Keywords: RADARSAT, Rice Area Estimation, Backscatter Change of Rice

Abstract:
The authors attempted to estimate rice-planted areas using RADARSAT mode data in an early stage after rice planting. The eastern part of Hiroshima City, Higashi-Hiroshima City, was selected as the test field, and three multi-temporal RADARSAT fine-mode images, taken on Apr, 8, May 26 and June 19, were used as the test data. From the first two temporal data, SAR backscatter intensity in rice-planted fields was found to decrease significantly from Apr. 8 to May 26. From the last two temporal data, SAR backscatter intensity increased significantly from May 26. From the last two temporal data, SR backscatter intensity increased significantly from May 26 to June 19 due to the growth of planted rice. We evaluated the actual performance for rice area estimation by above three RADARSAT data by comparing with the estimated area by SPOT multi-spectral data and also with National Digital Land Information.

1. Introduction
Rice is the most important agricultural product in Japan and a lot of man-power is still necessary to estimate rice planted areas in whole areas of Japan every year. Satellite remote sensing images, such as Landsat TM or SPOT HRV, have been expected to be used to estimate rice-planted areas. However, these optical sensors hardly have been able to get necessary data at a suitable timing due to cloud cover problem during rice planting season in Japan. Therefore currently only space-borne synthetic aperture radar (SAR) images might be practical data sources to realize rice area esimation by remote sensing in Japan.

According to high sensitivity of C-band radar image to surface roughness conditions, the backscatter intensity of RADARSAT images changes greatly from non-cultivated bare soil condition before rice planting to inundated condition just after rice planting (Suga et al, 1999). In addition, RADARSAT images are rather sensitive to the change of rice biomass in a growing period of rice (Ribbes et al, 1999 and Liew et al, 1999). Therefore, rice area estimation is expected to be realized in an early stage, namely only after rice planting or earlier than one month after rice planting. The authors attempted to estimate rice-planted areas using month after rice planting. The authors attempted to estimate rice-planted areas using month after rice planting. The authors attempted to estimate rice-planted areas using RADARSAT fine-mode data in an early stage after rice planting and evaluated the actual RADARSAT fine-mode data in an early stage after rice planting and evaluated the actual performance for rice area estimation by above three RADARSAT data by comparing with the estimated area by SPOT multi-spectral data and also with National Digital Land Information.

2. Test Site and Test Data
The eastern part of Hiroshima City, Higashi-Hiroshima City, was selected as the test field, and three multi-temporal RADARSAT fine-mode (FIF) images, taken on Apr. 8, May 26 and June 19 in 1999 were used as the test data. SPOT/HRV multi spectral data taken on June 21, 1999 were used to generate a reference image for rice-planted area extraction. The landuse data in the National Land Digital Information (NLDI) in the test site was also referred to compare rice field distribution with the result of rice planted areas by RADARSAT data.

Above three RADARSAT and one SPOT images in a part of the test site are shown in Figure 1. The rice fields are mainly distributed in the middle-right and bottom-left portions in the images. The land surface condition of May 26 is almost smooth water surface just after rice planting, and that of June 19 is a mixed condition of growing rice and water surface. It is easily found that the rice fields are shown in a dark tone in the RADARSAT images on APR. 8 and June 19.

RADARSAT F1F Apr. 08, 1999



RADARSAT F1F May. 26, 1999



RADARSAT F1F June 19, 1999



SPOT/HRV(Mu)-B3 June 21, 1999

(RADARSAT Image : @CSA&RADATSAT International 1999. SPOT Image: @ CNES 1999.
Fig.1 Multi-temporal RADARSAT images and SPOT/HRV image in a part of test site.