| The reclamation of cropland in mining areas is an effective way to solve the problem of cropland disturbance from the overlapped regions of grain production and mineral exploitation in China.The monitoring and evaluation of reclaimed cropland in mining areas is an important part of the reclamation project,which helps to guide the implementation of the reclamation project.The severe soil disturbance and gradual recovery of the cropland productivity in the mining area make the spatial variation of crop growth response more significant than the regular cropland,so it is necessary to carry out detailed monitoring and rapid evaluation.However,the traditional sampling analysis methods depend on the reasonable sampling point layout.The monitoring cost increases as the increasing sampling points,which is not conducive to long-term continuous monitoring.Satellite remote sensing approaches have a relatively low spatial and temporal resolution in regards of crop growth cycle,which makes it difficult for detailed crop monitoring in the reclaimed region.The emerging monitoring methods(Li DAR and ground spectrometer)is relatively expensive and captures a large amount of data,which is difficult to extensive popularization.These discussed methods have been a bottleneck for the detailed monitoring and rapid evaluation of reclaimed cropland in mining regions.In recent years,an emerging remote sensing technique using small unmanned aerial system(s UAS)has received a wide range of attention from all kinds of small-scale research due to its advantages such as low cost,high accuracy,prompt response,short cycle time,and easy operation.Numerous studies have developed many models for reclamation-concerned crop indicators,for example above-ground biomass(AGB),yield,plant height.This technique has showed up in detailed monitoring of mining reclamation on a small scale.It has the great potential for monitoring the reclaimed cropland in mining areas.Therefore,it should be paid the attention to introduce the s UAS remote sensing technology into the detailed monitoring and evaluation of the reclaimed cropland in the mining area.The reclamation of cropland in the mining area is constructed in stages and the cropland field is the basic cultivating unit.The severe disturbance of the soil profile leads to a greater spatial variance of crop growth response,and the gradual recovery of productivity makes it difficult to reveal the dynamic response characteristics of crop growth using the crop yield as an evaluation indicator.These problems block detailed monitoring and evaluation.This dissertation attempted to introduce the s UAS remote sensing technology into the monitoring and evaluation of the reclaimed cropland in mining areas at the basis of the research in related fields.This dissertation took the filling-reclaimed cropland using the Yellow River sediments in the mining-subsided region as an example,selected the crop AGB as the dynamic monitoring indicator for its linear relationship with crop yield.Then,wheat and corn AGB were estimated using the variables from the multi-feature data fusion,and detailed crop AGB monitoring was achieved and the rapid evaluation of the reclaimed cropland was conducted in the mining area according to automatically extracted cropland strips as basic analysis units.The main conclusions are as follows:(1)Considering the characteristics of phased construction of reclaimed cropland in the mining regions and the cultivated unit of the cropland strip,an automated mapping method of typical cropland strips was proposed.The cropland ridges and strips were extracted and verified after the analysis of s UAS image characteristics of cropland strips.The results showed that the proposed method was reliable and highly accurate.The accuracy of cropland strip extraction was greater than 98.9%and the Kappa coefficient was larger than 97.4%.The accuracy of cropland ridge detection has a high recall(>97%)and precision(>95%).(2)In regard of the characteristics of severe soil disturbance and significant spatial heterogeneity of the reclaimed cropland in the mining area,the s UAS multispectral images were used to construct a canopy height model of the cropland in the mining area through a series of processing,including two filtering steps of the dense point cloud generated from the multispectral images,filtered point cloud interpolation,raster operation and vegetation mask.The accuracy of the constructed CHM was verified after the determination of plant height extraction threshold(quantile 99%)to eliminating canopy porosity problem.The results showed that the multispectral data from s UAS platform can be used to construct the crop CHM,and the CHM accuracy is higher when the flight height of above ground level(AGL)is lower.At the AGL of 50m,the RMSE of wheat CHM is 5.3～8.2 cm,and the RMSE of corn CHM is 10～11.8 cm.(3)Aiming at the characteristics of the dynamic evolution of the soil environment and the gradual recovery of productivity of the reclaimed cropland in the disscussed regions,two new metrics were proposed to improve the estimation accuracy of above-ground biomass in the way of data fusion for wheat and corn,respectively.A multi-feature fusion model was proposed using the canopy spectral response,structural characteristics and meteorological data for the wheat,and a model was constructed using the canopy structure characteristics and meteorological data for the corn.Firstly,the volume-meteorology weighted canopy spectral response metric(VM-CSRM)was proposed for the wheat combining the canopy volume model(CVM),growing degree days(GDD)and the most correlative vegetation index(VI)with wheat biomass,and the meteorology weighted canopy volume model(M-CVM)was constructed for corn combining GDD using the suitable weight with CVM.Secondly,the optimal metrics were qualitatively and quantitatively selected for modeling and verification of AGB estimation.The results showed that the proposed metrics can improve the estimation accuracy of crop biomass.The optimal wheat estimation was obtained using binomial regression with the variable:CARI×CVM×GDD(R~2=0.8272,RMSE=0.1690kg/m~2),and the optimal corn estimation was obtained using exponential regression with the variable:CVM×GDD~2(R~2=0.8897,RMSE=0.1649kg/m~2).(4)In order to evaluate the reclaimed cropland in the mining regions,the characteristics of the spatiotemporal variation of the reclaimed cropand AGB were analyzed under two types of analysis units(the cropland strip and treatment plot).Firstly,the biomass grading criteria covering the whole growth stage of the crop was established to analyze the absolute changes of the crop biomass accumulation in response to the reclaimed soil environment.Secondly,the spatial autocorrelation analysis method was used to reveal the spatial relative variance of crop biomass accumulation in each period in response to the reclaimed soil environment.Thirdly,on the basis of this,the idea of the limit condition method is employed to evaluate of reclaimed cropland.The results showed that the cropland strip NO.B is relatively less stressed by the reclaimed soil environment in the view of analysis unit of the cropland strip.Treatment T09 was the optimal soil profile for inter-layers filling reclamation,and its configuration was“topsoil 30cm+subsoil 20cm+Yellow River Sediment 20cm+subsoil 20cm+Yellow River sediment30cm”. |
PDF Full Text Request