Analyzing Spatial Scale Problems Of Crop Growth Parameters For Growth Monitoring With Multi-scale Remote Sensing Data

Remote sensing technique is increasingly applied to the crop growth monitoring in the research field of precision agriculture, such as crop growth parameters inversion, crop growth ranking system construction, space consistency analysis for multi-scale crop growth monitoring results, etc. In this dissertation, the author, considering the actual demands of crop growth monitoring, investigated several problems arising from growth monitoring with multi-scale remote sensing data, and achieved the following research work:(1) According to the application requirements of crop growth monitoring, the author constructed two new growth paramters, i.e. CGMI1and CGMI2(Crop Growth Monitoring Index, CGMI), comprehensively and quantitatively describing crop growth condition from the aspects of morphology structure and biological activity of crop canopies. These two new crop growth paramters are constructed based on the abilities of Leaf Area Index (LAI) and Canopy Chlorophyll Density (CCD) in representing crop canopy morphology structure and crop canopy biological activity, respectively. Compared with the crop growth parameters available, these two new growth parameters can provide a much more comprehensive description of the characteristics of crop growth status.(2) According to the application requirements of constructing crop growth ranking system with multi-scale remote sensing observations, the author proposed a spatial adaptive threshold selection method, and carried out researches on crop growth monitoring and ranking based on the selected thresholds of CGMI1and CGMI2. The method, integrating data histogram and Gaussian distribution theory (or Skew-normal distribution theory) for thresholds selection according to the statistical characteristics of CGMI1and CGMI2, is able to strengthen the robustness of threshold selecting at different spatial scales. By applying the proposed method, the selected thresholds of CGMI1and CGMI2are utilized to constructe crop growth ranking system from the aspects of crop canopy morphology structure (large, medium and small) and crop canopy biological activity (strong, middle and weak).(3) In order to quantitatively describe the whole differences between inversed crop LAI datasets based on multi-scale remote sensing data, and describe the whole differences between inversed crop CCD datasets based on multi-scale remote sensing data, the author designed a data analyzing algorithm. In this algorithm, the whole differences of multi-scale inversed crop growth parameters are described from the following aspects:the differences caused by remote sensing reflectance datasets at different spatial scales, the differences caused by different crop LAI and CCD inversion models, and the differences caused by spatial scale effects of crop LAI and CCD. The substance of this algorithm is to carry out hierarchy analysis of the whole differences, to describe the contribution of each influence factors for the differences between multi-scale inversed parameters by means of statistical analysis theory, and then to provide references for multi-scale remote sensing observations correction.(4) In order to quantitatively describe the multi-scale inversion differences between crop growth parameters caused by remote sensing reflectance datasets at different spatial scales, the author proposed a data analysis and correction algorithm based on Gaussian distribution theory. Under the condition of the existence of smaller-scale remote sensing data, the algorithm makes the statistical analysis and Gaussian distribution patterns correction of larger-scale remote sensing data, so that the differences caused by remote sensing reflectance datasets at different spatial scales is reduced effectively. Furthermore, the algorithm can be applied to heterogeneity underlying surface, because the precondition of this algorithm only pay attention to data volume for analyzing, and there is no requirements on data spatial distribution characteristics.(5) In order to quantitatively describe the multi-scale inversion differences between crop growth parameters caused by spatial scale effects of crop LAI and CCD, the author proposed a spatial scaling model based on the theory of the collocation approximation of Chebyshev polynomial, to correct the scale effects of crop growth parameters. The proposed model not only quantitatively describes the scale effects from the aspects of the nonlinearity of LAI and CCD inversion models and the nonlinearity of the driving variables of inversion models, but also has better applicability in heterogeneity underlying surface compared with the spatial scaling model based on Taylor series expansion method.(6) According to the application requirements of space consistency analysis for multi-scale crop growth monitoring results, the author carried out corresponding research, aiming to quantitatively analyze the spatial distribution patterns and characteristics of crop growth status at different spatial scales based on multi-scale remote sensing observations correction and spatial scaling of crop growth parameters, and further providing references for feasibility and reasonableness analysis of multi-scale crop growth monitoring under heterogeneity underlying surface.All methods, algorithms, and models proposed above are verified by actual multi-scale remote sensing observations of there crops:wheat, corn, and barley, in four experimental fields located in Yingke Oasis in the middle reach of Heihe River Basin (oasis underlying surface), Labudalin farm of Hailaer Farming Cultivate Bureau in Inner Mongolia (farm underlying surface), major grain production areas in Hebei province (farmland underlying surface), and Shunyi District and Changping District of Beijing (suburban underlying surface) for multi-scale crop growth monitoring and corresponding space consistency analysis.