| Summary: | <p>Some of the perennial questions in hydrology relate to how land cover change and precipitation affect streamflow change. These questions have been addressed in many ways over the years, but some common assumptions remain insufficiently challenged or addressed. The work presented in this thesis seeks to address some of these gaps using a large sample of observed hydrological, climatological, and land surface data. The increased availability of large-sample hydrological data in recent years and growing interest in the use of statistical models for causal inference with observed data have made it possible to conduct research which sheds light on patterns and drivers of hydrological non-stationarity, a topic that is a matter of increasing importance for the community.</p>
<p>I take advantage of large-sample hydrological methods and data availability to produce novel insights into hydrologic sensitivity and elasticity, defined throughout as the expected change in streamflow associated with a one percent change in another variable. I address the following interrelated questions: 1. What is the effect of tree cover change and urbanisation on streamflow in the United States? 2. How robust are single-site regression models relative to causally interpretable panel regression models in this space? 3. How does streamflow elasticity to precipitation vary spatially and across the streamflow distribution? and 4. How do climate and hydrological behaviour co-evolve in the context of streamflow elasticity?</p>
<p>In this thesis, I find small but statistically significant effects of urbanisation on median and high streamflow across the U.S. The effects of tree cover change on flow and urbanisation on low flows are not statistically significant. I interrogate the relationship between streamflow and precipitation across many segments of the flow distribution, and create a new approach for investigating these relationships. Termed “elasticity curves”, this concept visually represents the elasticity of streamflow to average annual and seasonal precipitation across the entirety of the flow distribution. I cluster the curve according to their normalized shape and the resulting groups correspond, to some extent, with hydrologic signatures and catchment characteristics, including the baseflow index, slope of the flow duration curve, and the aridity index, among others. I posit that the shape of the curve corresponds to water storage capacity within a catchment. Thus, the normalized and clustered curve shapes might be used as a tool for understanding hydrologic behaviour relative to catchment storage. This work represents one of very few studies to investigate elasticity across different segments of the flow distribution simultaneously and offers new insights into hydrologic response to climate.</p>
<p>Finally, I investigate temporal variation in elasticity to precipitation for low, median, and high streamflow at a regional scale. This work shows high interannual and spatial variability with mean absolute year-to-year differences as high as 0.5 in some regions, relative to long-term averages typically ranging from about 1-2.5. Long term trends in regional-scale interannual elasticity were uncommon but present in some regions. Total absolute changes in elasticity based on Mann-Kendall trend test results in regions with significant trends range from 0.28 to 0.60 over the study period. The time period used to estimate elasticity may have an effect on the resultant value, because large year-to-year variability exists in some regions.</p>
<p>The results of this work have implications for hydrologic management and fundamental process understanding of hydrologic behaviour.</p>
|
|---|