Characterizing ERA-Interim and ERA5 surface wind biases using ASCAT

Abstract

Abstract. This paper analyzes the differences between ERA-Interim\nand ERA5 surface winds fields relative to Advanced Scatterometer (ASCAT) ocean vector wind\nobservations, after adjustment for the effects of atmospheric stability and\ndensity, using stress-equivalent winds\xa0(U10S) and air–sea relative motion\nusing ocean current velocities. In terms of instantaneous root mean square (rms) wind speed\nagreement, ERA5 winds show a 20\u2009% improvement relative to ERA-Interim\nand a performance similar to that of currently operational ECMWF forecasts.\nERA5 also performs better than ERA-Interim in terms of mean and transient wind\nerrors, wind divergence and wind stress curl biases. Yet, both ERA products\nshow systematic errors in the partition of the wind kinetic energy into\nzonal and meridional, mean and transient components. ERA winds are\ncharacterized by excessive mean zonal winds (westerlies) with too-weak mean\npoleward flows in the midlatitudes and too-weak mean meridional winds (trades)\nin the tropics. ERA stress curl is too cyclonic in midlatitudes and high latitudes,\nwith implications for Ekman upwelling estimates, and lacks detail in the\nrepresentation of sea surface temperature (SST) gradient effects (along the equatorial cold tongues\nand Western Boundary Current (WBC) jets) and mesoscale convective airflows (along the Intertropical Convergence Zone and the warm\nflanks for the WBC jets). It is conjectured that large-scale mean wind\nbiases in ERA are related to their lack of high-frequency (transient wind)\nvariability, which should be promoting residual meridional circulations in\nthe Ferrel and Hadley cells.