Detection of eruption-related microwave signals using a satellite-borne microwave radiometer AMSR2

Monday, 15 December 2014
Takashi Maeda, Japan Aerospace Exploration Agency, Earth Observation Research Center, Tsukuba, Japan
Japan Aerospace Exploration Agency (JAXA) launched GCOM-W1 (Global Change Observation Mission 1st - Water) satellite on 17 May, 2012 (UT). GCOM-W1 satellite carries the Advanced Microwave Scanning Radiometer 2 (AMSR2). AMSR2 is a multifrequency microwave radiometer, and has the single rotating antenna dish shared by multifrequency feedhorns from 6.9 GHz to 89 GHz. Due to this hardware design, a footprint size of each frequency significantly differs. Therefore, even if the multifrequency data obtained as brightness temperatures originate from the same geolocation, they represent responses of quite a different area one another. Accordingly, not considering the difference of footprint sizes in extracting significant information from the multifrequency brightness temperature will be one of implicit causes to deteriorate its quality.

In order to solve this problem, it is necessary to correct multifrequency brightness temperatures observed in native footprints as they are observed in the same footprint. This concept is known as the Backus-Gilbert method. In this method, weighting coefficients for small antenna patterns of high frequency densely distributing in a large antenna pattern of low frequency are first calculated. Then, the large antenna pattern is emulated by weighted average of the small antenna patterns. It is important to find well-modified weighting coefficients in this method. In AMSR2, the brightness temperature dataset corrected by the Backus-Gilbert method is officially defined as level 1R (L1R) product. We have investigated its implementation, especially the criterion to calculate `well-modified' weighting coefficients. The new version of L1R product implemented based on this investigation will be released by the next end of March.

Generally, L1R product produces a more effect in analysis of the land surface than the sea surface because microwave signals emitted from the land surface distribute more nonuniformly due to land cover. Therefore, we investigated L1R products related to volcanic eruptions as a reference analysis to verify microwave signals emitted from the land surface are more accurately evaluated by L1R product. In this presentation, we show the analysis results.