Journal of Geographical Sciences

Spatial scaling effects of gully erosion in response to driving factors in southern China

LIU Zheng, WEI Yujie, CUI Tingting, LU Hao, CAI Chongfa

地理学报(英文版), 2024, 34(5): 942-962. DOI: 10.1007/s11442-024-2234-y

Site	Latitude/longitude	Area (km²)	Elevation (m)	Mean annual temperature (℃)	Mean annual precipitation (mm)	Gullies	Gully area (km²)
Southern China	20°13′-34°38′N, 104°28′-120°40′E	1,241,500	-205-3097	14-24	754.5-2312.3	23,9125	1220
Jiangxi province	24°29′-30°05′N, 113°34′-118°29′E	166,900	-181-2191	16.3-19.5	1341-1943	48,058	207.40
Ganxian county	25°26′-26°17′N, 114°42′-115°22′E	2993.09	61-1158	16.3-20.5	1600.5-1700	4138	18.1
Tiancun township	26°04′-26°14′N, 115°02′-115°16′E	222.21	93-792	17.2-19.8	1622.8-1682.1	590	1.69

Table 1 Detailed characteristics of the study area in southern China

本文的其它图/表

Figure 1 Typical gully landforms in southern China
Figure 2 Location of the study area (a. southern China; b. Jiangxi province; c. Ganxian county; d. Tiancun township)
Figure 3 Flowchart of methodology in this study
Table 2 Conditioning factors of gully erosion
Table 3 Factor elimination using multicollinearity analysis and Boruta algorithm
Figure 4 Analysis of the importance of variables using the Boruta algorithm (N, total nitrogen; P, total phosphorus; K, total kalium; SOM, soil organic matter; TWI, topographic wetness index; HI, hypsometric integral; NDVI, normalized difference vegetation index; FVC, fraction vegetation coverage; EVI, enhanced vegetation index; GDP, gross domestic product; LULC, land use and land cover; DEM, elevation)
Figure 5 ROC curves of gully susceptibility maps by MLR, LMT, and RF models in different spatial scales (a-c. southern China; d-f. Jiangxi province; g-i. Ganxian county; j-l. Tiancun township)
Figure 6 Comparison of model accuracy in terms of scale and data input
Figure 7 Gully erosion susceptibility maps using the random forest (RF) model (a. southern China; b. Jiangxi province; c. Ganxian county; d. Tiancun township)