An effective BiLSTM-CNN model for predicting large-scale temporal-spatial dynamics of normalized difference...

Area of Science:

• Ecology
• Remote Sensing
• Data Science

Background:

• Terrestrial ecosystems and vegetation health are vulnerable to climate change and human activities.
• Remote sensing vegetation indices, like the Normalized Difference Vegetation Index (NDVI), are crucial for monitoring vegetation stress.
• Predicting NDVI variations is essential for attributing climate change impacts on vegetation growth.

Purpose of the Study:

• To develop an advanced deep learning model for temporal-spatial NDVI forecasting.
• To integrate meteorological and soil moisture data for enhanced vegetation activity prediction.
• To assess the model's capability in detecting subtle and diverse vegetation alteration trends.

Main Methods:

• Implementation of a combined Bidirectional Long Short-Term Memory (BiLSTM) and Convolutional Neural Network (CNN) model (BiLSTM-CNN).
• Utilizing meteorological and soil moisture data to inform the temporal-spatial NDVI modeling.
• Comparative evaluation of the proposed BiLSTM-CNN method against existing state-of-the-art techniques.

Main Results:

• The BiLSTM-CNN model demonstrates strong performance in forecasting NDVI.
• Experimental results confirm the model's competitiveness with current NDVI prediction methods.
• The model effectively captures complex temporal and spatial vegetation dynamics.

Conclusions:

• The proposed BiLSTM-CNN method offers a robust approach for NDVI prediction and vegetation monitoring.
• This advanced deep learning technique aids in understanding climate change effects on vegetation.
• The model shows potential for early detection of vegetation stress and alterations.