def correlations_lag_lat_or_lon(E,maxlag,lat_or_lon = 'lon',filter_order=50,climatology_option='NODA',hostname='taurus',verbose=False):
"""
compute correlations between U850 or OLR in a reference are and everywhere else,
as a function of lag and either latitude or longitude
INPUTS:
E - a standard DART experiment dictionary, with the variable field and level range corresponding to some MJO variable
maxlag: the limit of the lag (in days) that we look at
lat_or_lon: choose dimension to preserve after averaging -- 'lat' or 'lon'
climatology_option: choose which climatology to take the anomalies to respect with -- default is "NODA"
"""
# change the given daterange to daily resolution, because the lag is specified in days
E['daterange'] = dart.change_daterange_to_daily(E['daterange'])
# compute or load the daily climatology and deviation from climatology
anomalies,climatology,lat,lon,lev,DRnew = ano(E,climatology_option = climatology_option,hostname=hostname,verbose=verbose)
# filter daily anomalies using a Lanczos filter
AA,FA = filter(anomalies,filter_order,return_as_vector=False)
if E['variable'] == 'U':
variable_name = 'U'+str(E['levrange'][0])
else:
variable_name = E['variable']
# compute the zonal and meridional mean of the resulting field
# the regions we average over depend on whether we want lag-lat, or lag-lon plots
# also, note thatm by how the filtered anomalies are constructed, the 3rd dimension is always time
if lat_or_lon == "lon":
# select latitudes 10S-10N and average meridionally, then plot correlations as a function of lon
lat1,lon1,FAm = aave('TB',FA,lat,lon,None,variable_name,averaging_dimension='lat')
if lat_or_lon == "lat":
# average over the longitude corridor 80-100E and plot correlations as a function of lat
lat1,lon1,FAm = aave('ZB',FA,lat,lon,None,variable_name,averaging_dimension='lon')
# area averaging the desired variable over the Indian Ocean reference point
if (E['daterange'][0].month >= 10) or (E['daterange'][0].month <= 2):
season = 'winter'
else:
season = 'summer'
lat0,lon0,FA0 = aave('IO',FA,lat,lon,season,variable_name,averaging_dimension="all")
#------ compute field of correlation coefficients
# empty array size Lag by Lat
# plus an array to keep track of sample size
Lag_range = range(-maxlag,maxlag+1)
nlag = len(Lag_range)
n = FAm.shape[0]
R = np.zeros(shape=(nlag,n))
S = np.zeros(shape=(nlag,n))
# loop over latitudes
T = len(FA0)
for ii in range(n):
# loop over lags
for ilag,L in zip(range(nlag),Lag_range):
# the time points that we can check go from L to T-L
# so shorter lags have a larger sample size and are more significant.
if L < 0:
Tsel = range(-L,T)
if L > 0:
Tsel = range(0,T-L)
if L == 0:
Tsel = range(0,T)
# loop over the available time points and gather values to compare
IO = []
X = []
for k in Tsel:
IO.append(FA0[k+L])
X.append(FAm[ii,k])
# now compute the correlation from this list of samples and store in the lag vs lat array
rho = np.corrcoef(X,IO)
if rho != []:
R[ilag,ii] = rho[1,0]
S[ilag,ii] = len(IO)
else:
R[ilag,ii] = np.nan
S[ilag,ii] = np.nan
if lat_or_lon == 'lon':
space_dim = lon1
if lat_or_lon == 'lat':
space_dim = lat1
L = np.array(Lag_range)
return R,S,L,space_dim
