Newer versions of Matplotlib seem to fail to plot map scatter when al…

…l nan (#82)

Co-authored-by: danholdaway <danholdaway@users.noreply.github.com>

setup.py

@@ -16,7 +16,7 @@
 
 setuptools.setup(
     name='eva',
-    version='1.3.3',
+    version='1.3.4',
     author='Community owned code',
     description='Evaluation and Verification of an Analysis',
     url='https://github.com/JCSDA-internal/eva',

src/eva/diagnostics/map_scatter.py

@@ -2,6 +2,7 @@
 from eva.utilities.utils import get_schema, update_object, slice_var_from_str
 import emcpy.plots.map_plots
 import os
+import numpy as np
 
 
 class MapScatter():
@@ -25,6 +26,11 @@ def __init__(self, config, logger, dataobj):
         lonvar = lonvar.flatten()
         latvar = latvar.flatten()
         datavar = datavar.flatten()
+
+        # If everything is nan plotting will fail so just plot some large values
+        if np.isnan(datavar).all():
+            datavar[np.isnan(datavar)] = 1.0e38
+
         # create declarative plotting MapScatter object
         self.plotobj = emcpy.plots.map_plots.MapScatter(latvar, lonvar, datavar)
         # get defaults from schema

src/eva/plot_tools/dynamic_config.py

@@ -54,8 +54,14 @@ def vminvmaxcmap(logger, option_dict, plots_dict, data_collections):
 
     # Find minimum and maximum values
     cmap = option_dict.get('sequential colormap', 'viridis')
-    vmax = np.nanmax(datavar_check)
-    vmin = np.nanmin(datavar_check)
+
+    # If everything is nan plot some large min/max (plotting code should do the same)
+    if np.isnan(datavar_check).all():
+        vmax = 1.0e38
+        vmin = 1.0e38
+    else:
+        vmax = np.nanmax(datavar_check)
+        vmin = np.nanmin(datavar_check)
 
     # If positive and negative values are present then a diverging colormap centered on zero should
     # be used.

src/eva/transforms/transform_utils.py

@@ -47,11 +47,11 @@ def replace_cgv(logger, collection, group, variable, *argv):
 
     # Create dictionary with templates
     tmplt_dict = {}
-    if collection is not 'none':
+    if collection != 'none':
         tmplt_dict['collection'] = collection
-    if group is not 'none':
+    if group != 'none':
         tmplt_dict['group'] = group
-    if variable is not 'none':
+    if variable != 'none':
         tmplt_dict['variable'] = variable
 
     # Perform replace for the argument list