基于遗传程序设计方法的内陆湿地年蓄水量变化预测研究--以卧龙湖湿地为例

Genetic programming based method for predicting the inland-wetland water conservation capacity——A case study on Wolonghu Wetland

在湿地蓄水量概念及其内涵剖析的基础上,提出了基于遗传程序设计方法的半干旱内陆湿地年均蓄水量变化预测模型。以辽宁省最大的天然内陆湿地卧龙湖为例,通过对湿地系统的分析和参数确定,利用该区连续11年的气象和水文数据,尝试建立了该湿地蓄水量年际变化的预测模型,并以2004—2006连续3年的蓄水量实测数据对模型进行验证。结果表明,模型的平均拟合误差为8.2%,预测结果的相对误差分别为8.5%、9.6%和10.3%,符合湿地生态系统实际管理的需求。该方法对于实施湿地生态保护工程和水资源合理调配具有一定的参考价值。

Based on the analysis of the concept and its connotation of water reservation amount in wetland,a new method is proposed for modeling the change of WA in wetland by using the so-called genetic programming（GP） approach.Taking the Wolonghu Wetland,the biggest inland wetland in Liaoning Province,as a case study,the authors set up the prediction model of its annual water reservation amount by using 14 successive years of meteorologic and hydrological data scenarios.The precipitation and evaporation data have also been used as the input variable and WAA data as the output variable after systematic analysis and determination of parameters.In order to guarantee the precision of the GP model,we have divided the data scenarios of meteorology and hydrology into two categories,respectively.In the model validation phase,it is recommended to use the first category of 11 successive annual meteorological and hydrological data scenarios（from 1993 to 2003） for the model establishment,while the other category of 3 successive annual data（from 2004 to 2006） can be used for the model testing.The results of our prediction show that the average relative error during the validation phase is about 8.2% and during the testing phase,the relative errors are 8.5%,9.6% and 10.3%,respectively,which can be said to come to the need of the management of the wetland protection.Therefore,the method can be said of great significance for managing wetland ecosystem protection projects and rational allocation of the regional water resources.