Simulating the effects of mid- to upper-tropospheric clouds on microwave emissions in EC-Earth using COSP

Abstract. In this work, the Cloud Feedback Model Intercomparison (CFMIP) Observation Simulation Package (COSP) is expanded to include scattering and emission effects of clouds and precipitation at passive microwave frequencies. This represents an advancement over the official version of COSP (version 1.4.0) in which only clear-sky brightness temperatures are simulated. To highlight the potential utility of this new microwave simulator, COSP results generated using the climate model EC-Earth's version 3 atmosphere as input are compared with Microwave Humidity Sounder (MHS) channel (190.311 GHz) observations. Specifically, simulated seasonal brightness temperatures (T_B) are contrasted with MHS observations for the period December 2005 to November 2006 to identify possible biases in EC-Earth's cloud and atmosphere fields.

The EC-Earth's atmosphere closely reproduces the microwave signature of many of the major large-scale and regional scale features of the atmosphere and surface. Moreover, greater than 60 % of the simulated T_B are within 3 K of the NOAA-18 observations. However, COSP is unable to simulate sufficiently low T_B in areas of frequent deep convection. Within the Tropics, the model's atmosphere can yield an underestimation of T_B by nearly 30 K for cloudy areas in the ITCZ. Possible reasons for this discrepancy include both incorrect amount of cloud ice water in the model simulations and incorrect ice particle scattering assumptions used in the COSP microwave forward model. These multiple sources of error highlight the non-unique nature of the simulated satellite measurements, a problem exacerbated by the fact that EC-Earth lacks detailed micro-physical parameters necessary for accurate forward model calculations. Such issues limit the robustness of our evaluation and suggest a general note of caution when making COSP-satellite observation evaluations.