Geophysical Retrievals during OLYMPEX/RADEX Using the Advanced Microwave Precipitation Radiometer

The Advanced Microwave Precipitation Radiometer (AMPR) is a downward-pointing radiometer that was flown on the National Aeronautics and Space Administration’s ER-2 aircraft during the OLYMPEX/RADEX field campaign. AMPR performs cross-track scanning during flight and records mixed-polarization brightness temperatures (T_b) at four frequencies: 10.7, 19.35, 37.1, and 85.5 GHz, which can be converted to true horizontally- and vertically-polarized T_b via dual-polarization deconvolution using data from the two orthogonal receivers for each AMPR channel. This has allowed for the development of multiple-linear regression equations to calculate geophysical parameters based on these deconvolved AMPR T_b values. The regression equations were derived using atmospheric profiles from the National Centers for Environmental Prediction’s Global Data Assimilation System (GDAS) and simulated AMPR T_b values from a radiative transfer model, where atmospheric absorption was computed using Rosenkranz models and surface emissivity was computed using the Meissner and Wentz (2012) model.

Regression equations have been developed to retrieve three geophysical parameters using AMPR T_b: integrated cloud liquid water (CLW), atmospheric water vapor (WV), and ocean-surface wind speed (WS). The retrieval equations were applied to AMPR data from five ER-2 flights during OLYMPEX/RADEX to illustrate their capabilities in retrieving these atmospheric properties. To evaluate the performances of the retrieval equations, comparisons were made between the retrieved CLW, WV, and WS values and the values calculated for the same dataset using an independent one-dimensional variational inversion algorithm (1DVAR) developed at Colorado State University. Comparisons were also made between the retrieved values and those obtained using in-situ observations from dropsondes.

Using the simulated AMPR T_b values from the radiative transfer model, the calculated CLW, WV, and WS values had an average root-mean-square deviation (RMSD) of 0.11 mm, 1.28 mm, and 1.11 m s^-1, respectively, compared to the GDAS atmospheric profiles, with minimal crosstalk errors among these three variables and sea surface temperature. When applied to the five OLYMPEX/RADEX cases, the RMSD between the regression equations and 1DVAR were higher than the simulations, but the two methods still agreed well in most cases. Compared to past studies that have used AMPR T_b, the differences in retrieved WV values were lower and artifacts previously observed in the retrieved WV and WS were greatly reduced when using the new retrieval equations. These results indicate the potential for the regression equations to be utilized in radiometer-based retrievals of CLW, WV, and WS at microwave frequencies from an airborne or spaceborne platform, as well as the ability to retrieve these parameters at a relatively high spatial resolution when using aircraft-based AMPR data.