基于PROSAIL模型及HJ-1A-CCD2影像的MODIS/LAI产品精度验证

Validation of MODIS/LAI production based on PROSAIL model and high spatial resolution HJ-1A-CCD2 image

选取呼伦贝尔温带草甸草原为研究区,设计3km×3km试验样地,采集地面实测数据,用基于HJ-1ACCD2影像的PROSAIL模型反演的LAI值作为中间桥梁进行尺度上推,对MODIS/LAI产品进行验证与评估。结果表明,MODIS/LAI产品与地面实测LAI值相比,两者相对误差为28.1%,其精度受地表异质性、土壤背景影响较大,在草地植被相对均质的情况下,MODIS/LAI产品值高于地面实测值,高估约为0.04,相对误差为5.8%;在草地植被非均质条件下,MODIS/LAI产品值高于地面实测值,高估约为0.18,相对误差为32.7%。与基于尺度上推的1km HJ-1A-CCD2/LAI值相比,两者相对误差为13.5%。

The present study validated accuracy of the MODIS/LAI production with resolution of 1 000 m by compared with in-situ field measurements and retrieved LAI from the PROSAIL physical model using HJ-1A-CCD2 data from China Environment and Disaster Reduction Satellite. The field measurements achieved by designed and carried out measurements of LAI in 3 km×3 km areas from the temperate meadow-steppe in Hulunber region. Heterogeneity in remote sensing data was analyzed and LAI was resampled with scaling-up method and nearest-neighbor algorithm for comparison with MODIS LAI production. The results showed that the MODIS/LAI production was overestimated according to field measurements and underestimated according to HJ-1A-CCD2/LAI data. The relative errors were 28.1% and 13.5%, respectively. The MODIS/LAI production was overestimated about 0.04 in relatively homogeneous grassland and about 0.18 in heterogeneous grassland. The relative errors were 5.8% and 32.7%, respectively. The surface homogeneous, soil background and uncertain model possibly explained these overestimation and underestimation of the MODIS/LAI production. It was necessary to validate the accuracy of MODIS/LAI production as there was bias for relative error