| Wheat is one of the three major food crops in the world.Whether its yield is high or low is directly related to national food security.Timely prediction of its yield has great significance to wheat yield management,government grain policy formulation,market regulation and even foreign trade.Based on the analysis of the current situation and development trend of the crop yield estimation methods,and the advantages of statistical data,meteorological data and remote sensing data in yield estimation were fully considered.A coupled yield estimation model based on the trend model,climate index and EVI in critical period was proposed.Trend model reflects the yield changes caused by variety improvement,field management and related policies.Climate index expresses the impacts of precipitation,light and temperature on winter wheat yield in meteorology.Enhanced vegetation index(EVI)expresses the comprehensive impacts of pests,diseases and human factors on winter wheat yield.Then a coupled model of trend,climate and remote sensing is built by integrating the three factors.The yield of winter wheat was estimated on scale.In this study,11 winter wheat planting counties in Hebei Province were selected as study areas.Coupled yield estimation models of Winter Wheat during the whole growth period were constructed using county-level statistical yield data from 2001 to 2016,meteorological data of corresponding years and remote sensing data.The yield of Winter Wheat in 2017 was estimated and validated.Based on the coupled model of whole growth period,two different real-time yield estimation schemes were designed for different winter wheat growth period:1)The key period historical data replaced schemes for real-time yield estimation:before the maturity of winter wheat,without the data of whole growth period,based on the existing meteorological data,remote sensing data and historical meteorological data,remote sensing data structure.New data of whole growth period are generated,and a real-time yield estimation model is constructed by using the method of coupling model construction.2)Real-time yield estimation scheme of optimal impact factors per growth period:before the maturity of winter wheat,with the growth and development of winter wheat,the available meteorological data and remote sensing data gradually increase,and real-time yield estimation model is constructed by using existing data and newly acquired data.Firstly,the methods of establishing climate index in the coupled model are compared,and the most suitable method of establishing climate index in the study area is selected.Secondly,the accuracy of the coupled model of trend model,climate index and remote sensing data of key period is evaluated,and the goodness of fit of the coupled model of trend model,trend and climate index model,and trend,climate index and remote sensing model are compared,and the winter wheat yield in 2017 is estimated.Finally,for two different real-time yield estimation schemes,the coupling models of growth-by-growth period and fixed critical growth period were established for real-time prediction of winter wheat yield in 2017.The yield estimation results of the coupled model of the whole growth period of winter wheat,the prediction results of the real-time yield estimation schemes of the optimum influencing factors per growth period and the prediction results of the key period historical data replacing the real-time yield estimation schemes were compared and analyzed.The conclusions are as follows:1)For the method of establishing climate index,considering the correlation between climatic factors and the direct effect of each factor on yield,the method of constructing weighted summation climate index has achieved good results,which is superior to the product climate index.The correlation between the weighted sum climatic index and yield showed a steady growth trend from sowing date to maturity,which accorded with the rule of climate influence on Winter Wheat during its growth period.2)The yield estimation accuracy of the coupled model based on the whole growth period trend,climate index and remote sensing data in 10 research counties ranged from 95.8%to 99.6%.Comparing the applicability of the model in different areas,the results show that the coupled model is superior to the trend-based model and the climate index-based model in the areas where land resources are limited or pests and diseases occur.In the areas affected by precipitation and temperature,the accuracy of the coupled model is similar to that of the trend-based climate model,but it is superior to that of the trend-based model.In the region where irrigation is more frequent,the accuracy of the three models is not very different.3)Two real-time production estimation schemes are suitable for different regions.For arid regions,the real-time yield estimation scheme based on optimal impact factor per growth period is better than that based on critical period historical data instead of real-time yield estimation scheme;for irrigated or saline-alkali regions,the real-time yield estimation scheme based on critical period historical data is better. |
