Greetings,
I have discovered what appears to be a bug in the bandpass filter. The error is repeatable using the code below. By commenting out the filter line all is OK so I don't believe the response correction is involved.
If the data component selection is changed to LH* at 1 hz sample rate the filter works but at the MH* 40 sps sample rate it fails and creates NAN in the output.
Randy
################### Code Block###########Error line is commented out########
from obspy.fdsn import Client
from obspy import UTCDateTime
import numpy as np
t = UTCDateTime("2014-12-19T00:00:00.000")
client = Client("IRIS")
st = client.get_waveforms("TA", "MDND", "--", "BH*", t, t + 3600, attach_response=True)
print st
if np.isnan(np.min(st[0].data)):
print 'Warning: nan in st[0]'
if np.isnan(np.min(st[1].data)):
print 'Warning: nan in st[1]'
if np.isnan(np.min(st[2].data)):
print 'Warning: nan in st[2]'
st.remove_response()
##### Error caused by this next line######
#st.filter("bandpass", freqmin = 0.105, freqmax = .125, zerophase = True)
if np.isnan(np.min(st[0].data)):
print 'Warning: nan in filtered trace0'
if np.isnan(np.min(st[1].data)):
print 'Warning: nan in filtered trace1'
if np.isnan(np.min(st[2].data)):
print 'Warning: nan in filtered trace2'