def lines2bands(pnl, wbands=None):
'''
Cast PNL into wbands which might not be continuous,
monotonic, and non-overlapping, like CLIRAD-SW\'s wavenumber bands
INPUT:
wbands -- dictionary of {id: [(V1, V2), ...]}, where ID is a unique
label for a wavenumber band, and [(V1, V2), ...] is a list
of one or more wavenumber ranges in the wavenumber band, and
V1 and V2 are the lower and upper wavenumbers of each range,
respectively
'''
dict_pnlout = {}
for id, wranges in wbands.items():
V1s, V2s = zip(*wranges)
pnl_id = aeranalyse.lines2bands(pnl, wbands=wranges)
df_id = aeranalyse.sum_OUTPUT_RADSUM_over_wbands(pnl_id,
V1=min(V1s),
V2=max(V2s))
dict_pnlout[id] = df_id
return pd.Panel(dict_pnlout)