Identifier

etd-04052013-150210

Degree

Doctor of Philosophy (PhD)

Department

Geography and Anthropology

Document Type

Dissertation

Abstract

Wind storms cause significant damage and economic loss and are a major recurring threat in many countries. Maximum sustained and peak gust weather station data from multiple historic wind storms occurring over more than three decades across Europe were analyzed to identify storm tracks, intensities, and areas of frequent high wind speeds. Wind surfaces for maximum sustained and peak gust winds were estimated based on an anisotropic (directionally-dependent) kriging interpolation methodology. Overall, wind speed magnitudes and high intensity locations were identified accurately for each storm. Directional trends and wind swaths were also consistently located in appropriate locations based on known storm tracks. Anisotropic kriging proved to be superior to isotropic (non-directional) kriging when modeling continental-scale wind storms because of the identification of strong directional correlations across space. Results suggest that coastal areas and mountainous areas experience the highest wind intensities during wind storms. These same areas also experience high variability over short distances and thus the highest error measurements associated with concurrent interpolated surfaces. For this reason, various covariates were utilized in conjunction with the cokriging interpolation technique and improved the interpolated wind surfaces for five wind storms that impacted both the mountainous and topographically-varied Alps region and the coastal regions of Europe. Land cover alone reduced station-measured standard error most significantly in a majority of the models, while aspect and elevation (singularly and collectively) also reduced station standard error in most models as compared to the original kriging models. Additional comparisons between different areal scales of kriging/cokriging models revealed that some surface wind variability is muted at the continental scale, but identifiable at the local scale. However, major patterns and trends are more difficult to ascertain for local-scale surfaces when compared to continental-scale surfaces. Large station error can be reduced through local kriging/cokriging, but additional research is needed to merge local-scale semivariograms with continental-scale models. Results showed substantial improvements in wind speed surface estimates over previous estimates and have major implications for catastrophe modeling companies, insurance needs, and construction standards. Implications of this research may be transferrable to other geographies and create an impetus for database and covariate improvement.

Date

2013

Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Rohli, Robert

Share

COinS