ESTIMATION OF RICE CROP PRODUCTIVITY BY CROP VARIETY IDENTIFICATION USING REMOTE SENSING DATA

在本研究中，我们从统计学上讨论了利用水稻植物冠层多角度光谱反射率数据的地面数据进行水稻品种分类的可能性。在土壤、营养、供水和当地气候等环境条件均为均匀的实验稻田中种植了14个水稻品种，并测量了每个水稻冠层多角度的光谱反射率值。在400 ~ 850nm处每隔1nm采集一次反射率数据。我们测量了光谱反射率在45度，30度，15度的离底角在主平面的后向散射方向和前向散射方向，在主平面与正交平面的角度相同。计算各水稻品种各测量角度下绿波段（550nm ~ 560nm）、可见红波段（675nm ~ 685nm）和近红外波段（745nm ~ 749nm）的反射率数据。然后，在每一个测量角度的一对水稻品种平均值之间进行非配对t检验。结果表明，利用多角度光谱反射率数据可以统计区分的水稻品种对数量比仅利用最低点光谱反射率数据增加更多。结果表明，在多个角度测量的水稻冠层光谱反射率差异具有统计学意义。

In this study, we statistically discussed the possibility to classify rice varieties using the ground data of multiangular spectral reflectance data out of rice plant canopy. We grew fourteen varieties of rice in the experimental rice paddy field where environmental conditions like soil, nutrition, water supply and local climate were expected to be homogeneous, and then we measured the value of the spectral reflectance at multiangulars of each rice canopy. Reflectance data were collected from 400nm to 850nm in every 1nm. And we measured the spectral reflectance at the off nadir angle of 45 degrees, 30 degrees, 15 degrees in the direction of the backward scattering and the forward scattering in the principal plane, and at the same angle of the principal plane in the orthogonal plane.The reflectance data of the green band (550nm - 560nm), the visible red band (675nm - 685nm) and the near infrared band (745nm - 749nm) at every measuring angle were computed for each rice variety. And then, the unpaired t-tests were done between one pair of means of rice varieties at every measuring angle.As a result of this test, the number of pairs of rice varieties which can be statistically distinguished using multiangular spectral reflectance data increased more than that just using the spectral reflectance data at the nadir angle. It became clearer that the difference of spectral reflectance measured from the rice canopies at multiangulars is statistically significant.