Abstract
Increasing agroforestry areas with improper management has produced serious environmental problems, such as soil erosion. It is necessary to rapidly predict the spatial distribution of such erosion risks in a large area, but there is a lack of approaches that are suitable for mountainous regions. The objective of this research was to develop an approach that can effectively employ remotely-sensed and ancillary data, to map soil erosion risks in an agroforestry ecosystem in a mountainous region. This research employed field survey data, soil-type maps, digital elevation model data, weather station data, and Landsat imagery, for extraction of potential variables. It used the random forest approach to identify eight key variables—slope, slope of slope, normalized difference greenness index at leaf-on season, soil organic matter, fractional vegetation at leaf-on season, fractional soil at leaf-off season, precipitation in June, and percent of soil clay—for mapping soil erosion risk distribution in hickory plantations in Western Zhejiang Province, China. The results showed that an overall accuracy of 89.8% was obtained for three levels of soil erosion risk. Approximately one-fourth of hickory plantations were at high-risk, requiring the owners or decision makers to take proper measures to reduce the soil erosion problem. This research provides a new approach to predict soil erosion risk, based on the primary variables that can be extracted directly from remotely-sensed data and ancillary data. This proposed approach will be valuable for other agroforestry and plantations, such as Torreya grandis, eucalyptus, and the rubber tree, that are playing important roles in improving economic conditions for the local farmers but face soil erosion problems.
Highlights
Agroforestry area has continuously expanded globally, due to the requirement of human well-being and improved standard of living
Slope and slope of slope (SOS) were extracted from the DEM; NDGI, fractional vegetation, and fractional soil were extracted from the Landsat multispectral imagery; soil organic matter (SOM) and percent of soil clay were extracted from the soil data; and precipitation in June was gleaned from the interpolation of the weather station data
This research used the random forest (RF) approach to identify eight primary variables—slope, SOS, NDGI at the leaf-on season, fractional vegetation at the leaf-on season, fractional soil at the leaf-off season, SOM, clay, and average precipitation in June—that could be directly extracted from the DEM, the Landsat multispectral imagery, soil, and the weather station data
Summary
Agroforestry area has continuously expanded globally, due to the requirement of human well-being and improved standard of living. [F6o,7r]t.hFeosratkhee osaf kinetoefnisnivteenmsivane amgaenmaegnetmaenndt eaansdyecaoslylecotilolenctoiof nhiocfkohricyknourytsn, ufatrsm, fearrsmoefrtsenofrtemn orevme oshvreusbhsruabnsdagnrdasgsr,aasts,leaatslteatwstictwe ipceerpyeerayr,eainr, sinprsipnrgi,nbge, fboerfeortehethaedadditdioitnionofonfunturiterinetns,tsa,nadndinineaeralrylySSeepptetemmbbeer,r,bbeeffoorreetthhee nnuutt harvest (personal communiiccaattiioonnss with farmers in Zhejiang Province dduuring fifieellddwwoorrkk). Agroforestry has some common characteristics—single tree species and simple stand structure, lack of shrubs and grass under tree ctraeneopcayn, oapnyd, saenrdiousesrisoouils esrooislioenropsiroonblpemrosbl[e3m,4s]. These hickory plantations have serious soil erosion problems because there is almost noo uunnddeerrssttoorryy, most of the year, as shown in Figuurree 11..TThheessooilileerorosisoionnisisesepsepceicailalyllysesreioriuosuisnitnhethreairnayinsyeasseoanso, dnu, driunrginMgaMy anydanJudnJeu, nane,dainndthine ltehaef-loefaff-soeffassoenasofrnomfromNoNveomvebmerbetor tMo Marachrc/hA/pAripl,riiln, inthteheZZhehjeiajinagngPProrovvinincec,e,rreeqquuiririningg aa suitable approach to quickly evaluate the soil erosion proobblleemm [[55]]. TTwo photos showing the hickory plantations in leaf-on and leaf-off seasons
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have