The utility of fusing multi-sensor data spatio-temporally in estimating grassland aboveground biomass in the three-river headwaters region of China. Issue 18 (16th September 2020)

• Record Type: Journal Article
• Authors: Zeng, Na; He, Honglin; Ren, Xiaoli; Zhang, Li; Zeng, Yuan; Fan, Jiangwen; Li, Yuzhe; Niu, Zhongen; Zhu, Xiaobo; Chang, Qingqing
• Is Part Of: International journal of remote sensing. Volume 41:Issue 18(2020)
• Page Start: 7068
• Page End: 7089
• Publication Date: 2020-09-16
• Subjects: multi-source remote sensing data spatiotemporal fusion -- grassland aboveground biomass (AGB) -- Three-River Headwaters Region (TRHR) -- MODIS -- growing season averaged normalized difference vegetation index (GS-NDVI); Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2020.1752411
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Accurate grassland aboveground biomass (AGB) estimation is crucial for effective grassland utilization. However, most current satellites cannot provide data with high spatial and temporal resolutions simultaneously. Spatiotemporal fusion models can combine the resolution advantages of different remote sensing data and support high-precision vegetation monitoring. In order to obtain accurate grassland AGB maps with high resolution in the Three-River Headwaters Region (TRHR) of China, we developed an estimation method based on the synthetic 30 m growing season averaged normalized difference vegetation index (GS-NDVI), which was fused from 30 m Landsat 8 Operational Land Imager (OLI) and 250 m Moderate-resolution Imaging Spectroradiometer (MODIS) NDVI data. To choose the optimal fusion model, we investigated the performances of three spatiotemporal fusion models for NDVI fusion, the spatial and temporal adaptive reflectance fusion model (STARFM), the enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM), and the rule-based piecewise regression tree model (RPRTM). The three models all produced reasonable NDVI predictions, with the coefficient of determination ( R 2 ) ranging from 0.58 to 0.86. RPRTM had the highest efficiency and was more suitable for large-scale spatiotemporal data fusion. Compared with the models generated from 250 m MODIS GS-NDVI, the AGB estimation models based on 30 m synthetic GS-NDVI were more accurate, demonstrating the effectiveness of our methods. The resulting AGB map of 30 m resolution provides spatially detailed AGB information that will be useful for regional ecosystem studies and local land management decisions.