Prediction of Regional Forest Soil Nutrients Based on Gaofen-1 Remote Sensing Data

The study on the spatial distribution of forest soil nutrients is important not only as a reference for understanding factors affecting variability, but also rational use resources and establishment virtuous cycle ecosystems. rapid development remote sensing satellites provides an excellent opportunity to improve accuracy prediction models. This aimed explore utility Gaofen-1 (GF-1) satellite in mapping model Luoding City, Yunfu Guangdong Province, Southeast China. We used 1000 m resolution coarse-resolution map represent overall regional nutrient status, 12.5 terrain-hydrology variables reflect detailed nutrients, 8 surface vegetation status build artificial neural network (ANN) models full variable ANNs, respectively. objects were alkali-hydro-nitrogen (AN), available phosphorus (AP), potassium (AK), organic matter (OM) at five depths (0–20, 20–40, 40–60, 60–80, 80–100 cm). results showed that full-variable ANN was better than indicating reflecting can nutrients. had different effectiveness depths. In upper layers (0–20 20–40 cm), more useful AN, AP, OM, between 10%–14% (R2), less effective AK 8% 6% (R2). deep (40–60, improvement all significant, 3 RMSE ROA ± 5% decreased with depth soil. Remote ANNs (coarse maps + variables) further demonstrated predictive power data decreases depth. Compared terrain-hydrological variables, perform 0–20 cm, rapidly conclusion, integration could strongly (0–40 cm) NDVI, green band, types best predictors. addition, area rich AN while AP are scarce. Therefore, health, attention should be paid monitoring managing AK, OM levels.