Evaluation of the soil carbon sequestration potential and toward digital soil mapping under semi-arid Mediterranean ecological condition

• Published in: Euro-Mediterranean journal for environmental integration 2024-06, Vol.9 (2), p.997-1007
• Main Authors: Alaboz, Pelin, Dengiz, Orhan
• Format: Article
• Language: English
• Subjects: Algorithms; Aquatic Pollution; Atmospheric Protection/Air Quality Control/Air Pollution; Bulk density; Carbon; Carbon sequestration; Climate; Correlation coefficient; Correlation coefficients; Decomposition; Digital mapping; Distribution patterns; Earth and Environmental Science; Earth Sciences; Ecological conditions; Environmental Chemistry; Environmental Management; Environmental Science and Engineering; Laboratories; Land use; Machine learning; Macroporosity; Methods; Microporosity; Moisture content; Organic carbon; Original Paper; Physical properties; Root-mean-square errors; Salinity; Sand; Soil erosion; Soil fertility; Soil mapping; Soil physical properties; Soil properties; Soils; Spatial distribution; Statistical analysis; Support vector machines; Waste Management/Waste Technology; Waste Water Technology; Water content; Water Management; Water Pollution Control; Wilting; Wilting point
• ISSN: 2365-6433; 2365-7448
• DOI: 10.1007/s41207-024-00512-4

In this study, it was aimed to evaluate the relationship between the carbon sequestration potential (CSP) of soils and some soil physical properties. In addition, the predictability of CSP with the support vector regression (SVR) algorithm and the most successful interpolation method in distribution maps of observed and predicted values were determined. The CSP of the soils in the study area was determined to be 43.53 t C ha −1 and 78.09 t C ha −1 . Negative correlations were found between CSP and macroporosity, sand, and bulk density, and positive statistically significant correlations were found with organic carbon, available water content, permanent wilting point and microporosity. The CSP was predicted by the SVR algorithm. The root mean square error (RMSE), Lin’s concordance correlation coefficient (LCCC), and ratio of performance to deviation (RPD) were determined to be 7.67, 0.18, and 0.93, respectively. The predicted interval (PI) was determined to be 47.60 t C ha –1 and 67.03 t Cha –1 . In general, it was found that the error rates increased with a higher than 60% sand and 35% clay content. The simple kriging method with the lowest error rate was determined for the spatial distribution of the CSP. The distribution patterns of the predicted map and the actual value map were not found to be similar. It has been evaluated that it is crucial in the follow-up of sensitive areas, especially with the creation of digital maps created with forecasting models.