Investigating the interplay of changing landscapes and climate on the distribution of water (the “hydrology”) across the Puget Sound.

PRISM, in partnership with the UW Climate Impacts Group (CIG) is conducting a series of studies investigating the interplay of changing landscapes and climate on the distribution of water (the "hydrology") across the Puget Sound. This work is based on the application of DHSVM (Distributed Hydrology Soil Vegetation Model), a so-called geospatially-explicit, process based model. This work follows on from the initial "Hydromet" work that was the first-generation climate/DHSVM application of PRISM. DHSVM was developed originally for application to forested mountain watersheds. In this first study of the series, DHSVM was extended to the urban environment (Cuo et al 2008).
Urbanization modifies hydrological processes by replacing vegetated land cover with impervious surfaces and by extending the natural drainage network to include artificial ponds, ditches and conduits laid on the ground and underground. Impervious surfaces reduce infiltration, generally resulting in increased surface runoff and reduced base flow. Artificial ditches and conduits alter runoff pathways and change stormwater drainage. In urbanizing catchments, surface flow may be diverted to artificial ponds or flood detention ponds built to reduce flood risks or in some cases for irrigation purposes.
The objectives of this work were 1) to document adaptations to DHSVM that allow the model to be applied to partially urban or urbanizing conditions; and 2) to evaluate performance of the model under these conditions. The study lays a foundation for land cover change studies in the broader Puget Sound basin, Washington where urbanization is largely mixed with other land cover (e.g., forest).
To account for urban hydrological processes, a land cover category "urban" was added to the model, and for pixels with this land cover type, a fraction of impervious surface area is also specified. The fraction of impervious area determines the amount of surface runoff generated on impervious surfaces, whereas for the fraction that is not impervious, DHSVM handles infiltration using the same parameterizations as for non-urban pixels. A second parameter called the fraction of water stored in flood detention was also added. These two parameters allow the model to mimic, in a qualitative way, the dominant processes in urban runoff, specifically surface runoff generation and detention storage The runoff diverted to detention storage is allowed to drain as a linear reservoir, and re-enters the channel system in the pixel from which it is diverted. Surface runoff that is not diverted is assumed to enter the channel system directly. While it is assumed that the natural channel system remains intact, impervious surface runoff (and drainage from detention reservoirs) is assumed to be connected to the nearest stream channel directly so that overland flow reaches a stream channel at the same time step. Once impervious surface runoff has entered a stream channel, it follows the channel flow routing processes.
For testing purposes, the USGS gage 12113346 (Springbrook Creek near Orillia, southwest King County) was chosen, because it is a partially urbanized basin, is relatively small, and hourly precipitation data from a nearby climate station are readily available (Fig. 1). A thirty meter resolution USGS DEM was used as a base map for the watershed. Soil parameters from nearby relatively unurbanized catchments (Westrick et al, 2002), with land cover classifications for the basin from Alberti et al. (2004). Hourly precipitation measured at the Seattle-Tacoma International Airport was used to drive the hydrology model. Hourly temperature, relative humidity, incoming shortwave and longwave radiation are not routinely recorded, instead, they were generated from one sixteenth degree gridded daily precipitation, minimum and maximum temperature datasets. Hourly stream discharge (measured by USGS) between 1 January 1996 and 31 December 1999 at gage 12113346 was used for calibration, and from 1 January 2000 to 30 September 2002 was used for validation.
Progressive inclusion of urban structure improved model behavior (Fig. 2). Without the urban parameterization, storm peaks are underpredicted, and baseflow overpredicted, relative to observations (a1, a2). Inclusion of the fraction of impervious area, absent the detention parameterization, showed that peak flows are largely overpredicted, presumably because detention storage is not accounted for (b1, b2). Finally, including the fraction of runoff generated from impervious areas that goes to detention storage, and a detention recession coefficient to provide a plausible match with observations (c1, c2).
With these processes included, the model was calibrated and validated (Fig 3).

Following model testing, the overall effects of urbanization were investigated (Fig 4). Results show that that urbanization increased both seasonal and annual streamflows substantially. The largest increases occurred in the winter and the smallest increases were in the summer. Application of the model to a partially urbanized catchment in the Puget Sound drainage showed that using physically realistic values of impervious fraction, fraction of impervious runoff routed to detention, and detention drainage parameters resulted in good matches with observed flows. Simulations without the effects of detention storage greatly overestimated observed peak flows.

Cuo, L., D.P. Lettenmaier, B.V. Mattheussen, P. Storck, and M. Wiley. 2008. Hydrologic Processes 22(21):4205-4213 . DOI: 10.1002/hyp.7023.