A Model Research For Predicting Regional-Type Rice Yield In Hefei By Integrating Historical Meteorological Data

As one of the important grain production bases in China,Anhui Province plays a crucial role in ensuring grain crop output for food security in the province.In recent years,the frequent occurrence of complex weather conditions,especially extreme low and high temperatures,has had an adverse impact on rice production.Based on this,the study of meteorological factors and the relationship between Hefei and surrounding areas of rice growth development,make full use of the advantages of meteorological factors,for meteorological factors change suitability planting,and based on the weather factors,the yield of the rice harvest,to improve rice yield per mu has realistic value,for subsequent rice prices,interregional deployment and reserves have positive significance.This thesis mainly does the following work:(1)Determine the main areas and comparison areas of this thesis,and select the meteorological factors in the selected areas combined with the characteristics of the fertility period.Lujiang County is rich in climate factors and diverse rice varieties,which is a major rice planting county in Anhui Province.Based on the data obtained from practice in Meteorological Institute of Anhui Meteorological Bureau,Lujiang County of Hefei City is selected as the main research area and Feixi County and Shouxian County as the test and comparison area;temperature,precipitation and relative humidity are selected as the main influence factors to study the influence of meteorological factors on rice growth period.(2)Processed the selected meteorological data,and studied the rice yield prediction model.This thesis studies the influence of meteorological factors on rice yield,According to the specific meteorological data of the three counties,Based on the rice growth period and the combined selected three meteorological factors,15 meteorological factors of different reproductive periods were obtained;As the obtained rice yield data are multifactorial comprehensive statistical yield data,To exclude the effects due to other factors,The total rice yield was separated using the time regression equation,Meteorological output based on meteorological factors and time trend output affected by time factors;Applying the isolated meteorological yield data to the three models,Modeling by using meteorological data and meteorological yield,The predicted output value of rice based on the three models was IV obtained,respectively.(3)The three models selected were compared and analyzed to select the optimal model.Using the data of Feixi County as the test set,the Cat Boost yield prediction model,BP neural network yield prediction model and LSTM yield prediction model were established respectively.The research found that the accuracy of the three models was 64.2%,79.8%and 83.6%,respectively.The three models had the best accuracy LSTM effect on the test set,followed by the prediction accuracy of BP neural network.Based on this,the LSTM yield prediction model was selected as the model used in this research to predict the rice yield in Lujiang County.(4)The accuracy of the rice yield prediction model based on the LSTM was tested.The LSTM model was applied to the rice prediction in Lujiang County.Taking 2018 as the forecast year,the rice yield per unit area of the forecast year was 8722.7 tons,1003.5 tons different from the actual value,and the accuracy was 87%,which had a good prediction effect.In order to verify the accuracy of the model,Shouxian county with a small longitude difference but a large latitude difference was selected for comparative analysis,and the results show that the model still has good prediction results in Huaihe region of Anhui province.This thesis focuses on studying meteorological data related to temperature,precipitation,and relative humidity.Combined with the rice growth period,it constructs a rice yield prediction model based on LSTM to forecast rice yield in Hefei and its surrounding areas.This provides a method for regional rice yield prediction in Hefei and serves as a reference for local rice pricing and cross-regional allocation policy formulation.