With the increase in the availability of tools which enable researchers to merge GIS and geostatistical operations along with free remote sensing and relief data at the global scale with fine resolution, the pixel-based technique of digital soil mapping (DSM) has become increasingly preferred as an alternative to overcome some limitations of previously used soil mapping approaches. Although the most typical, a frequently used approach to understanding the complex changing behavior of soils even at unvisited sites in a vast area has been geostatistical methods however, this has changed in recent years.
Moreover, the classic sampling techniques and mapping approaches produce soil property maps with a spatial resolution that is not particularly useful for ecosystem management. Traditional vector-based or polygon maps of soil types and properties are cost and time-intensive to produce. It is concluded that, in modeling soil texture classes using RF models through a digital soil mapping approach, data should be balanced before modeling. Similar increasing trends were observed for the recall and F-scores.
Balanced data also improved the accuracies from 44% to 59% and 0.30 to 0.52 with regard to the overall accuracy and kappa values, respectively. Results showed that based on mean minimal depth (MMD), when imbalanced data was used, distance and annual mean precipitation were important, but when balanced data were employed, terrain attributes and remotely sensed data played a key role in predicting soil texture. Bioclimatic and remotely sensed data, distance, and terrain attributes were used as environmental covariates to model and map soil textural classes. A synthetic resampling approach using the synthetic minority oversampling technique (SMOTE) was employed to make a balanced dataset from the original data. The original dataset (imbalanced) included 6100 soil texture data from the surface layer of agricultural fields in northern Iran. The aim of this study was to investigate the effects of imbalance in training data on the performance of a random forest model (RF). While real-world datasets are inherently imbalanced, ML models overestimate the majority classes and underestimate the minority ones. Recently, demands for applying machine learning (ML) methods to improve the knowledge and understanding of soil behavior have increased. Soil provides a key interface between the atmosphere and the lithosphere and plays an important role in food production, ecosystem services, and biodiversity.
0 kommentar(er)
