Small improvements

Author: dnerini

examples/plot_extrapolation_nowcast.py

@@ -51,10 +51,11 @@
 # Plot the rainfall field
 plot_precip_field(R[-1, :, :], geodata=metadata)
 
-# Store the last frame for polotting it later later
+# Store the last frame for plotting it later later
 R_ = R[-1, :, :].copy()
 
-# Log-transform the data
+# Log-transform the data to unit of dBR, set the threshold to 0.1 mm/h,
+# set the fill value to -15 dBR
 R, metadata = transformation.dB_transform(R, metadata, threshold=0.1, zerovalue=-15.0)
 
 # Nicely print the metadata
@@ -111,12 +112,12 @@
 # Compute fractions skill score (FSS) for all lead times, a set of scales and 1 mm/h
 fss = verification.get_method("FSS")
 scales = [2, 4, 8, 16, 32, 64, 128, 256, 512]
-threshold = 1.0
+thr = 1.0
 score = []
 for i in range(n_leadtimes):
     score_ = []
     for scale in scales:
-        score_.append(fss(R_f[i, :, :], R_o[i + 1, :, :], threshold, scale))
+        score_.append(fss(R_f[i, :, :], R_o[i + 1, :, :], thr, scale))
     score.append(score_)
 
 figure()
@@ -126,5 +127,5 @@
 xlabel("Lead time [min]")
 ylabel("FSS ( > 1.0 mm/h ) ")
 title("Fractions skill score")
-show()
-# sphinx_gallery_thumbnail_number = 2
+
+# sphinx_gallery_thumbnail_number = 3

examples/plot_steps_nowcast.py

@@ -10,6 +10,7 @@
 
 from pylab import *
 from datetime import datetime
+from pprint import pprint
 from pysteps import io, nowcasts, rcparams
 from pysteps.motion.lucaskanade import dense_lucaskanade
 from pysteps.postprocessing.ensemblestats import excprob
@@ -29,7 +30,6 @@
 # transformed into units of dBR.
 
 
-
 date = datetime.strptime("201701311200", "%Y%m%d%H%M")
 data_source = "mch"
 
@@ -44,7 +44,7 @@
 
 # Find the radar files in the archive
 fns = io.find_by_date(
-    date, root_path, path_fmt, fn_pattern, fn_ext, timestep, num_prev_files=2,
+    date, root_path, path_fmt, fn_pattern, fn_ext, timestep, num_prev_files=2
 )
 
 # Read the data from the archive
@@ -60,12 +60,16 @@
 # Plot the rainfall field
 plot_precip_field(R[-1, :, :], geodata=metadata)
 
-# Log-transform the data to unit of dBR, set the threshold to 0.1 mm/h
-R = transformation.dB_transform(R, threshold=0.1, zerovalue=-15.0)[0]
+# Log-transform the data to unit of dBR, set the threshold to 0.1 mm/h,
+# set the fill value to -15 dBR
+R, metadata = transformation.dB_transform(R, metadata, threshold=0.1, zerovalue=-15.0)
 
 # Set missing values with the fill value
 R[~np.isfinite(R)] = -15.0
 
+# Nicely print the metadata
+pprint(metadata)
+
 ###############################################################################
 # Deterministic nowcast with S-PROG
 # ---------------------------------
@@ -96,24 +100,27 @@
 R_f = transformation.dB_transform(R_f, threshold=-10.0, inverse=True)[0]
 
 # Plot the S-PROG forecast
-plot_precip_field(R_f[-1, :, :], geodata=metadata, title="S-PROG")
+plot_precip_field(
+    R_f[-1, :, :],
+    geodata=metadata,
+    title="S-PROG (+ %i min)" % (n_leadtimes * timestep),
+)
 
 ###############################################################################
 # As we can see from the figure above, the forecast produced by S-PROG is a
 # smooth field. In other words, the forecast variance is lower than the
 # variance of the original observed field.
-# However, given applications demand that the forecast retain the same
+# However, certain applications demand that the forecast retain the same
 # statistical properties of the observations. In such cases, the S-PROG
-# forecsats are of limited use and a stochatic approahc might be of more
+# forecasts are of limited use and a stochatic approach might be of more
 # interest.
 
 ###############################################################################
 # Stochastic nowcast with STEPS
 # -----------------------------
 #
-#
 # The S-PROG approach is extended to include a stochastic term which represents
-# the variance linked to the unpredictable development of precipitation. This
+# the variance associated to the unpredictable development of precipitation. This
 # approach is known as STEPS (short-term ensemble prediction system).
 
 # The STEPES nowcast
@@ -141,25 +148,31 @@
 
 # Plot the ensemble mean
 R_f_mean = np.mean(R_f[:, -1, :, :], axis=0)
-plot_precip_field(R_f_mean, geodata=metadata, title="Ensemble mean")
+plot_precip_field(
+    R_f_mean,
+    geodata=metadata,
+    title="Ensemble mean (+ %i min)" % (n_leadtimes * timestep),
+)
 
 ###############################################################################
 # The mean of the ensemble displays similar properties as the S-PROG
-# forecast seen above, although the degree of smoothing strongly depends on
+# forecast seen above, although the degree of smoothing also depends on
 # the ensemble size. In this sense, the S-PROG forecast can be seen as
 # the mean of an ensemble of infinite size.
 
 # Plot the first two realizations
 fig = figure()
-for i in range(2):
-    ax = fig.add_subplot(121 + i)
+for i in range(4):
+    ax = fig.add_subplot(221 + i)
     ax.set_title("Member %02d" % i)
     plot_precip_field(R_f[i, -1, :, :], geodata=metadata, colorbar=False, axis="off")
 tight_layout()
 
 ###############################################################################
 # As we can see from these two members of the ensemble, the stochastic forecast
 # mantains the same variance as in the observed rainfall field.
+# STEPS also includes a stochatic perturbation of the motion field in order
+# to quantify the its uncertainty.
 
 ###############################################################################
 # Finally, it is possible to derive probabilities from our ensemble forecast.
@@ -175,7 +188,7 @@
     type="prob",
     units="mm/h",
     probthr=0.5,
-    title="Exceedence probability",
+    title="Exceedence probability (+ %i min)" % (n_leadtimes * timestep),
 )
 
 # sphinx_gallery_thumbnail_number = 5