Objective This study aims to evaluate the influence of remote sensing data from crop growth periods and topographic factors on improving the accuracy of black soil layer thickness inversion and mapping. Methods The study was conducted in a small watershed in Hailun Farm. 8 m high-resolution remote sensing data from both the bare soil and crop growth periods, SRTM digital elevation model (DEM) data, and in situ soil sampling data were comprehensively used. Three machine learning models,such as random forest (RF), gradient boosting decision tree (GBDT), and extreme gradient boosting (XGBoost) were applied for spatial prediction and mapping. Additionally, model performance was evaluated using the coefficient of determination (R²) and root mean square error (RMSE). Results 1) After incorporating remote sensing features from the crop growth period, the inversion accuracy of all models was significantly higher than that of models using data from the bare soil period alone. Specifically, the R² values of the RF, GBDT, and XGBoost models increased by 0.09, 0.11, and 0.10, respectively, while RMSE decreased by 1.15, 1.34, and 1.35, respectively. 2) By adding topographic factors to the integrated models, the accuracy was further improved. The R² values of the three models increased by 0.03, 0.12, and 0.05, respectively, while RMSE decreased by 0.38, 1.49, and 0.81, respectively. 3) The XGBoost model delivered the best performance across all feature combinations (R²=0.60, RMSE=12.31), significantly outperforming both GBDT (R²=0.53, RMSE=13.27) and RF (R²=0.46, RMSE=14.33), and exhibited stronger explanatory power when combined with topographic factors. Conclusion This study successfully constructs an 8 m high-resolution map of black soil layer thickness. The results demonstrate the effectiveness of the “multi-temporal remote sensing + topographic factors + machine learning” technical paradigm in improving the accuracy of soil property mapping. This provides high-precision spatial decision-making support for precision agriculture practices, sustainable utilization of black soil resources, national black soil conservation strategies, and food security assurance.
FANH M, CAIQ G, CHENG, et al. Comparative study of the soil erosion and control in the three major black soil regions in the world[J].Journal of Natural Resources,2005,20(3):387-393.
LIUR X, WANGZ Q, TANY P. The thickness of black soil layer and its influencing factors in a small watershed of typical black soil region in northeast China[J].Journal of Soil and Water Conservation,2024,38(4):346-353.
[8]
TESFAT K, TARBOTOND G, CHANDLERD G, et al. Modeling soil depth from topographic and land cover attributes[J].Water Resources Research,2009,45(10): W10438.
[9]
GAOQ Z, KANGM Y, XUH M, et al. Optimization of land use structure and spatial pattern for the semi-arid loess hilly-gully region in China[J].Catena,2010,81(3):196-202.
[10]
KURIAKOSES L, DEVKOTAS, ROSSITERD G, et al. Prediction of soil depth using environmental variables in an anthropogenic landscape, a case study in the western ghats of Kerala, India[J].Catena,2009,79(1):27-38.
[11]
ZHANGY, GUOL, CHENY Y, et al. Prediction of soil organic carbon based on landsat 8 monthly NDVI data for the Jianghan Plain in Hubei Province, China[J].Remote Sensing,2019,11(14):e1683.
[12]
DE SOUSA MENDESW, DEMATTÊJ A M, SILVERON E Q, et al. Integration of multispectral and hyperspectral data to map magnetic susceptibility and soil attributes at depth: A novel framework[J].Geoderma,2021,385:e114885.
[13]
YANGC, YANGL, ZHANGL, et al. Soil organic matter mapping using INLA-SPDE with remote sensing based soil moisture indices and Fourier transforms decomposed variables[J].Geoderma,2023,437: e116571.
[14]
KOBAYASHIY, YOSHIDAK. Quantitative structure-property relationships for the calculation of the soil adsorption coefficient using machine learning algorithms with calculated chemical properties from open-source software[J].Environmental Research,2021,196:e110363.
[15]
XUK K, DAIH M, ZHANGX J, et al. Black soil layer thickness prediction and soil erosion risk assessment in a small watershed in northeast China[J].PLoS One,2025,20(6):e0324368.
[16]
ZHANGS, LIUG, CHENS L, et al. Assessing soil thickness in a black soil watershed in northeast China using random forest and field observations[J].International Soil and Water Conservation Research,2021,9(1):49-57.
[17]
GONGZ T, CHENH Z, WANGH L. The distribution regularity of higher categories in Chinese soil taxonomy[J].Chinese Geographical Science,1997,7(1):1-10.
[18]
Soil Survey Division Staff. Soil survey manual[M]. Washington D C: US Department of Agriculture,1993.
[19]
SCHOENEBERGERP J, WYSOCKID A, BENHAME C,et al. Field book for describing and sampling soils[M]. USDA-NRCS: US Government Printing Office,2012.
[20]
BREIMANL. Random forests[J].Machine Learning,2001,45(1):5-32.
[21]
FRIEDMANJ H. Greedy function approximation: A gradient boosting machine[J].Annals of Statistics,2001,29:1189-1232.
[22]
CHENT Q, GUESTRINC. XGBoost: A scalable tree boosting system[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.San Francisco California USA. ACM,2016:785-794.
[23]
张之一.黑土开垦后黑土层厚度的变化[J].黑龙江八一农垦大学学报,2010,22(5):1-3.
[24]
ZHANGZ Y. The thickness changes of ah horizon after the phaeozems cultivated[J].Journal of Heilongjiang Bayi Agricultural University,2010,22(5):1-3.
[25]
BASTIDAF, ZSOLNAYA, HERNÁNDEZT, et al. Past, present and future of soil quality indices: A biological perspective[J].Geoderma,2008,147(3/4):159-171.
[26]
ZHANGY, LUOC, ZHANGY H, et al. Integration of bare soil and crop growth remote sensing data to improve the accuracy of soil organic matter mapping in black soil areas[J].Soil and Tillage Research,2024,244:e106269.
[27]
XIONGY L, LIY R, XIONGS Y, et al. Multi-scale spatial correlation between vegetation index and terrain attributes in a small watershed of the upper Minjiang River[J].Ecological Indicators,2021,126:e107610.
LIUH J, BAOY L, MENGX T, et al. Effect of topography on crop growth in black soil area[J].Transactions of the Chinese Society for Agricultural Machinery,2020,51(9):274-283.
LIUK, DAIH M, LIUG D, et al. Organic carbon content-baesd prediction and influencing factors of black soil layer thicknesses[J].Geophysical and Geochemical Exploration,2024,48(5):1368-1376.
GABETE J, MUDDS M, WOODR W, et al. Hilltop curvature increases with the square root of erosion rate[J].Journal of Geophysical Research: Earth Surface,2021,126(5):e2020JF005858.
[35]
ROERINGJ J. How well can hillslope evolution models explain topography? Simulating soil transport and production with high-resolution topographic data[J].Geological Society of America Bulletin,2008,120(9/10):1248-1262.
[36]
PATTONN R, LOHSEK A, GODSEYS E, et al. Predicting soil thickness on soil mantled hillslopes[J].Nature Communications,2018,9:e3329.
[37]
FENGZ Z, ZHENGF L, HUW, et al. Impacts of mollic epipedon thickness and overloaded sediment deposition on corn yield in the Chinese mollisol region[J].Agriculture, Ecosystems and Environment,2018,257:175-182.