Abstract
Utilizing reanalysis and high sensitivity W-band radar observations from CloudSat, this study assesses simulated high-latitude (55–82.5°) precipitation and its future changes under the RCP8.5 global warming scenario. A subset of models was selected based on the smallest discrepancy relative to CloudSat and ERA-I reanalysis using a combined ranking for bias and spatial root mean square error (RMSE). After accounting for uncertainties introduced by internal variability due to CloudSat’s limited four year day-night observation period, RMSE provides greater discrimination between the models than a typical mean state bias criterion. Over 1976–2005 to 2071–2100, colder months experience larger fractional modelled precipitation increases than warmer months, and the observation-constrained models generally report a larger response than the full ensemble. For everywhere except the Southern Hemisphere (SH55, for 55–82.5°S) ocean, the selected models show greater warming than the model ensemble while their hydrological sensitivity (fractional precipitation change with temperature) is indistinguishable from the full ensemble relationship. This indicates that local thermodynamic effects explain much of the net high-latitude precipitation change. For the SH ocean, the models that perform best in the present climate show near-median warming but greater precipitation increase, implying a detectable contribution from processes other than local thermodynamic changes. A Taylor diagram analysis of the full CMIP5 ensemble finds that the Northern Hemisphere (NH55) and SH55 land areas follow a “wet get wetter” paradigm. The SH55 land areas show stable spatial correlations between the simulated present and future climate, indicative of small changes in the spatial pattern, but this is not true of NH55 land. This shows changes in the spatial pattern of precipitation changes through time as well as the differences in precipitation between wet and dry regions.
Highlights
Warming in high latitudes is faster than in lower latitudes partly due to meridional heat transport and positive snow/ice-albedo feedback [1,2]
Behrangi et al [12] performed a comparative analysis between CloudSat total precipitation estimates and other products for the regions 55–82.5◦S/N ( “SH55” and “NH55”, the polar limit is based on the CloudSat orbit on a 2.5◦ × 2.5◦ latitude-longitude grid)
Mean precipitation and spatial statistics showed that both ERA-I and Global Precipitation Climatology Project (GPCP) generally agreed well with CloudSat, CloudSat does not have a formal rainfall product over land and so it was not considered robust over NH55 land
Summary
Warming in high latitudes is faster than in lower latitudes partly due to meridional heat transport and positive snow/ice-albedo feedback [1,2]. Palerme et al [24] compared CloudSat snowfall with that of CMIP5 models over Antarctica and identified that sea-ice extent is a key predictor of simulated Antarctic snowfall They considered a subset of CMIP5 models whose mean Antarctic precipitation was within ±20% of that from the CloudSat 2C-SNOW product. We find that models disagree more strongly in terms of their simulated spatial pattern of precipitation than in terms of their means, relative to the uncertainties introduced over a 4-year period by internal variability This indicates that additional use of spatial information provides a more robust method for ranking models by performance and determining whether better performing models indicate different future changes
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
