高分一号卫星影像特征及其在草地监测中的应用

Characteristics and Application of GF-1 Image in Grassland Monitoring

为评价高分一号卫星数据的草地监测能力,在分析传感器波段设置、辐射分辨率和光谱响应系数等特征的基础上,以草地为研究对象,提取草地分布信息,计算不同植被指数,结合地面同步观测的草地光谱、地上生物量、覆盖度和叶面积指数等实测数据,通过R^2和均方根误差筛选并建立最优估算模型。结果表明:波段设置与部分常用传感器保持了较好的一致性;空间分辨率的提高,增强了地物类型的识别能力,辐射分辨率的提高,增强了数据的层次性;光谱响应系数较好的涵盖了不同草地类型的光谱曲线特征;叶面积指数和生物量的最佳估算模型均为基于比值植被指数的三次多项式模型,覆盖度最佳估算模型为基于归一化植被指数的幂函数模型,并得到了较好的制图效果。

In order to evaluate the monitoring ability of GF-1 image in grassland,on the basis of analysis of the characteristics of the band setting,radiometric and spectral response coefficient of sensor,the distributed information was extracted,and the vegetation index of grassland was calculated.With the combination of field-measured spectrum,vegetation coverage,leaf area index and aboveground biomass data,the best vegetation index for grassland parameters was estimated.The optimal model was determined according to R2 and RMSE（root-mean-square error）.The results showed that GF-1 sensors kept consistency in band set comparing with other sensors.The improvement of spatial resolution enhanced the identification ability of object types,and the improvement of radiation resolution enhanced the levels of data.The spectral response coefficients covered better the spectral curves of different types of grassland.The correlation of different grassland vegetation parameters and GF-1 vegetation index reached a high level,and met the needs of remote sensing estimation or inversion.The regression analyses showed that the best estimation model for LAI and the biomass of the grassland were cubic polynomial regression model based on RVI（ratio vegetation index）,and the best estimation model for the vegetation coverage of the grassland were power function model based on NDVI（normalized difference vegetation index）,and the good mapping effect of the research region was obtained.