Variations in Simulated GMI Brightness Temperatures due to Varying Particle Size Distribution Models and Hydrometeor Characteristics in a Severe Hailstorm

Abstract The Atmospheric Radiative Transfer Simulator was used to conduct several simulations of Global Precipitation Measurement Microwave Imager brightness temperatures (BTs; 10.65–183.31 ± 7-GHz) over a severe hailstorm. Simulations were conducted to test the sensitivity of BTs to particle size distribution form and to the size, orientation, and shape of several hydrometeor types assuming constant S-band radar reflectivity. Results show an increase in BT (i.e., less scattering) at most frequencies when changing from a normalized gamma distribution (NGD) to exponential distribution (EXPD). This change causes a decrease in cumulative hydrometeor surface area, but not necessarily a decrease in number concentration, suggesting that surface area exerts a stronger influence on BTs than concentration. Simulated BTs at the highest frequencies (166.0–183.31 ± 7 GHz) agree better with observations when using an EXPD for cloud ice. At lower frequencies, especially 36.5–89.0 GHz, using an NGD for high-density graupel and hail leads to a better match between simulated and observed BTs. No clear preference is seen for low-density graupel, liquid precipitation, or snow. The impact of changing particle shape and/or orientation depends on the hydrometeor type. Changing the orientation of cloud ice from horizontal to a random orientation increases simulated BTs, while having no effect for high-density graupel. Assuming horizontally oriented, oblate-spheroid cloud ice results in simulated BTs that match better with observed. Finally, under the fixed reflectivity constraint, increasing the diameter of hail from 0.5 to 20 cm results in an increase in minimum BT up to 1.5-cm diameter with near constant BT at all frequencies thereafter. Significance Statement Accurate estimates of precipitation are important for numerous applications, and only satellite instruments can provide a global, uniform estimate of precipitation. This study seeks to use simulations to better understand how microwave radiation interacts with various hydrometeor types to improve the assumptions on which satellite-based precipitation estimates are based and ultimately improve the estimates themselves. Results indicate that an exponential distribution may be more appropriate for cloud ice and a normalized gamma distribution more appropriate for hail and high-density graupel. Other hydrometeor types (e.g., rain) show no clear preference for either distribution. Furthermore, assuming cloud ice is an oblate spheroid with horizontal orientation produces simulated brightness temperatures that better match those observed than assuming spherical ice or other orientations.