Alpine aboveground biomass dataset in Qinghai Lake Basin (2000-2015)

As a kind of important renewable resources, grassland resources have significant influence on human’s daily life. China is a country with abundant grassland resources. The scientific use of grassland resources would contribute to the sustainable development of animal husbandry, national unity and the stability of the country. However, grassland resources are facing with more and more problems, with the development of agriculture, industry, animal husbandry, population growth, and the impact of global warming. Therefore, obtaining accurate real-time information of the growth condition of grassland is quite important. People can use this information carrying on the scientific management of grassland resources, thus protecting grassland resources and keeping the sustainable development of animal husbandry. Traditional observation method is mainly ground experiment, which would cost lots of time and money. Remote sensing data has the advantage of near-real time, dynamic observation and contains image with large scale. But a single type of remote sensing data cannot meet the needs of high temporal-spatial grassland biomass observation. This study intends to use data fusion method to generate high temporal-spatial remote sensing data. Then combining with ground survey data , we established the parametric and non-parametric model. Eventually we developed the optimal aboveground biomass model for Qinghai Lake Basin and generate the biomass time series with 30 meter resolution and 8 day interval during 2000—2015. We then analyzed the grassland trend in Qinghai Lake basin during the past 16 years. The main work and the conclusions of our findings are as follows: (1) According to the actual situation of Qinghai Lake Basin, we developed the optimal fusion model from three prospects: the selection of generating synthetic NDVI, the comparison between different image (different MODIS product and TM image in different years), and the development of data fusion algorithm. We finally generated the synthetic NDVI series of Qinghai Lake Basin. The selection of the fusion scheme would directly affect the precision of the vegetation index, and then impact the accuracies of the construction of the biomass model. Based on Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) algorithm, we used MCD43A4 as the input MODIS file. We then chose data in the same year, in adjacent year, and data with 2-yr intervals. Based on the landcover type, we used decision tree to choose different windows for different vegetation types: 350m for croplands; 950m for forest; 750m for grassland and other vegetation types. We have synthetic NDVI time series with relatively high spatial and temporal resolution. It can tell more spatial details on the vegetation variation compared with MODIS data. (2) Based on the measured data and the fusion vegetation index data, the parametric models and a non-parametric model were established and compared. Finally, the experimental results show that the support vector machine (SVM) model has good accuracy. Based on this model, the data set of 30-m data series of grassland aboveground biomass in Qinghai Lake area in the past 16 years was established. We built the model in the following four steps: We generated the synthetic NDVI series with the optimal data fusion scheme; combined with the 291 field samples and vegetation index data, we generated the biomass estimation model of Qinghai Lake region; We chose the optimal model for biomass estimation according to the test data. We finally generated the biomass series with 320 scenes. Biomass estimation model with synthetic NDVI (r=0.85, RMSE=74.45g/m2) can not only maintain accuracies of the models based on MODIS NDVI (r=0.85, RMSE=73.20g/m2); it can also increase the spatial resolution of the biomass from 500m to 30m, and increase the time resolution up to 8 days. (3) The degradation condition of grassland in in Qinghai Lake area was analyzed. We found that during the past 16 years, grassland resources in this area have changed greatly. Grassland in the south lakeshore and the mountainous area in the northern part of the basin showed large degradation, while in the middle of the Qinghai Lake Basin, grassland showed growing tendencies. Grassland with apparent degradation accounted for 8.5% of the basin,while grassland with apparent growth account for 24.5% of the basin. The degradation of grassland were partly contributed by global warming; while the unscientific use of grassland resources is another critical issue caused the land degradation. In addition, as a tourist hot spot, In recent years, tourists number in Qinghai Lake Basin increased dramatically, which would also contribute to the grassland degradation in the local area.