Atmospheric river integrated vapor transport ground truth: dropsonde observations compared to reanalyses

Abstract:

Atmospheric reanalyses are often used as ‘truth’ in many areas of research including studies of atmospheric rivers. Modern reanalyses assimilate over a million observations each day, principally from satellites, but also conventional observations such as land stations and buoys. These reanalyses are vital to advancing the physical understanding of atmospheric rivers as well as assessing model performance and forecast verification. However, as reanalyses are model output they have inherent uncertainties and errors.

Here, we carry out a comprehensive comparison of integrated vapor transport (IVT) in a suite of atmospheric reanalyses including ERA5 and MERRA2 to dropsonde observations over the northeast Pacific. Using over 1000 dropsondes from 2014 to 2020 we examine the biases and random errors associated with each reanalysis and explore sensitivity to IVT calculation method, i.e. using pressure levels, model levels and vertically integrated model output.

We find that overall the reanalyses perform well simulating IVT at the dropsonde locations, with ERA5 outperforming the other reanalysis products. We also find that the differences in IVT calculated using the different model output are non-negligible.

The results from this study provide not only the atmospheric river research community with a thorough assessment of how well reanalyses simulate IVT, but also the wider atmospheric science community on reanalysis model performance.