Dynamic graph learning framework based seasonal and trend decomposition approach for potato crop evapotranspiration...

Area of Science:

• Agricultural Science
• Data Science
• Environmental Science

Background:

• Accurate estimation of crop water requirements (ETc) is crucial for sustainable agriculture, especially with increasing climate variability.
• Traditional methods often fail to capture the complex dynamic patterns of crop evapotranspiration.
• There is a need for advanced analytical approaches to improve ETc estimation accuracy.

Purpose of the Study:

• To propose a novel dynamic graph-based Dual-Graph Semantic Fusion (DG-DGSF) model for enhanced crop water requirement estimation.
• To address the limitations of traditional methods in analyzing dynamic evapotranspiration patterns.
• To improve the accuracy and efficiency of agricultural water management.

Main Methods:

• Decomposition of multivariate time series into trend and seasonal components to create Seasonal Dynamic Graphs (SDG) and Trend Dynamic Graphs (TDG).
• Application of Dual-Graph Semantic Fusion (DGSF) to extract semantic characteristics from the dynamic graphs.
• Integration of Dynamic Graph Learner (DGL) and Graph Convolutional based on Recurrent Unit (GC-GRU) for time-series analysis.
• Utilizing a contrastive learning model for semantic feature extraction and a multi-layer perceptron for supervised prediction.

Main Results:

• The DG-DGSF model demonstrated superior performance compared to state-of-the-art models (BiLSTM, GRU, GCN, etc.).
• Achieved the lowest forecasting error with significant improvements in ETc prediction rates.
• Key performance metrics include RMSE = 0.0469, MAPE = 0.120, NRMSE = 0.0431, KGE = 0.977, and NSE = 0.963.

Conclusions:

• The proposed DG-DGSF model offers a significant advancement in estimating crop water requirements.
• Its ability to capture dynamic patterns leads to more accurate and reliable ETc predictions.
• The model provides a valuable tool for optimizing water resource management in agriculture under changing climatic conditions.