Why is the atmosphere becoming drier? - An investigation of the role of dynamical drivers on recent trends in relative humidity

Relative humidity (RH) over land has declined steeply since 2000. The drying signal is relatively consistent from the edge of the deep tropics to the mid-latitudes of both hemispheres, whereas regions equatorward and poleward show increasing RH trends. The drying trend observed in the gridded global humidity dataset, HadISDH, could not be captured by the CMIP5 climate models [1, 2].

The drying trend finds partial explanation through thermodynamic drivers. Global warming causes an increase in both latent and sensible heat in the atmosphere. Over land, the increase in latent heat is much lower than that of sensible heat. Due to slower warming rates over the ocean compared to land, not sufficient humidity is evaporated and transported towards the coast to keep RH over land constant [3].

Temperature and moisture in many regions are influenced by the atmospheric circulation, therefore can influence RH. In this study, we investigate the potential influence of atmospheric circulation on the observed regional RH changes. We have done this for selected regions with a strong RH trend (including the western US, eastern Brazil, Greenland's coastal areas, southern Africa, the Caspian Sea, Mongolia and Tibet). We firstly calculate correlation and regression coefficients between gridded and regional RH and a range of dynamical drivers (including the Northern and Southern Annular Modes, ENSO and the PDO). We also explored the relationship between regional RH and global fields of sea surface temperature (SST), sea level pressure (SLP), and wind from the ERA-Interim reanalysis. We find a significant relationship between RH and the dynamical drivers in many regions (for example with the ENSO in eastern Brazil), as well as the impact of small-scale atmospheric circulations on land cover change, which then impacts RH (for example evaporation over the Caspian Sea). We will present these results, and try to quantify the contribution of these drivers to recent trends.

[1] Willet et al. (2014), HadISDH land surface multi-variable humidity and temperature record for climate monitoring

[2] Dunn et al. (2017), Comparison of land surface humidity between observations and CMIP5 models

[3] Sherwood and Fu (2014), A Drier Future?