snapshot...

modules/icing/pirep_goes.py

@@ -1490,6 +1490,50 @@ def run_mean_std_2(check_cloudy=False, no_icing_to_icing_ratio=5, params=train_p
     pickle.dump(mean_std_dct, f)
     f.close()
 
+
+def run_mean_std_3(train_file_path, check_cloudy=False, params=train_params_day):
+
+    # params = ['cld_height_acha', 'cld_geo_thick', 'cld_press_acha', 'supercooled_cloud_fraction', 'cld_temp_acha', 'cld_opd_acha',
+    #            'cld_reff_acha', 'cld_reff_dcomp', 'cld_reff_dcomp_1', 'cld_reff_dcomp_2', 'cld_reff_dcomp_3',
+    #            'cld_opd_dcomp', 'cld_opd_dcomp_1', 'cld_opd_dcomp_2', 'cld_opd_dcomp_3', 'cld_cwp_dcomp', 'iwc_dcomp',
+    #            'lwc_dcomp', 'cld_emiss_acha', 'conv_cloud_fraction']
+    #check_cloudy = True
+
+    params = ['temp_10_4um_nom', 'temp_11_0um_nom', 'temp_12_0um_nom', 'temp_13_3um_nom', 'temp_3_75um_nom',
+               'temp_6_2um_nom', 'temp_6_7um_nom', 'temp_7_3um_nom', 'temp_8_5um_nom', 'temp_9_7um_nom',
+               'refl_0_47um_nom', 'refl_0_65um_nom', 'refl_0_86um_nom', 'refl_1_38um_nom', 'refl_1_60um_nom']
+    check_cloudy = False
+
+    mean_std_lo_hi_dct = {}
+
+    h5f = h5py.File(train_file_path, 'r')
+
+    if check_cloudy:
+        cld_msk = h5f['cloud_mask'][:].flatten()
+
+    for dname in params:
+        data = h5f[dname][:,].flatten()
+
+        if check_cloudy:
+            keep = np.logical_or(cld_msk == 2, cld_msk == 3)
+            data = data[keep]
+
+        lo = np.nanmin(data)
+        hi = np.nanmax(data)
+        mean = np.nanmean(data)
+        data -= mean
+        std = np.nanstd(data)
+
+        print(dname,': ', mean, std, lo, hi)
+
+        mean_std_lo_hi_dct[dname] = (mean, std, lo, hi)
+
+    h5f.close()
+
+    f = open('/Users/tomrink/data/icing/mean_std_lo_hi_test.pkl', 'wb')
+    pickle.dump(mean_std_lo_hi_dct, f)
+    f.close()
+
 # def split_data(num_obs, perc=0.2, skip=1, shuffle=True, seed=None):
 #     idxs = np.arange(num_obs)
 #     idxs = list(idxs)