Estimation Of Available Nutrients In Orchard Soils Based On Hyperspectral

The levels of Available nutrients Available Nitrogen（AN）,Available Phosphorus（AP）and Available Kalium（AK）in orchard soils directly affect the growth and development of fruit trees and fruit quality.Blind fertilization not only increases production costs and pollutes the environment,but also reduces fruit quality.Therefore,timely and accurate monitoring of the content levels of major fast-acting nutrients in orchard soil is important for scientific fertilization and accurate management of orchard quality.Conventional soil sampling and analysis methods are prone to damage to plant roots and have disadvantages such as high laboratory costs,long cycle time and pollution,which cannot meet the needs of timely monitoring and large-scale management of modern orchards.Agricultural hyperspectral remote sensing provides the technical means to achieve nondestructive,rapid,accurate and large-scale monitoring of soil.In this study,a hilly apple orchard in Shuangquan Town,Changqing District,Jinan City,Shandong Province was used as the study area to explore the hyperspectral characteristics of apple orchard soil and its relationship with the content of major fast-acting nutrients,to screen the sensitive bands and determine the characteristic factors,to construct the best model for the estimation of the preferred AN,AP and AK,and to propose the technical process scheme for the estimation.The main research contents and results are as follows.（1）The statistical characteristics of the main fast-acting nutrients in sandy brown soils in apple orchards were analyzed.A descriptive statistical analysis of AN,AP and AK in the apple orchard revealed that the soil AN（98.00mg/kg）and AK（21.50mg/kg）were at a low level,which was related to the sampling time of 10 days before fertilization and the high demand and consumption of these two elements during the young fruit stage.（58.90mg/kg）was at a high level,which may be related to the preference for P fertilization in the orchard.（2）Soil hyperspectral characteristics and the relationship between spectral reflectance and major fast-acting nutrients in apple orchards were analyzed.The hyperspectral characteristics of soil in apple orchards were enhanced by breakpoint correction and denoising,and the relationship between spectral reflectance and soil fast-acting nutrient content was analyzed.The reflectance of the original soil spectrum was mainly concentrated in the range of 0.05-0.45,and the trends of the reflectance curves in the wavelength range of each soil sample were basically the same;the soil spectral curves in the range of 400-2100 nm showed an increasing trend in general,and those in the range of2100-2450 nm showed a decreasing trend in general;there were more obvious water absorption valleys near 1400nm,1900nm and 2200nm.There are more obvious water absorption valleys around 1400nm,1900nm and 2200nm.Soil AN content was positively correlated with spectral reflectance,and the higher the soil AN content,the higher the spectral reflectance.Soil AP and AK contents were negatively correlated with spectral reflectance,the higher the soil AP and AK contents,the lower the spectral reflectance.（3）The study scales,characteristic factors and sensitive wavebands of the main fast-acting nutrients in apple orchard soils were determined.Based on the coupling process of mathematical transformation and continuous wavelet transform of the original spectral reflectance,the research scale and characteristic factors with high spectral sensitivity were determined by correlation analysis,and the sensitive bands were screened based on Stepwise Multiple Linear Regression（SMLR）analysis.The research scales of soil AN were determined to be scales 6 and 7,and the feature factors were 1515,1518,1521,1567,1569,1853,1854,1887 nm;the research scales of soil AP were scales 5 and 8,and the feature factors were 1060,1080,1081,1096,2352 nm;the research scales of soil AK were scales 1 and 7 The best modeled feature bands for AN,AP and AK of orchard soil were737,803,1567 and 1671 nm;the best modeled feature bands for AP were 459,1237 and 1953nm;the best modeling feature bands of AK are 414,423,438,463,492,538,544,802,820,824,901,1047,1142,1277,1300,1393,1694,1995,2050,2123,2251,2327,2401,2449 nm.（4）A model for estimating the main fast-acting nutrients of sandy brown soil in orchards was constructed and optimized.Multiple Linear Regression analysis（MLR）and Partial Least Squares Regression（PLS）were used to construct and optimize the estimation models for AN,AP and AK contents in apple orchard soils,respectively.Among them,the best estimation model for AN content was（√R）′_AN-CWT-MLR with modeling set R~2of 0.7422 and RMSE of 17.6443;the best estimation model for AP content was type（√R）′_AP-CWT-MLR with modeling set R~2of0.6661 and RMSE of 28.2410;the best estimation model for AK was Ln R_AK-CWT-MLR,with modeling set R~2of 0.8707 and RMSE of 7.8748.（5）A technical solution for hyperspectral monitoring of soil fast-acting nutrients in apple orchards was proposed.Based on the key technical methods and whole-process links for hyperspectral estimation of major fast-acting nutrients in apple orchard soils,a process-oriented rapid hyperspectral monitoring technology scheme for fast-acting nutrients in apple orchard soils oriented to daily management needs is proposed by integrating and optimizing the research results.This paper explores the spectral characteristics of AN,AP and AK in apple orchard soils by using soil hyperspectral inversion technology,establishes and optimizes the estimation model,and proposes a technical scheme for rapid monitoring of major fast-acting nutrients in apple orchard soils,which provides a reference scheme for the application of hyperspectral technology to daily monitoring and management of soil nutrients in orchards.