.png){kind=icon}
.png){kind=icon}
Return
Spatial Associations Between Shallow Cropland Soil Organic Carbon and the Agro-Ecological Environment in the Plateau Lake Basin: A Dual-Perspective Assessment
.png){kind=icon}
PRE
AI
X
B
L
G
J
Z
DOI：10.1109/tgrs.2026.3716895.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Soil organic carbon (SOC) dynamics in croplands surrounding plateau lakes are crucial for maintaining ecological sustainability. However, the spatiotemporal evolution of shallow cropland SOC and its spatial associations with the agro-ecological environment (AEE) remain poorly understood due to uneven sampling, complex terrain, and climatic constraints. Taking the Erhai Lake basin as a representative area, this study integrates multisource satellite and ground data under a soil-pedogenic modeling framework to develop a dual-branch and dual-attention hybrid deep learning model, namely, convolutional neural network-long short-term memory with squeeze-and-excitation and temporal attention (CNN-LSTM-SE-TA, CLST). Five independent CLST models with identical inputs were separately trained to predict SOC, soil pH, hydrolyzable nitrogen (H_N), available phosphorus (AP), and available potassium (AK). Based on optimized time-series cropland extraction data, the spatiotemporal dynamics of SOC were estimated, and an attention mechanism was incorporated to enhance model interpretability. An annual estimation framework for shallow cropland SOC stocks in plateau lake basins was established. Furthermore, a cropland soil health index (CSHI) was derived from the predicted soil nutrient indicators (SNIs) and combined with the remote sensing ecological index (RSEI) to examine the spatiotemporal relationships between SOC and AEE from the perspectives of soil health (SH) and ecological environmental quality (EEQ). Results showed that the CLST model significantly outperformed benchmark methods, exhibiting superior prediction accuracy and spatiotemporal generalization ability. SOC content in the Erhai Lake basin exhibited an increasing-then-decreasing trend from 2007 to 2022 (32.242, 34.473, and 31.378 g <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\cdot $ </tex-math></inline-formula> kg−1, respectively), whereas total SOC stocks continuously declined. Direct changes resulting from in situ SOC depletion were +0.245 and −0.299 Tg C, while land-use transitions caused additional losses of 0.475 and 0.181 Tg C. CSHI remained relatively stable during the early stage but decreased locally in the later stage, showing generally consistent spatial patterns with SOC. In contrast, RSEI deteriorated in some regions, and SOC exhibited synergistic relationships with CSHI but tradeoff relationships with RSEI, highlighting divergent mechanisms underlying agro-ecological dynamics. These findings suggest that SOC dynamics, SH, and EEQ are not always synchronized in plateau lake basins, highlighting the importance of balancing agricultural production, soil carbon sequestration, and ecological conservation in fragile mountain ecosystems.
Keywords:
Cropland soil nutrient
multisource remote sensing data
plateau lake basin
soil health (SH)
soil organic carbon (SOC)
soil-pedogenic model
Journal
IF:
8.6
Papers:
2.1W
Citations:
10.7W
Organization
Cited Papers