Diagnosis and Improvement of Cryosphere Shortwave Radiation Biases in Global Climate Models.

Abstract

Faithful representation of cryospheric change is critical for accurate climate modeling, but there are complicating issues in representing snow extent and reflectance in physically realistic ways. This thesis is a collection of diagnostics and improvements of cryospheric shortwave radiation in climate models. Firstly, we incorporate a diagnostic called the cryosphere radiative effect (CrRE), the instantaneous influence of surface snow and sea ice on the top-of-model solar energy budget, into two released versions of the Community Earth System Model. CrRE offers a more climatically relevant metric of the cryospheric state than snow and sea ice extent and is influenced by factors such as the seasonal cycle of insolation, cloud masking, and vegetation cover. We evaluate CrRE during the late 20th century and over the 21st century, specifically diagnosing the CrRE contributions from terrestrial and marine sources. Present-day boreal CrRE compares well with observationally derived estimates. Similar present-day CrRE in the two model versions results from compensating differences in cloud masking and sea ice extent. Radiative forcing in future warming scenarios reduces boreal and austral sea ice cover, and boreal snow cover, which each contribute roughly 1 W/m-2 to enhancing global absorbed shortwave radiation. Similar global cryospheric albedo feedbacks between 0.41-0.45 W/m2/K indicate the models exhibit similar temperature-normalized CrRE change. Secondly, we incorporated a modified canopy scheme into the Community Land Model with snow interception as a prognostic variable and snow unloading tuned to in-situ measurements. The canopy radiation scheme has been updated from a direct temperature dependence of optical parameters to a dependence on the prognostic snow storage. With these improvements, boreal forest zones show large, significant albedo error reductions relative to MODIS observations. 13% gridcell RMSE reduction during spring results from a more gradual seasonal transition in albedo, while 27% reduction in winter is from a lower albedo. Over all North Hemisphere land area, error was also reduced. Thirdly, we assess the impacts of the snow canopy vegetation treatment in coupled model warming scenarios. Little change in global albedo feedback or climate sensitivity were shown, but significant alterations resulted that varied both regionally and temporally.