Research On Apple Frost Loss Assessment Based On High-temporal And Spatial Remote Sensing And Flowering Phenology Information

China has become the world’s largest producer and consumer of apples.With the continuous increase in the scale and output of China’s apples,the impact on the global apple supply pattern is increasing.However,in recent years,frost disasters have occurred frequently during the flowering period of apples in my country,causing serious production reductions and economic losses in some major apple-producing counties and cities.In response to my country’s major needs for strengthening agricultural information monitoring,early warning and release to improve the level of comprehensive agricultural information services.This paper takes regional-scale apple orchards as the research object,based on high-temporal and spatial remote sensing,meteorological data and field survey data,and focuses on key issues such as the frequent occurrence of frost disasters during the flowering period faced by the apple industry,the assessment of apple frost loss based on high-temporal and spatial remote sensing and flowering phenology information was carried out.According to the research content of this paper,the regional apple distribution extraction,planting year identification,initial flowering time prediction,and flowering period frost loss assessment research were carried out in sequence.The main conclusions are as follows:(1)This study proposes a method of combining"temporal and spatial fusion data"and"vegetation phenological characteristics"to obtain the distribution of apples in Qixia.The results show that compared with only using spatio and temporal fusion data for classification,adding vegetation phenology features can effectively improve the overall accuracy(OA)(OA:98.14%;Kappa coefficient:0.97).In addition,this study evaluated the performance of the enhanced spatial and temporal adaptive reflectance fusion model(ESTARFM)from the temporal and spatial dimensions.In terms of time,the accuracy of the fused images simulated by the algorithm during the summer and autumn seasons is relatively high,its coefficient of determination(R~2),root mean square error(RMSE)and normalized root mean square error(NRMSE)all reached more than 0.75,0.07 and 12.7%,respectively;spatially,the algorithm has high simulation accuracy for uniform distribution areas(R~2=0.74;RMSE=0.08;NRMSE=13.76%).The apple distribution results obtained in this study are of great significance for the fruit industry upgrade and spatial layout optimization in the service area.(2)According to the distribution of apples in Qixia as the area of interest to identify the planting year,this study combined the growth characteristic periods that can distinguish apple types and the euclidean distance(ED)algorithm to obtain the spatial distribution of apples from 2000 to 2017.Then use the temporal and spatial distribution characteristics of apples to identify the planting year information.In addition,this study evaluated the impact of images with different spatial resolutions on the accuracy of identifying the planting year.Compared with the use of Landsat data from 2000 to 2017,the accuracy of the apple planting year identification after using the 2017 Sentinel-2 data instead of the Landsat data for the same period is relatively better,R~2 increased by 0.019,and RMSE and NRMSE decreased by 0.11 years and 0.63%,respectively.Therefore,the use of remote sensing images with higher spatial resolution can effectively improve the recognition accuracy of apple planting years.The regional apple planting year identification method in this study can effectively support the formulation of plans for aging orchard transformation and variety renewal.(3)In order to effectively predict the phenological information of regional apples at the initial flowering stage,this study proposes a calibration method for the Sequential model that uses spatial sampling instead of long-term sampling to optimize the chill-heat requirements parameters of the model.Among them,Luochuan has 30 chill portion(CP)and 4900 growing degree hours(GDH)respectively,and Linyi has 57 CP and 7000 GDH respectively.Then,according to the Sequential model optimized by the data in 2020 and combined with the gridded temperature data in 2021,the regional apple initial flowering time is predicted,and the prediction accuracy is verified by the measured initial flowering data in 2021(RMSE≤4.7 days;NRMSE≤5.19%).The results show that compared with the Sequential model or the Chill overlap model calibrated by previous scholars,this study has obtained similar or better prediction accuracy.In addition,the simulation results of the initial flowering time obtained from the initial range of chill-heat requirements of the sample points found that when the chill requirement is low,the chill is the main factor that advances or delays the apple initial flowering time;and as the chill requirement increases further,the influence of heat on initial flowering time continues to increase.The method for predicting the initial flowering time of regional apples in this study is helpful to improve the frost disaster prevention mechanism during the flowering period of apples and the planning of suitable planting areas.(4)Based on the research of apple planting year identification and initial flowering time prediction,this study uses regional apple remote sensing information(planting year)and flowering phenology information(flowering time and chill accumulation before frost,as well as the minimum temperature and temperature difference on the day of frost),a method for evaluating frost loss during the flowering period of regional apples is proposed.Then,according to the results of leave-one-out cross-validation(LOOCV),the model has better simulation performance of frost loss during the regional flowering period(R~2=0.69;RMSE=18.76%;NRMSE=18.76%).In addition,the extended fourier amplitude sensitivity test(EFAST)method is used to evaluate the sensitivity of the simulation results to the modeling parameters.The results show that the minimum temperature on frost days,chill accumulation before frost and planting year are the most sensitive to frost loss during apple flowering.The sum of the first-order and total-order sensitivity index contributions of these three parameters accounted for more than 82%of the total,respectively.This study can effectively assist the government and enterprises in the implementation of apple market planning and agricultural insurance.In this paper,based on regional apple distribution extraction,planting year identification,and initial flowering period prediction research,to effectively carry out the post-frost loss assessment of apple flowering period based on high-temporal and spatial remote sensing and flowering phenology information,to improve the problem of abnormal market volatility caused by frequent frost disasters during the apple flowering period in my country,and promote the sustainable development of the apple industry.