Precipitation-Runoff and Streamflow-Routing Models for the ...

tion value of the HRU class by the observed meanannual precipitation of the precipitation gage, and thisadjustment was used as model input. Generally, theprecipitation gage nearest an HRU was used as theprecipitation station for that HRU. It should be notedthat precipitation gages are often subject to an undercatch of precipitation (especially snow), particularly ifthey are not protected by wind shields. In final calibration,a more general basinwide adjustment also wasapplied to each subbasin by adjusting the precipitationinput of all HRU’s within a subbasin equally to accommodatethe water balance. Generally, precipitationinput was adjusted upward.Temperature data adjustments (lapserates)—Minimum and maximum air temperature datawere used by PRMS to compute potential evapotranspirationat each HRU. To account for the air temperaturedifference between the elevation of thetemperature station and the mean elevation of theHRU, minimum and maximum air temperature lapserates (change in degrees per 1,000 ft change in elevation)were included as PRMS parameter input. Usingthe 30-year (1961–90) mean monthly minimum andmaximum air temperature data, 18 monthly lapse rateswere computed for various locations throughout theWillamette River Basin (fig. 8). These tables can befound in individual PRMS basin.g1 files used in modeling.An example can be found in Appendix 5 ingroups 22 (maximum rate) and 23 (minimum rate).The lapse rate table found in Appendix 5 was developedfrom temperatures collected at stations 33 and 16(fig. 8).Basinwide ParametersMost basinwide parameters in PRMS are relatedto potential evapotranspiration, subsurface flow,ground-water flow, snow, and snowmelt processes.Potential Evapotranspiration—The Hamon(1961) method was used in this study to estimatepotential evapotranspiration. The method incorporates12 monthly coefficients (CTS) that are used to adjustminimum and maximum air temperatures. The coefficients,shown in table 8, were adjusted during the calibrationprocess to help replicate an appropriateestimate of actual evapotranspiration losses from thebasin and define the water balance. Regional pan evaporationdata were used to identify extremes.Subsurface flow—Subsurface flow, as defined byPRMS, is the relatively rapid movement of water fromthe unsaturated zone to a stream channel. Subsurfaceflow is computed by using a conceptual reservoirroutingsystem. Two parameters, RCF and RCP, controlthe rate of subsurface flow as a function of storagevolume in the subsurface reservoir. If both parametersare used, the routing relation becomes nonlinear. Otherwise,if RCP is set to zero, the relation becomes linear.Although values for these parameters cannot bedetermined through field measurement, initial valuesfor RCP and RCF can be determined using graphicalflow-separation techniques. The PRMS manual(Leavesley and others, 1983) provides some guidancefor using these techniques. The RCF and RCP valueswere later adjusted during the calibration processusing the Rosenbrock optimization routine in PRMS(table 9).Ground-water flow—The ground-water systemin PRMS is conceptualized as a linear reservoir. Baseflowleaving the ground-water reservoir is controlledby a single parameter, RCB. RSEP determines flowentering the ground-water reservoir. As with subsurfaceparameters, these parameters could not be determinedthrough field measurement, so the value wasinitially estimated using graphical flow-separationtechniques and adjusted during the calibration process(table 9).Snow and snowmelt—The snow componentof PRMS was used to simulate the initiation, accumulation,and depletion of a snowpack on each HRU.Critical snow parameters used in PRMS includeTRNCF, a transmission coefficient for the vegetationcanopy over the snowpack; PAT, the maximum air temperaturethat causes all precipitation to become rain;CECN, a convection-condensation energy coefficient;BST, the temperature below which precipitation issnow and above which it is rain; and EAIR, the emissivityof air on days without precipitation. Precalibrationvalues used for these parameters were PRMSmodel default values. The values were later adjustedusing optimization techniques during the calibrationprocess. Most of the 10 calibration basins used in thestudy were in lower elevations where snowmelt wasnot a significant component of the hydrologic cycle.Snow accumulation and melt, however, was significantfor the Lookout Creek Basin located in the headwaters of the McKenzie River, so calibration of thesnow-related parameters was based on data from thisbasin only.Model Calibration and VerificationPrecipitation-runoff modeling was used in thestudy to estimate inflows from ungaged basins for use22