Uncertainties of global moderate resolution Leaf Area Index (LAI) products derived from remote sensing data

Global moderate resolution leaf area index (LAI) products are derived from different satellite remote sensing data, for example the MODIS, CYCLOPES and GLOBCARBON projects. For an optimal use of these products, knowledge about the data consistency based on product intercomparison is necessary. The objective of this study is to quantitatively compare the global leaf area index (LAI) products derived from the TERRA and AQUA MODIS (2000-2007) and SPOT/VEGETATION (1999-2007), and GLOBCARBON projects. A global LAI field measurement database was created from literature review and major validation campaigns. Remote sensing LAI products agree very well with field data for grasses and cereal crops, shrubs and savannas. The Collection 5 MODIS LAI has improved over the earlier Collection 4 products, especially for needleleaf forest. While more field data are necessary, particularly for broadleaf crops and broadleaf forests, our study finds no significant discrepancies between Collection 5 MODIS and CYCLOPES LAI products for grasses and cereal crops, shrubs, savannas, and needleleaf forests. The CYCLOPES and the TERRA+AQUA main algorithm MODIS LAI products display the best overall fit with ground measurements (RMSE=1.23 and 1.25, respectively) in spite of sparse field points and scale differences. A triple collocation error model was introduced to compare the monthly 0.05° LAI products produced by MODIS, CYCLOPES and GLOBCARBON. The resultant uncertainty information for each product gives a good indication of using these products in global land surface, hydrological and climatic models. The uncertainty information can be used directly as quantitative evaluation of the LAI products as well as input to feed global models. Intercomparison of mean (+ 1 standard deviation) MODIS and CYCLOPES LAI with common field observations for each biome type. The number of LAI observations for each biome is also shown.