Validation of the MODIS LAI product in Qinghai Lake Basin combined with field measurements using Landsat 8 OLI data

The leaf area index (LAI) has been widely used in various scientific studies to characterize the dynamics and functions of vegetation using remote sensing. Validation of NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) LAI products is of high significance, due to its role as an important input parameter for climate and eco-hydrological models at global and regional scales. However, recent validation activities have mainly been located in North America, Europe, and Africa. Furthermore, the validations principally focused on forests and savanna. As one of the representative areas in which unique and sensitive alpine ecosystems are widely distributed, the Qinghai Lake Basin maintains the ecological security of the eastern Tibetan Plateau. Therefore, exploring the precision and applicability of MODIS LAI applied to the alpine ecosystem of Qinghai Lake Basin is a crucial prerequisite to using the LAI parameter for weather forecast and water resource management models. In this paper, the inversion accuracy of the Collection 5 MODIS LAI (MOD15A2) products was examined for three typical vegetation types, i.e., alpine meadow, steppe, and subalpine shrub, in the Qinghai Lake Basin. On the basis of the validation of the plots' representation of MODIS pixels, we measured LAI of 84 plots covering the entire basin during August 10 and 20, 2015, in order to establish the Normalized Difference Vegetation Index (NDVI)-LAI exponent regression model using Landsat 8 OLI remote sensing data. The Landsat 8 LAI map with 1-km spatial resolution, which is the same as that of the MODIS LAI product, was then obtained by aggregating Landsat 8 LAI data with 30-m resolution derived from the NDVI-LAI model. The aggregation progress was performed using a scaling-up method, referred to as average aggregation. The validation and applicability analysis of the MOD15A2 LAI product then were carried out through a pixel by pixel method. The results can be summarized as follows. (1) The MODIS LAI product in the study area agrees well with the Landsat 8 inverted results for alpine meadow and steppe. (2) Validation of the product varies with different vegetation types. Specifically, the MODIS LAI product is underestimated by approximately 10% for alpine meadow (R2 = 0.77), overestimated by approximately 9% for steppe (R2 = 0.59), underestimated by approximately 51% for subalpine shrub (R2 = 0.42), and underestimated by approximately 16% when the vegetation of the entire basin is not distinguished (R2 = 0.36). (3) The precision of the MODIS LAI product may be influenced by the uncertainty of the sampling strategy, surface heterogeneity, surface reflectance, and the NDVI-LAI statistical model. Of these, spatial heterogeneity has the major impact on the uncertainty of the MODIS LAI product. Consequently, MODIS LAI products should be combined with vegetation maps when applied for a particular vegetation type in the Qinghai Lake Basin. Simultaneously, correction is necessary for MODIS LAI products to improve the accuracy. The results of this study can provide an important basis for researching the selection and correction of MODIS LAI products used in the Qinghai Lake Basin.