Closed bradyrx closed 4 years ago
Currently the temporal.to_annual() replaces any NaNs in a DataArray with zeros after converting. My current workaround is to replace the original NaNs afterwards with a simple .where() but that shouldn't be the standard.
temporal.to_annual()
.where()
import numpy as np import xarray as xr import esmtools as et np.random.seed(999) space = np.arange(5) time = xr.cftime_range('1990-01', '1992-01', freq='MS', calendar='noleap') da = xr.DataArray(np.random.rand(len(time), len(space)), dims=["time", "space"], coords=[time, space]) da[:, 0:1] = np.nan print(da) >>> <xarray.DataArray (time: 25, space: 5)> array([[ nan, 0.5275223 , 0.11911147, 0.63968144, 0.09092526], [ nan, 0.42738095, 0.55438581, 0.62812652, 0.69739294], [ nan, 0.13189035, 0.34277045, 0.20155961, 0.70732423], [ nan, 0.90925004, 0.40516066, 0.76043547, 0.47375838], [ nan, 0.75129249, 0.09708994, 0.41235779, 0.28163896], [ nan, 0.87110921, 0.08124512, 0.55793117, 0.54753428], [ nan, 0.97326881, 0.2862761 , 0.5082575 , 0.14795074], [ nan, 0.84082001, 0.0037532 , 0.78262101, 0.83347772], [ nan, 0.97260166, 0.83282304, 0.06581761, 0.40379256], [ nan, 0.50750135, 0.97787696, 0.81899021, 0.18754124], [ nan, 0.68261077, 0.99909815, 0.48263116, 0.73059268], [ nan, 0.26139168, 0.16107376, 0.69850315, 0.89950917], [ nan, 0.31244902, 0.95412616, 0.7242641 , 0.02091039], [ nan, 0.58165923, 0.9545687 , 0.74233195, 0.19750339], [ nan, 0.85836332, 0.44904621, 0.82365038, 0.99726878], [ nan, 0.5890016 , 0.42402702, 0.89548786, 0.44437266], [ nan, 0.66019353, 0.30244304, 0.02295771, 0.83766937], [ nan, 0.37552193, 0.18172362, 0.83135182, 0.18487429], [ nan, 0.69644561, 0.60566253, 0.49600661, 0.70888438], [ nan, 0.65267488, 0.62297362, 0.83609334, 0.3572364 ], [ nan, 0.06336003, 0.86713576, 0.55147501, 0.8895871 ], [ nan, 0.20783523, 0.48823397, 0.59087233, 0.87087296], [ nan, 0.95453522, 0.17335265, 0.106837 , 0.6271594 ], [ nan, 0.65200627, 0.68657226, 0.47579499, 0.66581477], [ nan, 0.18154475, 0.70021069, 0.44733626, 0.52452094]]) Coordinates: * time (time) object 1990-01-01 00:00:00 ... 1992-01-01 00:00:00 * space (space) int64 0 1 2 3 4 da = et.temporal.to_annual(da) print(da) >>> <xarray.DataArray (year: 3, space: 5)> array([[0. , 0.654351 , 0.4019154 , 0.54661122, 0.49807322], [0. , 0.55067952, 0.55752298, 0.59013334, 0.57021443], [0. , 0.18154475, 0.70021069, 0.44733626, 0.52452094]]) Coordinates: * space (space) int64 0 1 2 3 4 * year (year) int64 1990 1991 1992
Currently the
temporal.to_annual()replaces any NaNs in a DataArray with zeros after converting. My current workaround is to replace the original NaNs afterwards with a simple.where()but that shouldn't be the standard.