A Mixed-Method Approach to Detecting and Attributing Changes in Streamflow
Abstract
Changes in streamflow can be caused by climatic change (e.g. temperature, precipitation) or by development and the associated land use change or infrastructure construction. Disentangling the combined effects of climatic and urbanization changes can be challenging in part because these variables interact non-linearly. For example, catchment properties altered by urbanization influence the relationship between watershed inputs (e.g. precipitation, temperature) and streamflow and groundwater response. Additionally, streamflow time series can include seasonally varying components, gradual changes, and abrupt changes. Decomposing a changing streamflow time series into these components is a key step in attributing the causes of change, as different drivers of change alter hydrological signatures in distinct ways. However, this decomposition is challenging because multiple decompositions may fit the data reasonably well and the statistically optimal decomposition is not guaranteed to be correct. Here we apply a Bayesian approach to identifying changes in both the trend and seasonal component of streamflow. We then use Bayesian model averaging, as it can approximate complex time series well and avoids the hazard of selecting a single best model. An additional output is the probability of change in the trend and seasonal component at each time step. We then pair these outputs with a detailed case history documenting land use, infrastructure, and policy changes that may alter hydrological properties and processes. Pairing the probability of change estimates and case history allows us to evaluate case specific hypotheses of change with hydrological knowledge and other relevant data sets (e.g. precipitation, land use). To demonstrate this method, we apply it to the case of Salt River (and its tributary the Verde River), which is a water source for metropolitan Phoenix, AZ. This case serves to identify benefits, weaknesses and opportunities for improvement.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMH113.0001G
- Keywords:
-
- 1803 Anthropogenic effects;
- HYDROLOGY;
- 1833 Hydroclimatology;
- HYDROLOGY;
- 1834 Human impacts;
- HYDROLOGY;
- 1878 Water/energy interactions;
- HYDROLOGY