The Research And Application Of Spectra Identification On Several Kinds Of Economic Crops

The agriculture will change from traditional to modern, management mode fromextensive to intensive operations, with man’s stepping into the information society inthe21st century. At present, our country is facing the more and more sharp resource andenvironmental problems, it is imperative to promote the development of agriculturalscience and technology and implement the strategy of sustainable development toensure the food’s security of13hundred million people. Geologists experts think thatModern agriculture always be the only way ofAgricultural Sustainable Development.As the human environments gained a growing influence over the growth process ofthe crop, Identifying types and measuring crop chlorophyll content rapidly and timelybecomes more and more important. Hyper-spectral technology has become a powerfultools of crop growth monitoring by studying the relationships between crop reflectancespectra and chlorophyll content, It has been the research hotpot of the Precisionagriculture to monitor and diagnose crop nutrition timely and rapidly based on thereflective spectral characteristics of crop.Economic crop production is a low efficiency of agricultural economy, Thefarmers’ consumption comes from economic crop production. since the policy of “donot slacken efforts to food production, develop diversied economy actively”, manadjust the agricultural structure, and the ration of economic crop and farmers’ incomehave increased. therefore, the study of Spectra identification on several kinds ofeconomic crops, with practical significance.The thesis measured hyper-spectra reflectivity and crop chlorophyll content ofseeding-watermelon, beets and gourd by ASD FieldSpecPro spectrographs of analyticalspectral device company and the SPAD·502chlorophyll meter between May andSeptember,2012,analyzed their spectrum feature; did spectra identification study and contrasted the PCA-FDA and PCA-BP; estimated the chlorophyll content by pluralisticstatistic analysis methods, the characteristic spectrum and the vegetation index; Thenobtained the following results：1.Pretreat the Field Spectral Data, process the data by dimension reduce andprinciple components analysis, the first five principal components accounted for92.16%of total data, which present all the raw data; With five principal componentscores as the input variables, crop types as output variables, build PCA-BP andPCA-FDA model, the former’s fitting residual is5.2156×10-6, the exact recognition rateis100%; the latter has a rate is100%for verification rate and exact recognition rate. incontrast, PCA-FDAmodel bring better result.2.Chlorophyll content and spectral index had significantly close relativity, the bestis MSAVI2(R2=0.82), the Trinomial model is set up, which include dependent variableof MSAVI2and the independent variables, and it has the most high accuracy(R2=0.8066) and to be the best for the plant. The logarithmic model based on principalcomponent projection deterministic coefficient R2is0.8201, RMSE is0.9126, as thebest chlorophyll estimation model of the Gourd.3.Spectral index VARI of each growth period and beet yield had significantly closerelativity, it has a higher correlation coefficient between beet yield and VARI duringexpanding period of root tubers, then foliage-luxuriating period, seedling period, thesugar accumulation period, The correlation coefficient were0.8306,0.8107,0.8076and0.8015; Regression equation were set between beet yield and VARI of each growthperiod, and all had significantly close relativity, The cubic’s RMSE were0.0882,thelowest, the highest precision of single phase spectrum evaluates yield; Amongcompound spectrum estimate the yield models of spectrum variable agronomyParameters-yield, their models precisions of four periods compound spectrum estimatethe yield models are all the best (R2,0.960; RMSE,0.127).