To improve the use of remote sensing for the diagnosis of crop growth status over a growing period, we analyzed image data from different types of remote sensing on a specific area over different seasons and developed a method for comparing this data through normalization of the acquired data. Here, normalization refers to the conversion of digital numbers representing image data into ground surface reflectance. The digital numbers representing the image data derived from different types of remote sensing technologies, including satellite imaging, airborne digital imaging, and aerial photographs, cannot be used for a comparative analysis because of inherent differences in the gain and offset values and atmospheric correction values, obtained in different seasons. To enable a comparison of image data from different types of remote sensing technologies generated in different seasons, we normalized the digital values acquired at various wavelength bands to correct for differences in both the gain and offset values and atmospheric correction values based on the ground surface reflectance of asphaltic concrete. This is because there is little difference in the ground reflectance of asphalt, depending on the season; the area of asphalt was sufficiently large for the objects to be identified with the naked eye at a resolution used for satellite data. From 2006 to 2009 in the area of Tayagawa, Chikusei City, Ibaraki Prefecture, we photographed wheat fields (for autumn sowing) over the course of the growing season from stem-growth start to maturity. We obtained data from AVNIR-2 multispectral imagery from the Japanese Advanced Land Observing Satellite (ALOS), images from the ADS 40 airborne digital scanner sensor, and images from aerial photography using film. We normalized this data and examined the validity of our normalization process to compare data from the different types of remote sensing. Following normalization, we calculated the Normalized Difference Vegetation Index (NDVI) based on the ground surface reflectance. Since there was no marked difference in the NDVI for image data acquired from the different types of remote sensors in different seasons, we conclude that the normalization method described here is valid. The NDVI for the stem-growing period was positively correlated with crop yields for each year. In conclusion, the NDVI calculated using the new method described here for normalization of data serves as a meaningful index to diagnose crop growth status and has potential to help growers improve crop management practices.
View full abstract