|
<br />002.224
<br />
<br />~~
<br />~~j:
<br />'"t.
<br />
<br />Sensitivity Analysis
<br />
<br />~:~,;
<br />
<br />Results of the sensitivity analysis are listed in table 3. Sensitivity
<br />is defined as the change in error variance resulting from a 10-percent change
<br />in the value of the parameter. In table 3, the error variance between
<br />observed and simulated daily streamflow and the increase in that variance for
<br />a 10-percent change in the particular parameter are listed for each station.
<br />The most sensitive parameters for all six drainage basins used for calibration
<br />amd varification were divided almost equally between category 1, parameters
<br />from regional climatic characteristics, and from category 2, parameters
<br />determined from physical characteristics within individual HRU's.
<br />
<br />~)2..:.
<br />
<br />~~
<br />
<br />, ;.1'
<br />
<br />(:;::(
<br />~~':~i:~
<br />
<br />~~~!~
<br />,{.~:~!
<br />
<br />Two parameters from category 1 (table 2) are sensitive in all six
<br />drainage basins. The first parameter is CTS, the air temperature-
<br />evapotranspiration coefficient used in the Jensen-Haise equation (Leaves ley
<br />and others, 1983, p. 20). CTS affects the volume of water lost to evapo~
<br />transpiration and, because of the resulting soil-moisture deficits, affects
<br />the volume of surface runoff.
<br />
<br />~d
<br />
<br />;.{t
<br />, t:~t~?
<br />~~~}
<br />
<br />,;....,<
<br />'::::;':;~.
<br />
<br />Optimization
<br />
<br />ii
<br />~1:l:
<br />g;~
<br />~f:~;~~}
<br />',,"::,:,
<br />
<br />~r.~
<br />
<br />~"';:;~':
<br />
<br />t~~i~
<br />'.:::':::::.
<br />',~>':
<br />.:::,-;':':,:
<br />, ~:~<J~:
<br />
<br />i
<br />
<br />
<br />;.;:,,'.'~~'
<br />
<br />The second parameter from category 1 that is sensitive in all six drain-
<br />age basins is BST, the air temperature above which precipitation is considered
<br />all rain and below which precipitation is considered all snow (Leaves ley and
<br />others, 1983, p. 13). BST is particularly important at the onset of spring
<br />snowmelt, although it is important for each storm occurrence. Precipitation
<br />is identified in the model as rain, snow, or a mixture, depending on the value
<br />of BST and the observed air temperature. Depending on the type of precipi-
<br />tation, large volumes of meltwater can leave the drainage basin in a short
<br />period of.time or can be stored in the snowpack.
<br />
<br />Two parameters from category 2 (table 2) that are determined from phys-
<br />ical characteristics of the HRU's are sensitive in five. of the six drainage
<br />basins. The first parameter is SMAX, the maximum available water-holding,
<br />capacity of the soil profile. SMAX substantially affects the runoff component
<br />in the model. The eff,ect of SMAX in the model's water balanc.e was emphasized.
<br />by Norris and Parker (1985), and a special optimization routine for calibra-
<br />tion was developed for this parameter. The second sensitive parameter from
<br />category 2 is TRNCF, the transmission coefficient for shortwave radiation
<br />through the vegetative canopy (Leavesley and others, 1983, p. 42). TRNCF
<br />affects the energy budget in computations and, thus" the rate of snowmelt.
<br />
<br />Initial optimization was done using the parameter, SMAX, for all six
<br />drainage basins following the procedure described by Norris and Parker (1985).
<br />The objective function for this optimization was annual volume of streamflow,
<br />and parameter adjustment primarily resulted in changes in volume, although the
<br />timing of the water at the gage also was affected.
<br />
<br />~;~~~:;~ -:
<br />'-~ ;~~:.
<br />,-:;-. .
<br />,~:,~:."~..
<br />~~:'tl~
<br />
<br />:f~/,~,
<br />;:~~:':;3"
<br />:.......,
<br />, ~,,'
<br />, -, ,~
<br />.:..'"
<br />
<br />~.~~~'-?:'
<br />~.. ..'
<br />~"<\'
<br />
<br />13
<br />
<br />~M~~~~
<br />k~.T"
<br />
<br />':'-,
<br />
<br />..',>-,
<br />
<br />:,i;
<br />
|