Near-surface air temperature (Ta) is an important parameter in agricultural production and climate change. Satellite remote sensing data provide an effective way to estimate regional-scale air temperature. Therefore, taking Gansu section of the upper Weihe River Basin as the study area, using the filtered reconstructed high-quality long-time series normalized difference vegetation index (NDVI), interpolated reconstructed land surface temperature (LST), surface albedo, and digital elevation model (DEM) as the input data, the back-propagation artificial neural network algorithm (BP-ANN) was combined with a multiple linear regression method to estimate regional air temperature, and the influencing factors of air temperature estimation were analyzed. This method effectively compensates for the fact that air temperature data provided by a single station cannot represent regional air temperature information. The result shows that the temperature estimation accuracy is high. In terms of interannual variation, the air temperature in the study area showed a slightly increasing trend, with an average annual increase of 0.047°C. The calculation results of the interannual variation rate of temperature showed that the area with increased air temperature accounted for 75.8% of the total area. In terms of seasonal variation, compared with that in summer and winter, the air temperature rising trend in autumn was obvious, and the air temperature in the middle of the study area decreased in spring, which is prone to frost disasters. LST and NDVI in the study area were positively correlated with air temperature, and their positive correlation distribution areas accounted for 93.62% and 94.34% of the total study area, respectively. NDVI, LST and DEM influence the temperature change in the study area. The results show that there is a significant positive correlation between NDVI and air temperature, and the change of NDVI has a positive effect on the spatiotemporal variation of air temperature. The correlation coefficient between LST and air temperature in the southeast of the study area is negative, and there is a difference. In addition, the correlation coefficient between LST and air temperature in other areas of the study area is positive. The air temperature decreased with elevation, air temperature decreases by 0.27°C every hundred meters.

No abstract

Abstract. The Carnegie-Ames-Stanford Approach (CASA) model is widely used to estimate vegetation net primary productivity (NPP) at regional scale. However, the CASA is still driven by multi-source data, e.g. satellite remote sensing (RS) data, and ground observations that are time-consuming to obtain. RS data, can conveniently provide real-time surface information at the regional scale, thus replacing ground observation data to drive CASA model. We attempted to improve the CASA model in this study using DEM data derived from radar RS and RS products data generated from Moderate Resolution Imaging Spectroradiometer satellite sensor. We applied it to simulate the NPP of alpine grasslands in Qinghai Lake Basin, which is located in the northeastern Qinghai-Tibetan Plateau, China. The accuracy of the RS data driven CASA, with mean absolute percent error (MAPE) of 23.32 % and root mean square error (RMSE) of 26.26 g C•m-2•month-1, was higher than that of the multi-source data driven CASA, with MAPE of 49.08 % and RMSE of 65.21 g C•m-2•month-1. The NPP simulated by RS data driven CASA in July 2020 shows an average value of 110.17 ± 26.25 g C•m-2•month-1, which is similar to published results and comparable with the measured NPP. The results of this work indicate that simulating alpine grassland NPP with satellite RS data rather than ground observations is feasible. We may provide a workable reference for rapidly simulating grassland, farmland, forest, and other vegetation NPP to satisfy the requirements of precision agriculture, precision livestock farming, accounting carbon stocks, and other applications.