UC Berkeley

Catchments integrate incoming hydrological and geochemical fluxes via the mixing and reaction processes occurring within their boundaries. The catchment science community still seeks realistic and internally consistent models which explain integrated catchment behavior. It is known that the amount of streamflow responds quickly to rainfall, that stream water is predominantly "old" water which has been stored for long periods within the catchment, and that streamflow chemistry varies with flow regime. To quantify these observed patterns of catchment behavior, I examined precipitation-runoff relationships, the distribution of travel times of water parcels falling on the catchment, and reactive tracer concentration-discharge relationships.

By examining water and solute fluxes across a wide range of catchments, I identified characteristics which catchment models should include in order to accurately represent these systems. Specifically, I showed that long-tailed travel time distributions are common in at least 20 European and North American catchments, and I tested the methods that are commonly used to estimate such distributions. I demonstrated that concentration-discharge relationships consistently exhibit power-law scaling at timescales of rainstorms and on an interannual basis in 59 hydrochemically diverse catchments. Finally, I showed that in snow-dominated catchments in California's Sierra Nevada mountains, summer low flows often respond more-than-proportionally to changes in snowpack volume.

I also showed that, although simple exponential mixing models are commonly used to represent catchment behavior, they cannot reproduce observed long-tail travel time distributions or power-law concentration-discharge relationships. Instead I showed that catchment travel time distributions are well-characterized by gamma distributions, and that a new reaction and transport model matches concentration-discharge relationships across our study catchments. Finally, I examined the effects of changing the phase of precipitation in a historical snowpack-low flow model. As more precipitation fell as rain instead of snow, the historical model was less accurate. I considered a mechanistic model to explore potential catchment responses to climate warming in Sierra Nevada catchments, and found that changes in the processes of groundwater recharge and evapotranspiration that were not considered in the historical model affected low flow responses to perturbed catchment inputs.