[A-AS10] General Meteorology

[AAS10-03] High Spatial Resolution Observation of Oceanic Heavy Rain by ALOS-2/PALSAR-2 Full Polarimetry

★Invited Papers

*Ando Megumi¹, Masato Furuya¹ (1.Hokkaido University)

Keywords:Full-Polarimetry, L-band, Observation of Oceanic Rain, High Spatial Resolution

Traditional rain radar measures the scattering process of microwaves by raindrops. In domestic land and surrounding seas, the radar of the Japan Meteorological Agency plays a central role. On the other hand, for global sea area observations, global rain observations using multiple satellite data have progressed since the Tropical Rainfall Measuring Mission(TRMM)and are currently carried out by the Global Precipitation Measurement(GPM)project. However, precipitation observations over the sea are still insufficient (e.g., Sun et al. 2017). This study presents several cases where L-band Synthetic Aperture Radar (SAR), which is completely different from existing rain radar, can map heavy rain over the sea with high spatial resolution that traditional rain radar cannot achieve. In a previous study by Melsheimer et al.(1998), experimental observation data of rainfall areas over the sea using multiple frequencies(L, C, X - band)and multiple polarizations with the Shuttle Imaging Radar were used to compare the scattering intensity images and the phase difference of HH-VV for each polarization in L, C, X - bands for precipitation cells over the sea. Since the volume scattering and attenuation by precipitation particles can be almost ignored in L-band SAR, the roughness of the surface of the seawater is basically observed regardless of the presence or absence of rainfall. Theoretical studies (Manton, 1973; Le Mehaute and Khangaonkar, 1990)supporting that the higher the precipitation intensity, the more flattened the surface roughness of the seawater becomes, and Melsheimer et al.(1998)showed that the scattering intensity decreases in the L-band. However, since then, there has been no study confirming heavy rain over the sea with multiple polarizations in L-band SAR, and polarization information has not been utilized. The L-band SAR data we use is full-polarimetry(4 polarizations: HH, VV, HV, VH; abbreviated as FP)data from PALSAR-2, mounted on ALOS-2(Daichi-2)launched by JAXA in 2014, with high spatial resolution of less than 6 m. The PALSAR-2 data used for analysis was searched through the JAXA Earth Observation Satellite Data Distribution System "G-Portal". For comparison, we also use the nationwide composite radar echo intensity GPV (250 m resolution) created by the Japan Meteorological Agency. While Melsheimer et al.(1998)showed intensity images and phase difference profiles, this study also uses Coherence images that reflect the degree of phase coherence in addition to intensity images. As an example, we describe the results of heavy rain over the sea off the coast of Aichi Prefecture at 14:44 UTC on August 18, 2017. By combining intensity images of HH polarization, VV polarization, and Coherence images of HH-VV, radar signatures reflecting differences in precipitation intensity were detected. From the coherence images that are averaged over 3 × 3 pixels, which means a resolution higher than 20 meter, we could clearly identify strong precipitation areas that were not clear only in the intensity images. Since the sea is basically weak in intensity, there are cases where it is difficult to read the weakening of intensity due to precipitation from intensity images alone, but by using Coherence images together, it is possible to clearly detect radar signatures due to precipitation. It should be noted that areas with weak scattering intensity and low Coherence in L-band SAR data do not necessarily indicate precipitation areas (False Positive), and during strong winds such as typhoons, the effect of surface roughness on the seawater due to wind is significant. Therefore, L-band FP data is not omnipotent for detecting precipitation over the sea. However, if the wind speed is relatively low, it is possible to capture precipitation areas with spatial resolution of less than 10 m. Estimating quantitative precipitation intensity distribution and utilizing deep learning are future challenges.