墒情诊断模型的评价

发布时间：2018-05-31 14:48

本文选题：墒情监测 + 土壤含水量　；参考：《生态学杂志》2017年12期

【摘要】：对6个模型进行系统评价的目的是为模型优选提供理论、方法和参数依据。采用优选模型比例、验证方法指标、验证方式指标和异常值指标4项指标对6个独立墒情诊断模型进行了综合评价。结果是:模型优先顺序为差减统计模型、间隔天数统计模型移动统计模型比值统计模型统计模型平衡模型;逐日模型好于时段模型。合格率高的模型包括4个特征,即自变量数量不超过3个、自变量相对独立、模型能处理不固定监测天数、人为不决定参数。按监测点建模的6个独立模型都可以单独使用,解决了模型不通用的问题;可以实现逐日诊断和预测,方便与遥感信息、作物长势信息等实时匹配;差减统计模型和间隔天数统计模型精度最高,前者是基于质量守恒定律的统计模型,后者有效地解决了间隔天数不固定所带来的预测误差。
[Abstract]:The purpose of systematic evaluation of six models is to provide theory, method and parameter basis for model selection. Six independent soil moisture diagnosis models were evaluated synthetically by four indexes: optimal model proportion, verification method index, verification method index and outlier value index. The results are as follows: the priority of the model is the difference subtraction statistical model, the interval statistical model is the moving statistical model, the ratio statistical model is balanced, and the daily model is better than the time period model. The model with high qualified rate includes four characteristics, that is, the number of independent variables is not more than 3, independent variables are relatively independent, the model can handle the indefinite monitoring days, and the parameters are not determined artificially. The six independent models modeled according to the monitoring points can be used separately to solve the problem that the models are not universal and can be diagnosed and predicted day by day, which is convenient for real-time matching with remote sensing information and crop growth information. The difference subtraction statistical model and the interval days statistical model have the highest accuracy. The former is a statistical model based on the law of conservation of quality, and the latter effectively solves the prediction error caused by the uncertainty of the interval days.
【作者单位】：农业部环境保护科研监测所;北部湾环境演变与资源利用教育部重点实验室(广西师范学院)广西地表过程与智能模拟重点实验室(广西师范学院);
【基金】：“中国农业科学院科技创新工程”(2016-cxgc-hyl) 广西科技开发项目(14125008-2-24) 天津市科技支撑计划(15ZCZDNC00700) 全国农业技术推广中心节水处项目(2016-hx-hyl-5)资助
【分类号】：S152.7

【相似文献】