Laserfiche WebLink
<br />the basin. This data will be disaggregated both spatially and seasonally to develop monthly <br />flows for approximately 180 gaged locations. The form of the model will insure that some <br />statistical properties are preserved while other properties are not implicitly preserved. <br />Therefore, developing a stochastic model must be evaluated for each site to assure that all the <br />appropriate statistical properties are preserved. For example, spatially desegregating the annual <br />flows will assure that the total annual flows at gaged and ungaged locations sum to the <br />downstream gage. But if each gage is then seasonally disaggregated individually, then the sum · <br />of anyone month's gaged flows may result in an ungaged flow which is unreasonable at a <br />downstream gage. This anomaly may be resolved by developing more parameters in the model, <br />but estimating additional parameters may be limited by the length of the historic data record <br />and the degrees of freedom of the model. Therefor the model must preserve these relationships <br />and not have so many parameters that the historic data cannot be used to estimate them. <br /> <br />Developing climatic data to estimate demands is not expected to be performed in this subtask. <br />Rather the program developed previously, which estimates demands as a function of a wet, dry <br />or average streamflow, is expected to be used to estimate demands that correspond to the <br />stochastic streamflows. <br /> <br />12.2.Stochastic Model Verification In this subtask, the stochastic models will be verified to insure <br />the characteristics and interrelationships of the data have been preserved. Data will be generated <br />at each gage location and the characteristics ofthe generated data will be evaluated and <br />compared to historical data. Discrepancies will be noted, adjustments to the models will be <br />made, and results verified. <br /> <br />This subtask may be an extensive, iterative process since many models will not preserve key <br />historical statistical characteristics such as drought length or severity. The process becomes <br />iterative because a property such as a drought characteristics can only be evaluated after the <br />model is executed and the results are evaluated. <br /> <br />12.3.Generation of Data Sets In this subtask, the stochastic models developed and verified <br />previously will be used to generate data sets for the evaluation of drought characteristics and <br />comparison of stochastic data to the historic and extended data sets. This subtask will be <br />complicated only by the number of gaged sites where flows are generated and not by the <br />complexity of the data. Several data sets representing design drought sequences, stochastic data <br />sets of varying duration, and others will be developed for use in CRDSS. There may be separate <br />sequences for each major basin that allows the most severe droughts to be used in each basin <br />with less severe conditions to be assumed in adjacent basins. It is assumed that these sequences <br />will be selected from a stochastically generated trace on the order of 10,000 years length, i.e. <br />100 traces 100 years in length. <br /> <br />12.4.Interface Design In this subtask, the Graphical User Interface (GUI) designed previously will <br />be revised to accommodate stochastic data development. <br /> <br />12.5.Draft Stochastic Model Documentation In this subtask, the stochastic model development <br />will be documented in a memorandum containing approximately 15 pages of text. The <br />memorandum will describe the model selection, the algorithms used, and the checks performed <br />to validate the models. <br /> <br />0338 <br /> <br />c: \w\crdss\scope\boyI Ilb\crdss3 .doc <br /> <br />5/30/97 - Page 19 <br />