Mapping Soil Organic Carbon by Integrating Time-Series Sentinel-2 Data, Environmental Covariates and Multiple Ensemble Models
Despite extensive use of Sentinel-2 (S-2) data for mapping soil organic carbon (SOC), how to fully mine the potential of time-series S-2 data still remains unclear. To fill this gap, this study introduced an innovative approach for mining time-series data. Using 200 top soil organic carbon samples a...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-03-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/25/7/2184 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Despite extensive use of Sentinel-2 (S-2) data for mapping soil organic carbon (SOC), how to fully mine the potential of time-series S-2 data still remains unclear. To fill this gap, this study introduced an innovative approach for mining time-series data. Using 200 top soil organic carbon samples as an example, we revealed temporal variation patterns in the correlation between SOC and time-series S-2 data and subsequently identified the optimal monitoring time window for SOC. The integration of environmental covariates with multiple ensemble models enabled precise mapping of SOC in the arid region of southern Xinjiang, China (6109 km<sup>2</sup>). Our results indicated the following: (a) the correlation between SOC and time-series S-2 data exhibited both interannual and monthly variations, while July to August is the optimal monitoring time window for SOC; (b) adding soil properties and S-2 texture information could greatly improve the accuracy of SOC prediction models. Soil properties and S-2 texture information contribute 8.85% and 61.78% to the best model, respectively; (c) among different ensemble models, the stacking ensemble model outperformed both the weight averaging and sample averaging ensemble models in terms of prediction performance. Therefore, our study proved that mining spectral and texture information from the optimal monitoring time window, integrated with environmental covariates and ensemble models, has a high potential for accurate SOC mapping. |
|---|---|
| ISSN: | 1424-8220 |