A 10-m Fractional Vegetation Cover Monthly Dataset of the Kherlen River Basin in 2022

Due to this, The BiLSTM model, rooted in deep learning techniques, endeavors to accurately generate the vegetation cover in this watershed. Firstly, a comprehensive feature dataset is assembled using Sentinel-2 data, integrated with spectral indices and elevation data. This feature dataset effectively captures a range of vegetation characteristics such as chlorophyll content, water status, and terrain attributes.

Subsequently, this feature dataset undergoes division into training and testing subsets to comprehensively assess the performance of four models: BiLSTM, Random Forest Regression, Multilayer Perceptron, and LSTM. Then, the BiLSTM model achieved the best performance with an R2 and RMSE of 0.716 and 0.103, respectively. It is then utilized to generate a monthly vegetation cover dataset for 2022 for the Kherlen River Basin.

This dataset consists of 12 months of vegetation cover inversion results for the Kherlen River Basin, and the data have been completed with operations such as merging and mask extraction. Moreover, it is a valuable resource for monitoring surface vegetation growth and assessing ecosystem vitality in the Kherlen River Basin, thereby furnishing essential data support for ecological studies within the watershed.