Establishment And Application Of Soil Moisture Diagnosis Model Based On Precipitation

The government has always attached great importance to the soil moisture monitoring,diagnosis and forecast,and the Ministry of Water Resources,Ministry of Agriculture and Meteorological Bureau established soil moisture monitoring and forecasting system successively,which has played a very important role in guiding the real-time agricultural production.The seventeenth of 100 projects in “the 13 th Five-Year Plan” is to ensure the completion of high standard farmland 800 million acres,even reach to 1 billion acres(irrigation is the basic guarantee).And the nineteenth is to increase the efficient water-saving irrigation area of 100 million acres(The soil moisture monitoring,diagnosis and prediction is the scientific basis for reasonable irrigation).Both guidelines have direct and important relationships with soil moisture.The internet,the internet of things,big data,cloud computing and so on not only provide advanced technology and guarantee for the platform of the soil moisture monitoring,diagnosis and forecast but also make the platform modern,informational and real-time.A difficult problem is,however,the theory and method of soil moisture diagnosis and forecast are not mature.Therefore,there is an urgent need to develop innovative theory,method,technical system and service mode for soil moisture diagnosis and forecast.In this study,based on the present situation of soil moisture forecast model,we conducted a systematic study and established 6 independent diagnostic models of soil moisture and an integrated diagnostic model using the data from 2012-2014(Some of the monitoring point data from the beginning of 2010 or 2011).The six models are built based on balance method,statistical method,subtractive statistics method,ratio statistic method,interval days statistics method and movable statistics method,respectively.And the integrated model is composed by the six methods to diagnose the soil moisture.These models are validated using soil moisture monitoring data in 2015.Based on the long-term observation data,we established the quantitative relation between the local precipitation and soil moisture,and these prediction models can predict soil water content accurately and provide diagnosis and forecasting services for soil moisture,achieving the goal of real-time monitoring of soil moisture,.The results showed that:(1)Seven soil moisture monitoring diagnosis models,established based on the data of nearby weather stations rainfall,can diagnose the soil moisture at interval and daily.According to the prediction results,the daily model is better than the time interval model.(2)Considering the comprehensive diagnosis process and avoiding the abnormal value cases,the accuracy of six diagnosis models is as follows,the subtractive statistics method,the interval days statistics method> the movable statistics method> the ratio statistic method>the statistical method> the balance method.The accuracy of integrated diagnostic model is significantly higher than other models.(3)In the process of modeling and parameter optimization,every monitoring point is separate operation,resulting in specific models and parameter.These parameters implied the environmental impact factors of the monitoring such as climate,geomorphology,soil texture,crops and farming methods and so on.As a result,six diagnosis models and the integrated model have universal suitability in seven different provinces.(4)From the view of mathematics,the models with high accuracy in this paper should have four conditions at least.First,the variables were relatively independent and not more than three.Second,the method should follow the law of conservation of mass and statistical method.Third,the method of unfixed monitoring date should be handled.Last,the parameters should not be determined artificially during the data processing.(5)The annual precipitation affects the precision of the model,and the less the annual precipitation,the higher the accuracy of the model will be.Soil moisture monitoring diagnosis models,established based on the data of nearby weather stations rainfall,are scientific and practical.The prediction accuracy of 6 independent models needs to be improved,and the integrated model effectively avoids the shortcomings of the independent model,and improves the prediction accuracy to some extent.The soil moisture models provide a theoretical basis and model methods for the establishment of soil moisture monitoring network and real-time dynamic soil moisture prediction,which have a broad application prospect.