Data, Scale, and Change: A Study of Hydrologic Dynamics in the Colorado River Basin

Abstract

Hydrology is increasingly reliant on large-scale and long-term data to understand past, present, and future water system change. As climate change intensifies pressure on water resources globally, challenges remain in detecting, quantifying, and interpreting surface water dynamics across time and space. For example, the Colorado River Basin (CRB) is facing unprecedented drought conditions. While many studies document patterns of change such as declining streamflow, earlier snowmelt, and reservoir depletion, less attention has focused on how spatial and temporal depth shape the hydrologic insights we derive. This dissertation explores three complementary perspectives on hydrologic change in the CRB, examining how different datasets and scales influence our understanding of water availability and variability. The first study investigates temporal predictability using tree-ring reconstructed streamflow across multiple Upper Basin gauges, extending beyond the commonly analyzed, Lees Ferry gauge. By applying wavelet filtering and nonlinear methods, results reveal that streamflow predictability varies across decadal and multidecadal time scales. Moreover, predictability windows can be misaligned even in gauges with highly correlated flow. Notably, increasing variance in recent decades suggests a possible transition into a lower predictability regime, potentially linked to warming induced hydrologic shifts. The second study quantifies long-term surface water change using nearly four decades of Landsat derived inundation maps (1984 – 2021). Moving beyond point observations, it analyzes spatial patterns of permanent and seasonal water across the Basin. Results show a net loss of 10% in total inundated area, with two-thirds of losses occurring outside major reservoirs (Lake Powell and Lake Mead). Declines in permanent water in the lower basin while expanding seasonal water in headwater regions. This highlights the basin wide drying trends that extend far beyond monitored locations. The third study evaluates the influence of spatial resolution on surface water detection by comparing 10-meter Sentinel-derived and 30-meter Landsat-derived surface water mapping products. Results show that Sentinel identifies more than twice as many low-order stream segments as Landsat, with up to five times more segments particularly in headwater regions. These differences are more noticeable in ephemeral, narrow streams, or small water bodies. This improved detection capacity highlights the importance of high resolution data for capturing fine scale hydrologic features that are often missed in coarser imagery, especially in arid and semi-arid environments like the CRB. Together these findings demonstrate that our understanding of hydrological change is fundamentally shaped by the data, tools, scales, and temporal periods we choose. This work highlights seasonal water changes as major drivers of water availability shifts in the CRB, shows the values of extended temporal records for understanding predictability transitions, and emphasizes the importance of multi-resolution approaches for comprehensive surface water monitoring. These contributions provide new insights for hydrologic monitoring in water stressed regions and emphasize the critical need for multi-scale perspectives in understanding and managing water resources in an era of accelerating change.