When I was running this code, some of the biases counted toward the mean were very large (~ -8.0), but had only a few points, some even just 2 points, which meant often the standard deviation was low enough (<minSig) to be included in the mean.
It would be nice if a threshold could be included to mask files that also had a low number of points. I am still learning python, so there may be a better way to do this, but I did:
if zdr_bias['volume_standard_deviation'] > minSig:
Stats[index,0] = np.ma.masked
Stats[index,1] = np.ma.masked
Stats[index,2] = np.ma.masked
elif zdr_bias['volume_number_good'] < 1000.0:
Stats[index,0] = np.ma.masked
Stats[index,1] = np.ma.masked
Stats[index,2] = np.ma.masked
else:
Stats[index,0] = zdr_bias['volume_average']
Stats[index,1] = zdr_bias['volume_standard_deviation']
Stats[index,2] = zdr_bias['volume_number_good']
When I was running this code, some of the biases counted toward the mean were very large (~ -8.0), but had only a few points, some even just 2 points, which meant often the standard deviation was low enough (<minSig) to be included in the mean.
It would be nice if a threshold could be included to mask files that also had a low number of points. I am still learning python, so there may be a better way to do this, but I did: