科研之友
微信
新浪微博
Facebook
分享链接
摘要
Multiple variables from simulated climate fields are widely used in hydrological and ecological models and climate impact assessments, yet the importance of multivariate climate relationships is not widely recognised. This study evaluates climatic outputs from the Canadian Coupled Global Climate Model (CGCM3) in the southern boreal forests of western Canada, by comparing the simulated multivariate relationships with those observed at three representative forest sites. Monthly mean data for five near-surface climate variables (net radiation RN, air temperature TA, relative humidity RH, wind speed WS and surface pressure P) are analysed and compared using visual inspection, hypothesis testing and principal component analysis. The projections of the 1st and 2nd principal components, which explain about 75% of the variation in the data, show remarkable similarities in the observations from the three forest sites (with some subtle differences between the evergreen and deciduous plant functional types), but some broad differences between the observations and model outputs. The model reproduces the observed relationships among RN, TA and P, but not between RH or WS and the other variables. In particular, RH is strongly and negatively related to TA and RN in the forest observations but independent in the model outputs; RH is negatively related to WS in the observations but positively related in the model output; and P is uncoupled from the other variables in the observations but negatively related to RH and WS in the model output. The broad scope of the differences indicates a divergence of process representation at large time and space scales. We explore possible reasons for the observed discrepancies, which indicate some limitations in using climate model outputs directly to drive hydrological models.
- 出版日期2014-11-27
- 单位Saskatchewan; Saskatoon
