Publications

Yu, X; Xie, JC; Jiang, RG; Zhao, Y; Li, FW; Liang, JC; Wang, YP (2021). Spatiotemporal variation and predictability of vegetation coverage in the Beijing-Tianjin-Hebei metropolitan region, China. THEORETICAL AND APPLIED CLIMATOLOGY, 145(2-Jan), 47-62.

Abstract
Vegetation coverage and its dynamic response to climatic ecologic indicators are critical for the monitoring and management of terrestrial ecosystem. This paper aims to investigate the spatial and temporal variations and relationships between vegetation coverage and climatic factors during the growing season for the period of 2000-2017 in the Beijing-Tianjin-Hebei metropolitan region (BTH) of China, using the Normalized Difference Vegetation Index (NDVI) and related climate data. The multivariate linear regression model (MLR), support vector regression model (SVR), and gradient boosting regression tree model (GBRT) were applied to explore the predictability of NDVI, using climatic factors as predictors. The results showed that the mean NDVI during growing season has significantly increased in the BTH over the past 18 years, with about 66% of the total vegetation cover in the study area evidencing increasing trends. Significant increasing trends were mainly located in the northwest mountainous regions and southeastern plains, and the trends significantly decreased in big cities and the surroundings. The precipitation during growing season and summer has increased. However, no significant trend of the temperature was detected during growing season. NDVI was mainly positively correlated with precipitation at about 85.8% area of the BTH and negatively correlated with temperature during summer. Precipitation and temperature were the dominant influencing factors for vegetation growth in BTH, and therefore these factors were used as potential predictors to estimate NDVI during growing season. The comparative results of three prediction models showed that the prediction accuracy of GBRT was higher than other two models, indicating that it should be applicable to use GBRT model to predict the NDVI in the study area.

DOI:
10.1007/s00704-021-03616-x

ISSN:
0177-798X