Monitoring And Predicting Growth Status By Integrating Remote Sensing And Wheat Model

Coupling remote sensing data with crop growth models to predict crop yield is a hot research topic.Integration of the real-time,regional remote sensed data and growth model of mechanism and predictability can achieve crop condition monitor and yield estimation in large area.At present,the coupling methods of remote sensing and crop model are mostly based on the initialization/parameterization method,namely,the coupling points are agronomic parameters of remote sensing inversion.This method is based on the assumption that remote sensing inversion values are more accurate than the simulation values of model,but in fact there are large errors in the agronomy parameter of remote sensing inversion.Therefore,errors produced by agronomy parameter inversion can be avoided if directly using remote sensing information(e.g.,vegetation index)as the coupling point.In addition,the process coupling remote sensing with crop growth model to invert initial parameters mainly adopts the method of grid computing,which need to run the growth model many times repeatedly,bringing large amount of calculation.In recent years,the problem of calculation efficiency has become one of hot topics in the study of coupling remote sensing with crop model.There are two coupling strategies in this research:one is based on the initialization/parametrization method,using multi-source through statistical model for remote sensing information inversion agronomy parameter(LAI and LNA),namely "remote sensing inversion value".By comparing the "remote sensing inversion value" and simulation values of wheat growth model(WheatGrow),some management parameters which are difficult to accurately obtained by inverting,including sowing date,sowing rate and nitrogen rate;The other one is based on the assimilation method,directly comparing to the canopy spectra,which are simulated by the coupling model of PROSAIL model and wheat growth model(WheatGrow),with the canopy spectra from multi-source remote sensing to invert the management parameters in the study area.By comparing the two coupling strategies,correctness verification results show that when "remote sensing inversion value" are accurate enough,inversion of sowing date,sowing rate and nitrogen rate can be more accurate,and inversion strategy using the coupling point based on "remote sensing observation value" did not show obvious advantages;The coupling prediction results using external assimilation data based on ground spectral measurement data show that prediction results using the coupling index of "remote sensing observation value" are obviously better than that using " remote sensing inversion values".Possible reasons are that remote sensing inversion agronomy parameter process is not accurate.Based on the simulation zoning method,wheat canopy RVI at different growth stages and soil nutrient indices,including organic matter content,total nitrogen content and available potassium content,were selected as data sources to delineate management zones.Results show that the values of variation coefficients of each zone,including organic matter content,total nitrogen content,were between 2.8～6.6%,34.2～55.1%and 5.7～8.2%,which were less than their values in the whole area range of 7.63%,65.99%and 9.89%respectively;The variation coefficients values of wheat canopy RVI at different growth stages of each zone,were between 4.4～17.8%,3.1～5.7%,5.6～9.5%respectively,which were less than their variation coefficient values range of 25.38%,9.61%and 16.52%in the whole area.This suggests that the partition achieved good effects.By running coupling model on each zone,the research improve the problem of large amount of calculation when inverse parameters in regional scale.This research uses the high space-time resolution images,which are fused by the high spatial resolution and high temporal resolution images,as the information fusion point of remote sensing and model.By building lookup table based on the coupling model between the management parameters(sowing date,sowing rate and nitrogen rate)and vegetation indexes.The research implemented the inversion of parameters which are difficult to accurately obtain on regional scale,and ultimately improved the WheatGrow simulation accuracy on regional scale.At the same time,measuring data which comes from different years and different ecological point are used to study the optimum coupling vegetation index and timing based on the integration of remote sensing and growth model.Results show that:(1)the 3 bands vegetation index RNir/(RRed+RGreen)or the 2 bands vegetation index SAVI(RNir,RRed)is the best fit coupling vegetation index;(2)if research only have one remote sensing image,heading oranthesis is the best stage of coupling;(3)if research can get more high space-time resolution images of different growth periods,it can achieve the highest prediction accuracy when the assimilation of 3-4 images in the late jointing stage to initial filling stage.Based on partition method,the research is better to simulate the time and space distribution of winter wheat growth and productivity index.