The accurate prediction of the spatial distribution of soil organic carbon (SOC) and the identification of the mechanisms underlying its spatial differentiation are of paramount significance for the conservation and utilization of land and regional sustainable development. A total of 512 soil samples were collected from Wuchang and Shuangcheng County in Harbin City, Heilongjiang Province, China, which served as the study area. Six machine learning models, including Random Forest (RF), AdaBoost, Support Vector Regression (SVR), weighted average, Stacking, and Blending, were utilized to predict the spatial distribution of SOC and analyze its spatial differentiation. The result reveals that 12 environmental variables, including soil type, bulk density, pH, average annual precipitation, average annual temperature, net primary productivity (NPP), land use type, normalized difference vegetation index (NDVI), slope, elevation, soil parent material, and distance to rivers, are effective influencing factors on SOC in the study area. It turns out that the Stacking model, with an R² of 0.4327, performed the best in this study, followed by the weighted average, Blending, RF, AdaBoost, and SVR models; a heterogeneous integrated learning model may be more robust than an individual learner. The predicted SOC content is generally lower in the northwestern arable land and higher in the southeastern forest land. In addition, SOC differentiation shows that forest land and grass land with dark brown soil or swamp soil, soil covering igneous and metamorphic rocks with various minerals, higher elevation and slope, and suitable water-thermal and soil intrinsic conditions for aerobic microbial activity benefit the enrichment of SOC in the study area. The enrichment and depletion of SOC are jointly influenced by pedogenesis, microbial activity, and biodiversity.

As the need to discovers new mineral deposits and occurrences has intensified in recent years, it has become increasingly apparent that we need to map potentials via integrated information on the basis of metallogeny. Occurrences of mineralization such as tungsten (W), tin (Sn), columbium (Nb), tantalum (Ta), gold (Au), copper (Cu), lead (Pb), zinc (Zn), manganese (Mn) and monazite (Mnz) have been discovered in Rwanda. The objective of this study was to present a regional quantitative mineral prospectivity mapping (MPM) of W, Sn and Nb-Ta mineralization in Rwanda using the random forest (RF) method on the basis of open source data, such as geological maps, Bouguer gravity anomalies, magnetic anomalies, Landsat 8 images, ASTER GDEM, Globeland30, and OpenStreetMap. In addition, a newly introduced interpolation–density–delineation (IDD) process was applied to deal with the blank (masked) areas in remotely sensed mineral alteration extraction. Additionally, a k²-fold cross-validation method was also proposed to obtain more reasonable test errors. Firstly, the metallogenic regularity of W, Sn and Nb-Ta in Rwanda was summarized with the help of articles online. Secondly, original geological, geophysical, and remote sensing data were utilized to generate secondary data. Specifically, the IDD process was applied subsequent to the directed principal component analysis method (DPCA) to reconstruct the alteration anomaly map, and a relevant dataset was formed by the combination of original and secondary data. Thirdly, specific predictor layers for W, Sn and Nb-Ta were selected from relevant data via spatial correlation with known deposits, respectively, and the predictive models were established. Finally, near 26,000 squares were zoned in Rwanda, and RF was optimized and applied, the k²-fold cross-validation method was utilized to assess test errors, metallogenic belts and prospective areas for W, Sn, and Nb-Ta were delineated on the basis of total mineralization potential map and likelihoods map. Results proved that the open source data online were valid for drawing a preliminary mineralization potential map. Furthermore, it was also shown that the IDD method is suitable for the postprocessing of masked alteration anomaly maps. Belt IV-4 in the northwest and belt IV-2, IV-1 in the middle-east of Rwanda, containing a number of prospective areas, possess considerable likelihoods of deposits, and mining in Rwanda is at its dawn, with potential worth expecting.

The performance of remote sensing-based mineral alteration extraction is significantly restricted in the vegetation area. Spectral unmixing is one of the effective methods to address the vegetation problem during mineral alteration extraction. However, the spectral curves of different tree species vary a lot; if multiple tree species are regarded as a whole during the spectral unmixing stage, the proportions of vegetation would be estimated with more errors. The purpose of this study was to verify the effects of dominant tree species classification on spectral unmixing and reconstruction, and to apply the proposed method to the mineral alteration extraction practice. To accomplish this, the Shabaosi gold deposit region in Mohe City, China, with an area of 650 km², was selected as the study area. Firstly, reference spectral curves, GaoFen-1/6 (GF-1/6) satellite imageries, ZiYuan-1F (ZY-1F) satellite imageries, Sentinel-1B satellite synthetic aperture radar (SAR) data, the ALOS digital elevation model (DEM), and sub-compartment dominant tree species data were collected; subsequently, simulated mixed-pixel reflectance images of ZY-1F, reflectance images of GF-1/6, ZY-1F, backscattering data of Sentinel-1B, slope, aspect, and 5484 tree species samples were derived from the collected data. Secondly, to verify the effect of dominant tree species classification on mineral alteration extraction, the reference spectra of pine, oak, goethite, and kaolinite were used to construct a simulated ZY-1F mixed-pixel image, and spectral unmixing and reconstruction experiments were conducted. Thirdly, fourteen independent variables were selected from the derived data, five dominant tree species classification models were trained and tested using tree species samples via the ResNet50 algorithm, and the pine- and birch-dominated parts were segmented from the ZY-1F images. Fourthly, minimum noise fraction (MNF), pixel purity index (PPI), n-dimensional visualizer auto-clustering, and spectral angle mapper (SAM) methods were separately applied to the pine- and birch-dominated parts of ZY-1F images to extract and identify endmembers; subsequently, the fully constrained least squares (FCLS) and linear spectral unmixing (LSU) methods were separately applied to the pine- and birch-dominated parts to estimate endmember proportions and generate spectrally reconstructed ZY-1F images. Fifthly, the pine- and birch-dominated parts of spectrally reconstructed ZY-1F images were mosaiced, and the SAM was utilized to extract mineral alteration in the study area. The result showed that in the spectral unmixing and reconstruction experiment, the spectral reconstruction error declined from 0.0594 (simulated ZY-1F image without segmentation) to 0.0292 and 0.0388 (simulated ZY-1F image that was segmented by pine- and oak-dominated parts), suggesting that dominant tree species classification could improve the accuracy of spectral unmixing and reconstruction and help obtain a more reliable mineral alteration extraction result. In the study area, the tested overall accuracies (OA) and Kappa coefficients of the five dominant tree species classification models were 0.75 ± 0.03 and 0.50 ± 0.05, respectively, suggesting that conducting dominant tree species classification was feasible in dense vegetation areas and could facilitate mineral alteration extraction. After segmenting the ZY-1F image by pine- and birch-dominated parts and spectral reconstruction, eight main types of alteration, including kaolinite, vesuvianite, montmorillonite, rutile, limonite, mica, sphalerite, and quartz, were identified, and nine mineral alteration areas (MA) were delineated accordingly.