Heat as a Hazard to Human Health: A Multiple Dataset Assessment of Extreme Heat Indices Relevant to Human Health.

Abstract

Extremely elevated air temperatures are one of the most deadly types of weather. Future climate predictions along with predictions in urbanization and global population require research in this area to be relevant to both the climate science community and other scientific fields. This thesis looked at extreme heat within an urban region, through time over the continental United States and within three high-resolution gridded datasets. First trends were examined in extreme heat event characteristics between 1930 and 2010 using a unique form of the United States Historical Climate Network dataset. Using station data distributed across the United States it was found that the center of the continental United States has not increased and extreme heat event trends are sensitive to which daily temperature extreme they are based on. Future work investigating the differences in the trends as a function of timing within summer would be useful. Subsequently three gridded observational climate datasets of high resolution were compared to the aforementioned USHCN dataset because they had not previously been rigorously compared to a high quality climate dataset. The differences were large and highly variable at small spatial scales, which implies downscaled products using these datasets must consider how this translates as an uncertainty in those produces. The differences also exist at large spatial scales, and thus future work should focus on how to homogenize the underlying data network without reducing the number of stations included. Lastly the spatial variability in summer temperatures was explored throughout a metropolitan region using a network of observing stations. The amount of variability during the daily minimum temperature was largest and linked to city-wide weather variables within a reanalysis dataset, which suggests future studies might use forecast models to predict the amount of spatial variability. This framework of linking temperature observations to land attributes was demonstrated to work well and might be applied elsewhere.