Monitoring vegetation dynamics (2010-2020) in Shengnongjia Forestry District with cloud-removed MODIS NDVI series by a spatio-temporal reconstruction method

Shengnongjia Forestry District is the national natural reserve with abundant biological resources in China. Monitoring its variations of vegetation for the intimate connection with eco-environmental changes is of great significance. In this paper, the 16-day composite MODIS normalized difference vege-tation index (NDVI) products (MOD13A1) with 500 m resolution from 2011 to 2020 were selected to investigate the vegetation dynamics in Shengnongjia Forestry District. To alleviate the cloud contamina-tion in NDVI products, a spatio-temporal prefill method with harmonic analysis of time series (ST-HANTS) is proposed. ST-HANTS first prefills the raw NDVI time series using spatio-temporal information to reduce data gaps, which effectively improves the reconstruction performance of the subsequent HANTS algorithm. In the simulation experiments, ST-HANTS shows the highest average correlation coef-ficient and the lowest average root mean square error compared with other mainstream methods, includ-ing HANTS, Savitzky-Golay filter, wavelet transform, and data assimilation. The reconstruction curves are close to the upper envelope of the NDVI time series, which is more consistent with the vegetation phe-nology and can effectively capture the key points in the growth cycle. By analyzing the cloud-free NDVI time series reconstructed with ST-HANTS, results reveal the overall trend of Shennongjia vegetation cov-erage is high in the middle while low at the edge. Except for the population centers and Hongping Airport, the NDVI of most areas is above 0.7 and shows a remarkable increasing tendency. Moreover, the fluctu-ation degree of NDVI in the whole study area is low, indicating that the ecological environment of Shennongjia is relatively stable. The vegetation variations are influenced by land surface temperature and precipitation, and the vegetation growth response to precipitation exhibits an apparent hysteresis. & COPY; 2023 National Authority of Remote Sensing & Space Science. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).