Representation of Dropsonde?Observed Atmospheric River Conditions in Reanalyses

Atmospheric rivers (ARs) are the primary mechanism for mid-latitude water vapor transport, and identified by a key variable, integrated transport (IVT). The ability of atmospheric reanalyses in providing ground-truth dataset IVT field is assessed comparing ERA5, MERRA-2, JRA-55 data against large sample (>1,700) dropsonde profiles deployed around ARs. Bias error increase with magnitude, although asymmetrically AR core. A partitioning source reveals that humidity contributes more to difference above 800 hPa, while wind dominant lowest levels (to 950 hPa). This quantification reanalysis bias identifies ERA5 as errors demonstrates remaining challenges representing observed state