Understanding the responses of precipitation, evaporative demand, and terrestrial water availability to planetary temperature in climate models
MetadataShow full item record
In many models of land hydrology, precipitation (P) and potential evapotranspiration (PET, a.k.a. evaporative demand) are the main inputs that determine actual evapotranspiration, runoff, soil moisture, and aridity or drought. In the first three chapters of this work, we attempt to understand the robust subtropical P declines, planet-wide PET increases, and widespread P/PET declines projected under strong greenhouse warming in CMIP5, a large suite of global climate models (GCMs). Motivated by the apparent absence of this aridification during past greenhouse eras (and the apparent aridity of the ice ages), in the final chapter we use a very simple land model coupled to an atmospheric GCM and a slab ocean to evaluate the relevance and robustness of the P/PET responses to warming across a wide range of boundary conditions and modeling choices. In the CMIP5 projections, robust P declines are almost entirely found within the equator-side flanks and extensions of the model extratropical P belts (including both dry and wet regions), not in the centers of the subtropical dry zones nor on the dry margins of the tropical wet belts. This implies that they are primarily caused by the dynamic poleward retreat of extratropically driven P, not by the thermodynamic increase in dry-zone moisture divergence (which occurs largely as an evaporation increase.) The robust P declines are largely found over the oceans and intersect land only in certain regions; most land locations see non-robust changes in P or robust increases in P. On the other hand, Penman-Monteith PET robustly increases everywhere on land, usually by a low double-digit percentage. This is because the simulated Penman-Monteith PET response is almost always dominated by the response to the local warming itself, not by the responses to concurrent changes in surface radiation, relative humidity (RH), or wind speed. For given values of the latter three variables, warming increases the numerator of the Penman-Monteith equation at a roughly Clausius-Clapeyron rate, ~ 6% K<super>-1</super>, but it increases the denominator more slowly, especially in colder base climates. Thus, evaporative demand increases with local warming at around 1.5-4 % K<super>-1</super>, where the larger values occur in colder regions. A simple analytic scaling for this sensitivity very accurately predicts the PET response field of each model. This PET increase is large enough that in each of the 16 CMIP5 models examined, the ratio P/PET declines with global warming in most land areas in the tropics, the subtropics, and the midlatitudes, implying aridification. However, in our idealized-land GCM, the weakly increasing land P response and strongly increasing PET response that enable this are not general. Depending on the prescribed ocean heat transport, continental configuration, and base planetary temperature, greenhouse warming often causes our modeled land P to strongly decrease, or sometimes to increase so strongly as to entirely suppress the PET increase (even as global-mean P increases weakly in all cases.) The former occurs when the basic-state terrestrial climate is already drier, and the latter occurs when it is quite wet. Future work may investigate what drives this broad range of land P and PET responses to warming, and whether this idealized-model behavior sheds any light on the tension between non-arid past greenhouses and the arid future projections.
- Atmospheric sciences