日本地球惑星科学連合2023年大会

講演情報

[J] オンラインポスター発表

セッション記号 M (領域外・複数領域) » M-IS ジョイント

[M-IS08] 南大洋・南極氷床が駆動する全球気候変動

2023年5月26日(金) 15:30 〜 17:00 オンラインポスターZoom会場 (10) (オンラインポスター)

コンビーナ:草原 和弥(海洋研究開発機構)、箕輪 昌紘(北海道大学・低温科学研究所)、野木 義史(国立極地研究所)、関 宰(北海道大学低温科学研究所)

現地ポスター発表開催日時 (2023/5/26 17:15-18:45)

15:30 〜 17:00

[MIS08-P12] Assessment of the resolution enhancement method for AMSR2 data and its application to thin ice thickness estimation

*中田 和輝1可知 美佐子1栗原 幸雄1 (1.宇宙航空研究開発機構地球観測研究センター)

キーワード:AMSR2、高分解能化、沿岸ポリニヤ、薄氷厚

Satellite passive microwave radiometers, which can observe the earth's surface globally every day regardless of nighttime or cloud cover, have been widely used for estimating sea ice physical parameters such as sea ice concentration. On the other hand, their spatial resolutions (about a few to several tens of kilometers) are often insufficient for sea ice regions with surface complexities. Several studies have developed a method to synthesize higher-resolution brightness temperatures (TBs) using the sensor antenna pattern. Among them, the microwave version of the Scatterometer Image Reconstruction (rSIR) method developed by Early and Long (2001) can effectively improve spatial resolution at low computational cost, especially in polar regions where measurements are acquired from multiple passes per day. However, this method assumes that the TB field does not change between satellite passes. For GCOM-W/AMSR2, the minimum time interval between two passes for a given point is about 1 hour and 40 minutes, during which sea ice typically moves about 1-5 km. Even if measurements from four passes could be acquired within a day, depending on AMSR2 channels, sea ice moves larger than the footprint size during the time difference of >6 hours, which is not negligible for applying the rSIR method. In this study, we have checked the sensitivity of the performance of the rSIR method to sea ice motion and determined the best value of parameters such as the number of satellite passes for synthesizing enhanced-resolution TBs on sea ice regions. Also, we focused on improving the resolution of AMSR2 36GHz and 89 GHz TBs (the footprint size of 12 km × 7 km and 5 km × 3km, respectively), which are the input of the thin ice algorithm. The thin ice algorithm can estimate ice thickness of <20 cm and sea ice production in coastal polynyas (thin ice or low ice concentration area). Since the offshore width of Antarctic coastal polynyas, typically ranging from a few to several tens of kilometers, is close to the resolution of AMSR2, it is expected that the estimates would be greatly improved by combining with the rSIR method.
The details of the sensitivity analysis are as follows: First, an arbitrary true TB distribution with representative features was generated. Next, the true TB distribution was filtered using antenna patterns obtained from a gaussian function to generate TB measurements from multiple passes, while taking into account sea ice motion of 0-3 km/h. Finally, the rSIR method was used to make enhanced-resolution TBs, and the Root Mean Squares Error (RMSE) between the enhanced-resolution and true TBs was calculated. The Sensitivity analysis for 36 GHz shows that the use of measurements from two passes provides the most accurate enhance-resolution TBs. On the other hand, the RMSEs for three or more passes are higher than that for single-pass and two passes when sea ice moved about >0.8 km/h. This sensitivity analysis also shows that the RMSE for single-pass and two-pass are reduced by up to 30% and 38%, respectively, compared with the original measurements. Based on these results, we created AMSR2 36GHz and 89GHz enhanced-resolution TBs mapped onto the NSIDC polar stereographic projection with grid resolutions of 5 km and 2.5 km, respectively, by using measurements from up to two passes in each grid cell. This product provides TB fields more than three times a day, depending on the oceans, since extra TB data that would degrade enhanced-resolution performance is used to improve temporal resolution. In this presentation, we show some examples of applying the thin ice algorithm to the enhanced-resolution TB product and comparing them with the thin ice thickness distribution obtained from the MODIS thermal infrared sensor.

Reference:
Early, D.S.; Long, D.G. Image reconstruction and enhanced resolution imaging from irregular samples. IEEE Trans. Geosci. Remote Sens. 2001, 39, 291–302.