Machine Learning and Feature Selection for soil spectroscopy. An evaluation of Random Forest wrappers to predict soil organic matter, clay, and carbonates

Canero, Francisco M.; Rodriguez-Galiano, Victor; Aragones, David

doi:10.1016/j.heliyon.2024.e30228

Search citation statements

Order By: Relevance

Paper Sections

Select...

Citation Types

Supporting

Mentioning

Contrasting

Year Published

2024

Publication Types

Select...

Article2

Preprint1

Relationship

Self Cite0

Independent3

Authors

Journals

Cited by 3 publications

References 93 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Mapping the carbon mitigation potential of photovoltaic development in the Gobi and desert regions of China

Lyu,

Li,

Zhang

et al. 2024

Energy

View full text Add to dashboard Cite

Mapping the carbon mitigation potential of photovoltaic development in the Gobi and desert regions of China

Lyu,

Li,

Zhang

et al. 2024

Energy

View full text Add to dashboard Cite

Application of chemometrics based on digital image analysis for simultaneous determination of tartrazine and sunset yellow in food samples

Hosseini,

Heidari,

Zendegi-Shiraz

et al. 2024

Food Chemistry

View full text Add to dashboard Cite

Estimating forest aboveground carbon sink based on Landsat time-series and its response to climate change

Yang,

Luo,

Zhang

et al. 2024

Preprint

View full text Add to dashboard Cite

Accurately estimating forest carbon sink and exploring their climate-driven mechanisms are essential for achieving carbon neutrality and sustainable development. Taking Pinus densata in Shangri-La as the research object, we established three Random Forest (RF) dynamic models based on Landsat time series and ground data with 5-year interval variation, 10-year interval variation, and annual average variation. Then, Genetic Algorithm (GA) was applied to optimize the parameters of RF to establish GA-RF dynamic models, and selected the optimal model to estimate the carbon sink intensity (CSI) of Pinus densata. Finally, climate-driven mechanisms were explored by correlation analysis. We found that 1) the GA-RF model based on the annual average variation had the highest accuracy with an R2 of 0.83. 2) The CSI of Pinus densata in Shangri-La was 7.84–12.35×104 t C·hm− 2 from 1987 to 2017. 3) Precipitation had the greatest effect on CSI. The joint weak drive of CSI by precipitation, temperature and surface solar radiation was the most dominant form of CSI drive for Pinus densata. These results suggest that the GA-RF model can be used for large-scale long-term estimation of above-ground carbon sinks in highland forests. In addition, the precipitation-led multifactorial synergistic driving mechanism will stabilize the carbon sink capacity of Pinus densata in the long term.

show abstract

Machine Learning and Feature Selection for soil spectroscopy. An evaluation of Random Forest wrappers to predict soil organic matter, clay, and carbonates

Cited by 3 publications

References 93 publications

Mapping the carbon mitigation potential of photovoltaic development in the Gobi and desert regions of China

Mapping the carbon mitigation potential of photovoltaic development in the Gobi and desert regions of China

Application of chemometrics based on digital image analysis for simultaneous determination of tartrazine and sunset yellow in food samples

Estimating forest aboveground carbon sink based on Landsat time-series and its response to climate change

Contact Info

Product

Resources

About