Validation of Global LAnd Surface Satellite (GLASS) leaf area index product

Abstract

Long time series Global LAnd Surface Satellite (GLASS) Leaf Area Index (LAI) product (1981-2012) is generated from time-series MODIS and AVHRR reflectance data using general regression neural network method. This study assesses the performance of the GLASS LAI product in two ways: (1) by comparing the spatial and temporal characteristics of the GLASS LAI product with those of other moderate-resolution LAI global products and (2) by comparing the GLASS LAI values with ground measurement data. The results show that the GLASS LAI product achieved a good consistency with the MODIS and CYCLOPES LAI products at the global scale, although with differences in magnitude of LAI values. The largest differences occur between the GLASS and CCRS LAI products in high northern latitudes and close to the equator, followed by differences between the GLASS and MODIS LAI products and the GLASS and CYCLOPES LAI products. The GLASS and MODIS LAI products have more complete temporal trajectories than the CYCLOPES LAI product, while the GLASS and CYCLOPES LAI products have more continuous and realistic trajectories than the MODIS LAI product. The GLASS LAI product maintains reasonable profiles in contrast to the MODIS LAI product, which shows dramatic fluctuations, particularly during the growing seasons. Compared with 20 ground-measured LAI reference maps at 17 sites, the GLASS LAI product shows lower uncertainty, with an R-square of 0.76 and RMSE of 0.51, than the MODIS and CYCLOPES LAI products.

在国家高技术研究发展计划（863计划）重点项目的支持下,已利用MODIS和AVHRR地表反射率数据生成了1981年—2012年的GLASS（Global LAnd Surface Satellite）叶面积指数（LAI）产品。本文从两个方面对GLASS LAI产品的质量进行分析和评价:（1）与现有的全球LAI产品进行比较,分析GLASS LAI产品的时空变化特征;（2）利用LAI的地面测量数据,对GLASS LAI的精度进行评价。研究结果表明:GLASS LAI与CCRS LAI在高纬度和赤道附近区域的差异较大;相对而言,GLASS LAI与MODIS（主算法反演）和CYCLOPES LAI在空间分布上具有更好的一致性;GLASS和CYCLOPES LAI的时间序列曲线连续平滑,MODIS LAI在一些区域的植被生长季节存在剧烈的跳跃;与LAI的地面测量数据进行比较,GLASS LAI产品的R2为0.76,RMSE为0.51,结果明显优于MODIS和C YCLOPES LAI产品。