Analysis of Airborne and Tower-Based Cross-Polarization Backscatter Data at X- and Ku- Bands for SWE Retrieval Algorithms.

There have been many airborne and tower-based radar field campaigns at X and Ku bands that have studied the effects of snow on the radar backscattering. These campaigns have been essential in the development of dual frequency X and Ku band algorithms to retrieve snow water equivalent (SWE) from the radar backscattered signal under different snow conditions as well as a wide range of SWE values. A dual frequency X and Ku band algorithm has been validated using nine airborne and tower-based measurement datasets. This dual frequency algorithm is based on measurements of X and Ku band co-polarized signal to retrieve two parameters: the SWE and the scattering albedo ωX.

To further improve the retrieval algorithm, we propose in this paper to incorporate use of cross-polarized backscatter measurements. Currently, the algorithm uses a binary prior on scattering albedo (either ωX = 0.35 or 0.65). The use of cross-polarization is important because it can reduce the requirement for prior information through use of the cross polarization to co polarization ratio. The inclusion of cross polarization, particularly at Ku band, is also important for deep snow layers because the cross-pol Ku-band backscatter increases at a higher rate than co-pol for increasing SWE values.

In this paper, we use the bi-continuous DMRT model to predict radar backscatter from a snow volume. This model has shown to be accurate in predicting both co-polarized and cross-polarized signals for different snow conditions. We also report an analysis of the cross-polarization signatures from airborne (TVCExp 2013) and tower field campaigns (NoSREx). For airborne data, the cross-pol data is around 8-9 dB lower than co-pol, while for tower data, cross-pol is around 10-11 dB lower than co-pol. We also compare these cross-polarization field measurements to the bi-continuous DMRT predicted cross-polarized values to assess the accuracy of the model.

To improve the prior estimate of albedo used in the SWE retrieval algorithm, we also analyze the ratio of cross-polarized signal to the co-polarized signal. We see that, for a SWE range of 50-250 mm, we have an increasing trend of the cross-pol to co-pol ratio with a dynamic range of -9 to -5dB. Based on the ratio, we can identify the regions of low SWE and high SWE to reduce requirements for prior information in SWE retrieval processing.