Enhanced gap-filling for satellite-derived crop monitoring using temperature-driven reconstruction techniques
A solid understanding of plant growth is crucial for maintaining future crop productivity in the face of climate change. Remote sensing of crop functional traits using optical satellite imagery provides a valuable tool for determining effective management techniques that reduce risk and enhance agro...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Smart Agricultural Technology |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2772375525000504 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | A solid understanding of plant growth is crucial for maintaining future crop productivity in the face of climate change. Remote sensing of crop functional traits using optical satellite imagery provides a valuable tool for determining effective management techniques that reduce risk and enhance agroecosystem resilience. However, atmospheric disturbances limit the availability of imagery, leading to gaps and noise in the trait time series. Thus, accurate crop growth modelling necessitates time series reconstruction methods. One promising approach is incorporating physiological priors, such as the influence of environmental variables like air temperature on plant growth. We propose a novel method that combines Sentinel-2 Green Leaf Area Index (GLAI) observations with three temperature driven reconstruction techniques that describe the physiological relationship between growth and temperature in winter wheat. By employing a probabilistic ensemble Kalman filtering data assimilation scheme, we can integrate high-resolution air temperature data and satellite imagery while quantifying uncertainties. Our results suggest that assimilating Sentinel-2 GLAI and temperature-response-based growth rates allows for the reconstruction of physiologically meaningful GLAI time series. Moreover, our proposed method outperforms state-of-the-art approaches based on logistic functions in terms of physiological plausibility, fitting requirements, and representation of high GLAI values. Our approach requires fewer satellite observations compared to traditional remote sensing time series algorithms, making it suitable for agricultural areas with high cloud cover. Consequently, it has significant potential to improve the reliability of crop productivity assessments based on remote sensing data. |
|---|---|
| ISSN: | 2772-3755 |