Prediction of Grass Production by Using LANDSAT TM Data in the Kyushu Region

Japanese Journal of Crop Science

Tsunehisa INOUE¹  Yasuyuki WAKIYAMA¹  Nobuaki KOYAMA¹ 

关键词: Effective accumulated temperature;    イネ科牧草;    Grass;    収量予測;    Landsat TM data;    植生指数;    Prediction of production;    有効積算気温;    Vegetation index;    ランドサットTMデータ;   

DOI  :  10.1626/jcs.68.301

学科分类：农业科学（综合）

来源: Nihon Sakumotsu Gakkai / Crop Science Society of Japan

【 摘 要 】

The production of grass in the Kyushu region was predicted by using Landsat TM data. Explanatory variables to be used for the multiple regression model for the prediction were determined by simple regression analysis. The degrees of correlation (r) between 2 vegetation indexes (RVI, NDVI), TM4 band, and the grass production ranged from 0.35 to 0.71. The grass production and mean air temperature were highly correlated during the growing period (r=0.83). The growth of grass from day of satellite observation to day of harvest was estimated on the basis of effective accumulated temperature (over 5°C) of this period. The multiple regression models were obtained in each year of 1989, 1990, 1992, and 1996 by using the TM4 band, the TM2 band, and effective accumulated temperature as explanatory variables. The coefficients of determination (r2) ranged from 0.64 to 0.72 in each model. The production of grass in the other years was then predicted based on these models, but the accuracy of the prediction was poor. This inaccuracy might be attributed to the above method of estimating grass growth from the day of satellite observation to the day of harvest and to an atomospheric effect on the TM data. These results suggested that the accuracy of prediction could be improved by using a crop growth model and atmospheric correction.

【 授权许可】

Unknown