| Physiological and ecological parameters of rice will change under heavy metal stress.Using rice phenologyto conduct segmented assimilation that extracted from remote sensing data to study on the different responses to heavy mental stress in different rice growth stage.The Harris algorithm was used to get the Remote Sensing-Crop Growth Model assimilation frame to time scale optimization,and then extract the optimal assimilation time for the assimilation.The optimal assimilation can guarantee the model simulation accuracy and improve the model operation efficiency.In this research,3 rice fields with different heavy metal stress levels in Zhuzhou,Hunan Province were selected as the study area.And 16 HJ-1A/B CCD data in 2013 and the field measured data were also selected for the research.Using the HJ-1A/B data to construct the EVI time-series curve.Then,smoothed the curve of EVI time-series by non-Symmetric Gauss function method,in order to explore the response relations between the change of rice phenology and EVI time-series.So the rice phenology can be extracted.The stress factor f was imbedded into the WOFOST model to monitor the stress level,which can represent the response mechanism of rice physiological and ecological parameters under heavy metal stress.Based on the above research,get the WOFOST model to be localization and regionalization.According to the rice growth key phenology,we could determine the rice growth stage by the DVS.And we could establish the RS-Crop Growth Model segmented assimilation framework,based on the PSO algorithm.Drive the WOFOST model to simulate WRT one by one pixel,through getting different f to different rice growth stages.In this research,the segmented assimilation of RS-Crop Growth Model was evaluated by the heavy metal stress factor f and the simulated WRT.The temporal scale of the segmented assimilation model was conducted.Based on Harris corner algorithm,the LAI curve was converted to grey values to detect dominant points.In order to verify the accuracy of the 4 detected dominant points,the points were linearly fitted.R2 value was 0.89 of the fitting curve and LAI curve,and R2 value was 0.78 of the fitting curve and LAI ratio curve.The fitting curve can explain main information of the original curve.We calculated LAI by 4 CCD data,corresponding to the 4 dominant points,and drive the RS-Crop Growth Model assimilate.Compared with the previous assimilation results,the assimilation frame’s time efficiency was greatly improved by the 4 CCD data.What’s more,the simulation accuracy of the WRT was above 95%.Results showthe phenology segmented assimilation methods can improve the accuracy of RS-Crop Growth Model.Based on the model,4 detected dominant points can improve the operation speed of RS-Crop Growth Model and guarantee the assimilation precision. |
