UC Davis

This dissertation is a three-part geographic study of the historical ecology and applied ecological restoration of Lower Putah Creek, California. It uses state-of-the-art tools in geographic information systems (GIS) to analyze historical and modern datasets. Findings are intended to inform riparian management, restoration planning, and restoration design. In Chapter 1, a century of geomorphic change is analyzed using flow frequency analysis and a suite of topographic metrics to characterize channel shape over time. Most notably, it uses height above river (HAR), which is the height (z) of an x-y location above the nearest channel’s thalweg or edge of low-flow (baseflow) river surface. Results show that the dams on Putah Creek reduced channel-forming peak flows by at least 82% in volume, and now maintain a perennial baseflow. Between 1905 and 2005, there was significant channel incision along the entire creek, a change from a U-shaped channel to a V-shaped channel, and a narrowing of the baseflow channel by half, although with significant reach-scale variation. These findings can inform channel-appropriate restoration designs or dam diversion strategies to manage geomorphic processes. In Chapter 2, land cover change is analyzed between 1937 (pre-dam) and 2009 (post-dam) using machine learning image classification tools and forest-based classification methods. Those techniques were used to identify the importance of three topographic variables in predicting land cover: HAR, bank slope, and distance-to-baseflow surface. Results support previous studies, which show that a regulated flow regime enables longer-term plant succession and establishment of woody vegetation. Analysis by topographic variables show that woody vegetation tracked downward channel incision, and that the riparian zone became compressed in a narrower channel with steeper banks. Bank slope, HAR, and distance-to-baseflow surface had equal importance in predicting land cover types in both 1937 and 2009 image classifications. Findings demonstrate that these methods can be used in other systems to characterize vegetation change. In Chapter 3, first, inundation modeling and machine learning image classification are used to establish that HAR is well-correlated to discharge and vegetation. Then, HAR is used in a random forest classification to predict land cover types, and the output classification is used to create HAR zones relevant for restoration planning. Next, 25 reaches of Lower Putah Creek are delineated based on relatively homogeneous geomorphic characteristics and each is then ranked according to the sum of two independent rankings: (1) the in-channel relative area of their combined core riparian and marginal riparian zones; and (2) the in-channel relative area of their combined aquatic and transition zones. While 18 of the 21.22 miles of creek analyzed qualify as “degraded,” lowering of just half of the transition zone to floodplain HAR level could double the riparian zone, indicating a significant opportunity to recover endangered riparian forest in the Sacramento Valley. Finally, the relative area of aquatic and transition zones were used to prescribe geomorphological restoration actions. Of the degraded sections, 13.64 miles qualify for floodplain lowering, 7.05 miles qualify for baseflow narrowing, and 3.21 miles qualify for both, indicating that radical geomorphological change is needed to maximize the riparian habitat potential of Lower Putah Creek. The HAR zones created in this study can be directly incorporated into existing terrain design tools for restoration on Lower Putah Creek, and these methods can be implemented in many other river systems using publicly-available GIS datasets.