Assessment of Six Machine Learning Methods for Predicting Gross Primary Productivity in Grassland

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(14), P. 3475 - 3475

Published: July 10, 2023

Grassland gross primary productivity (GPP) is an important part of global terrestrial carbon flux, and its accurate simulation future prediction play role in understanding the ecosystem cycle. Machine learning has potential large-scale GPP prediction, but application accuracy impact factors still need further research. This paper takes Mongolian Plateau as research area. Six machine methods (multilayer perception, random forest, Adaboost, gradient boosting decision tree, XGBoost, LightGBM) were trained using remote sensing data (MODIS GPP) 14 factor carried out grassland GPP. Then, flux observation (positions stations) non-flux reference data, detailed evaluation comprehensive trade-offs are on results, key affecting performance explored. The results show that: (1) six highly consistent with change tendency demonstrating applicability prediction. (2) LightGBM best overall performance, small absolute error (mean less than 1.3), low degree deviation (root mean square 3.2), strong model reliability (relative percentage difference more 5.9), a high fit (regression determination coefficient 0.97), closest to bias only −0.034). (3) Enhanced vegetation index, normalized precipitation, land use/land cover, maximum air temperature, evapotranspiration, evapotranspiration significantly higher other determining factors, total contribution ratio exceeds 95%. They main influencing study can provide for also support

Language: Английский

High-Performance Computing and Artificial Intelligence for Geosciences

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(13), P. 7952 - 7952

Published: July 7, 2023

Geoscience, as an interdisciplinary field, is dedicated to revealing the operational mechanisms and evolutionary patterns of Earth system [...]

Language: Английский