中国农业气象

• 论文 • 上一篇    下一篇

基于GLUE方法的植被界面过程模型(VIP)的不确定性分析

王昆;莫兴国;林忠辉;舒畅;   

  1. 中国科学院地理科学与资源研究所陆地水循环及地表过程重点实验室;中国科学院研究生院;
  • 出版日期:2010-08-10 发布日期:2010-08-10
  • 基金资助:
    国家“973”计划项目(2010CB428404);; 国家“863”计划项目(2006AA10Z228);; 科技部国际合作项目(0911)

Uncertainty Analysis of Vegetation Interface Processes(VIP) Model Based on GLUE

WANG Kun1,2,MO Xing-guo1,LIN Zhong-hui1,SHU Chang1,2(1.Key Laboratory of Water Cycle and Related Land Surface Processes,IGSNRR,Chinese Academy of Sciences,Beijing 100101,China;2.Graduate School of Chinese Academy of Sciences,Beijing 100039)   

  • Online:2010-08-10 Published:2010-08-10

摘要: 生态系统模型一般参数较多,且在应用时存在时空尺度问题,易产生不确定性。通过模型不确定性分析,可以加深对模型结构的理解,提高模型预报的可靠性。植被界面过程模型(VIP)是一个综合考虑了陆地生态系统能量收支、水文循环和碳氮等生命元素吸收转化等过程的生态/水文动力学模型。本文采用GLUE(General-ized Likelihood Uncertainty Estimation)方法,以拟合度系数作为似然判据,利用华北平原冬小麦生长季内的田间观测数据分析VIP模型中的作物生长、土壤水分运动以及光合速率模块中8个参数以及模型预报的不确定性。研究表明,最大光合速率Vmax、饱和含水量wcsat、田间持水量wcfield参数为敏感性参数,其对似然判据的影响大,其余参数是相对不敏感参数。在置信度为95%水平下,发现观测值大都接近或者包含在置信预报区域内,说明可以通过参数校准得到很好的模型模拟效果。

关键词: GLUE方法, VIP模型, 预报不确定性, 敏感性参数

Abstract: There are many parameters in the ecosystem model.The variability of parameters at different space-time scales resulted in uncertainties.The uncertainty analysis of parameters could help to understand the structure of the model deeply and improve the reliability of the model predictions.The Vegetation Interface Processes(VIP) model is an ecohydrology dynamic model,which includs energy budget,hydrology cycle,absorption and transformation of carbon and nitrogen in the terrestrial ecosystem.In this paper,the generalized likelihood uncertainty estimation(GLUE) methodology was used to analyze the uncertainty of the parameters.In the VIP model,we chose eight parameters,which came from the crop growth module,soil water dynamic process and photosynthesis module.An index of agreement was chosen to be the likelihood weight.The field data(LAI,biomass,soil water content etc.) in the North China Plain were used.The results showed that maximum catalytic activity of Rubisco,saturated water content and field capacity were sensitive parameters,which influenced the value of likelihood weight greatly.The others were non-sensitive parameters.Almost all the observations approached were included in the confidence interval with the 95% confidence level,which indicated that better simulations could be got by calibrating the model parameters.

Key words: GLUE methodology, GLUE methodology, VIP model, Prediction uncertainty, Sensitive parameters