|
<br />0022.31
<br />
<br />.,,:-~
<br />:,~'~ ;'~.
<br />
<br />:':<,'
<br />
<br />"
<br />
<br />Additional difficulty in predicting low streamflows may be the result of
<br />a lack of appropriate algorithms in PRMS. The model does not include alluvial
<br />aquifers or the transfer of water into or out of such an aquifer during low-
<br />streamflow conditions. If this inclusion is an important component in causing
<br />low streamflow, PRMS cannot duplicate the situation. The importance of allu-
<br />vial aquifers or the transfer of water is unknown in the drainage basins
<br />studied. PRMS assumes a closed system; therefore, it is difficult to simulate
<br />ground-water source areas, influencing low streamflow, that are larger than
<br />the basins being modeled.
<br />
<br />)!:'.,'"
<br />',',"
<br />~:';'-\
<br />:~:;;i
<br />t:ii
<br />~',;',:
<br />:: :;"~.,
<br />
<br />'.',~
<br />,...,'
<br />
<br /><:..-.:
<br />',:::'
<br />;'i::
<br />,;;';':-,
<br />
<br />~:i>
<br />~~.;:
<br />
<br />Flow-duration curves provide a logical extension of the singular, statis-
<br />tic, percent of days of zero streamflow. The flow-duration curve includes a
<br />frequency of all streamflow values and not just the frequency of zero flow
<br />days. Flow-duration curVes are shown for each of the s,ix streamflow-gaging
<br />stations (figs. 3 through 8) for the period of record.
<br />
<br />'.:--:::::
<br />::~i
<br />~~;;~~
<br />, !1-Y:
<br />-,,:.'.
<br />..:.'"
<br />f~1j;:
<br />, ~-..
<br />
<br />i
<br />~J;
<br />[!~
<br />I
<br />
<br />~~i
<br />i
<br />~~
<br />;,,~
<br />
<br />fl,
<br />
<br />:::I';--:;'
<br />,:'~'X':
<br />~,;::::,j
<br />'f;t,~~:
<br />;{;:":~
<br />/.<;::,]
<br />
<br />The steep flow-duration curves indicate the ephemeral nature of the
<br />streams, with the possible exception of Wilson Creek (fig. 7). Most simulated
<br />flow-duration curves generally correspond to the observed curves; the corre-
<br />spondence is greatest for the midrange streamflows. At high flows (less than
<br />1 percent of the time that flow is equaled or exceeded); there is a deviation
<br />of the simulated streamflow from the observed streamflow for most of the
<br />drainage basins. For flows less than 1.0 ft3js, deviations of,the simulated
<br />streamflow from the observed streamflow also tend to occur. These deviations
<br />for the smaller streamflows have been described in the paragraphs above about
<br />the zero-flow statistic.
<br />
<br />Additional difficulty in predicting streamflow is indicated when the
<br />water balance produced by the model is studied (table 7). In the water
<br />balance, net precipitation is total precipitation minus losses from inter-
<br />ception. Basin storage is the sum of all storage in the soil profile, the
<br />subsurface reservoirs, and the ground-water reservoirs. In the overall water
<br />balance, much of the net precipitation was lost to evapotranspiration. The
<br />average loss for all drainage basins, for all years, was 96 percent. During
<br />some years, evapotranspiration exceeded precipitation because water was
<br />extracted from basin storage. Actual streamflow was a small component of
<br />the water balance and streamflow comprised the remaining' 4 percent of net
<br />precipitation.
<br />
<br />Because streamflow is such a small percentage of the water balance,
<br />substantial difficulty occurs when calibrating PRMS or any model for this
<br />semiarid environment. The error generated when distributing the point values
<br />of precipitation in time and space probably is larger than the annual volume
<br />of streamflow. Furthermore, the error generated when predicting evapotrans-
<br />piration values probably is larger than the annual volume of streamflow. Both
<br />of these errors are combined in this analysis. Neither source of error can be
<br />defined effectively without intensive data collection. In hydrologic regimes
<br />where streamflow is a more substantial part of the water balance, errors in
<br />precipitation distribution and evapotranspiration may not result in large
<br />difficulties in calibration, but, in a semiarid region, such errors can easily
<br />hinder streamflow prediction.
<br />
<br />";"z:,;',
<br />
<br />':~"::';:':
<br />~.{;::.~~:
<br />:~.:~':::;
<br />::;~\~\,
<br />~:t;'>,',:~
<br />
<br />20
<br />
<br />" ,,~':,
<br />~?>,;,;~
<br />A}{
<br />[t't:
<br />~~~;~::
<br />;".,'
<br />
|