Session Information
| Session | Poster Session | | Date | Monday (2008-04-07) | | Time | 5:00 PM - 7:00 PM | | Room | Grand Terrace |
Presentation Information
| Presenter | Wei Wu | | Title | Comparing two spatial models in deriving the probability of cloud cover in north-eastern Puerto Rico | | Affiliation | University of Southern Mississippi | | Authors | Wei Wu, Lianjun Zhang, Charles Hall | | Keywords | AIC, Cloud cover, GLMM, GWR, Moran | | Presentation Type | Poster | Abstract:
Understanding the spatial patterns of cloud cover is a key to understand solar radiation incident on the earth surface, which ultimately drives all ecological processes. In this paper, we compared two spatial models: generalized linear mixed model (GLMM) and geographically weighted regression (GWR) logistic model in deriving the probability of orographic cloud cover at the Luquillo Experimental Forest in north-eastern Puerto Rico on relatively clear (March 27 of 2000 and July 20 of 2001) and cloudy (January 9 of 2001 and March 4 of 2003) days. In the models, the dependent variable is binomial: 0 representing non-cloudy areas and 1 representing cloudy areas, which were derived from MODIS images. The independent variables include the difference between elevation and lifting condensation level, aspect and slope. The model results indicated that the GWR model is a consistently better fit model compared to the generalized linear mixed model based on significantly lower Akaike Information Criteria (AIC), and Bayesian Information Criteria (BIC) in the GWR model. In addition, Moran’s I statistics in the residuals from the GWR reduced to 0.13 compared to 0.28 from the GLMM, suggesting that there existed smaller spatial autocorrelation in the residuals from the GWR, mainly because that spatial heterogeneity of cloud cover was accounted for in the GWR, but not in the GLMM. |
|