, std, transformation_method, normalised)
@@ -175,7 +175,7 @@ class PostProcessing(RunEnvironment):
                 ols_prediction = self._create_ols_forecast(input_data, ols_prediction, mean, std, transformation_method, normalised)
 
                 # observation
-                observation = self._create_observation(data, None, mean, std, transformation_method, normalised)
+                observation = self._create_observation(data, observation, mean, std, transformation_method, normalised)
 
                 # merge all predictions
                 full_index = self.create_fullindex(data.data.indexes['datetime'], self._get_frequency())
@@ -195,12 +195,19 @@ class PostProcessing(RunEnvironment):
         getter = {"daily": "1D", "hourly": "1H"}
         return getter.get(self._sampling, None)
 
-    @staticmethod
-    def _create_observation(data, _, mean, std, transformation_method, normalised):
+    def _create_observation(self, data, observation, mean, std, transformation_method, normalised):
         obs = data.observation.copy()
         if not normalised:
             obs = statistics.apply_inverse_transformation(obs, mean, std, transformation_method)
-        return obs
+        window_lead_time = self.data_store.get("window_lead_time", "general")
+        obs_w = []
+        for w in range(window_lead_time):
+            obs_w.append(obs.shift(datetime=-(w+1)))
+        if observation is None:
+            observation = data.label.copy()
+        observation.values = np.concatenate(obs_w, axis=0)
+        return observation
+
 
     def _create_ols_forecast(self, input_data, ols_prediction, mean, std, transformation_method, normalised):
         tmp_ols = self.ols_model.predict(input_data)
@@ -212,7 +219,7 @@ class PostProcessing(RunEnvironment):
         return ols_prediction
 
     def _create_persistence_forecast(self, data, persistence_prediction, mean, std, transformation_method, normalised):
-        tmp_persi = data.observation.copy().sel({'window': 0})
+        tmp_persi = data.observation.copy().sel({'window': 0})#.shift(datetime=1)
         if not normalised:
             tmp_persi = statistics.apply_inverse_transformation(tmp_persi, mean, std, transformation_method)
         window_lead_time = self.data_store.get("window_lead_time", "general")