Laserfiche WebLink
<br />Memorandum To: Ray Alvarado March 10, 2003 <br />Page 3 <br />• Yield of exchanges that typically operate in sub-monthly windows <br />• Peak flow rates and volumes for environmental uses <br />• Demands or rights that cease or change mid-month (e.g., instream flow filings) <br />One element that potentially mitigates the preference for a daily model is availability and curacy of <br />basic daily data. If accurate daily data were as easily available as monthly data, daily time s ~p modeling <br />would perhaps often be preferable. However, since considerable amounts of daily data gen ally need to <br />be estimated, the overall accuracy of a daily model may not be significantly greater than a nthly <br />model. In other words, one should not select a daily approach solely on the assumption that >better <br />calibration can be achieved. Depending on the data available and the estimation techniques sed, that <br />may or may not be the case. Ultimately, the decision to use a daily model comes down to t importance <br />of sub-monthly phenomena in the application, the availability of daily data, and the ability estimate or <br />disaggregate monthly values to daily values in a realistic way. <br />2.0 The routing issue <br />Streamflow routing, that is, accounting for travel time and attenuation of reservoir releases <br />inflows, is not an issue in monthly time step modeling. In large systems such as the Colora <br />where it can take two to three days to deliver water from Green Mountain Reservoir to the <br />Valley, routing can be an issue on a daily basis. <br />While it is recognized that StateMod is not currently capable of flow routing, this is not co idered a <br />significant drawback when modeling for planning, as opposed to operational, purposes. In e case of <br />reservoir releases, the volume of water potentially delivered early because travel time is i red is offset <br />by the potential for over release after demand is satisfied. <br />Furthermore, if StateMod's baseflow estimation method is applied directly to daily data, it <br />for routing impacts. StateMod calculates reach gains and losses by doing a mass balance be <br />flow points, including the effects of diversions, reservoir storage and releases, return flows, <br />to the extent they are known or estimated. All other effects and estimation errors are reflect <br />baseflow, which is the residual term in the balance. Thus, in the circumstance where water 1 <br />released from a reservoir but is not yet "seen" by the downstream gage, the baseflow gain/1~ <br />intervening reach will be lower than it would be otherwise, until the released water shows u <br />total amount of water in the river will be correct. <br />3.0 The computer resources issue <br />Another potential drawback of a daily model, at least in the past, has included the practical <br />computer resources. Computer performance time is much less important now than just a fe~ <br />However, it is still true that a daily model will require several orders of magnitude more tir. <br />than the analogous monthly model. 'Since monthly model execution time varies widely amc <br />models, the daily execution time maybe significant in some cases and less so in others. Ea <br />suggests that running the daily Rio Grande or Colorado models for 25 years could take sev~ <br />awell-equipped (e.g. a P4 xx MHz processor) computer. <br />account <br />in the <br />> been <br />> for the <br />Thus the <br />itations of <br />;ars ago. <br />hours on <br />CS-B99-190-00/B /TASKMEM03PINAL.DOC E3OYLE <br />