Laserfiche WebLink
<br />OJJ386 - <br /> <br />12 <br /> <br />IV - 4,J - 2 <br /> <br />WARNING OR FORECASTING <br /> <br />Relevant factors in design <br /> <br />Even though it was decided not to attempt benefit-to-cost analysis, it perhaps would have been feasible (0 <br />make rough estimates of the relative value of the forecast service at each point and warp the network density in <br />favour of those points with greatest potential benefits. Only two classes of service value were used objectively - <br />those points requiring only Oood forecasts. and those for which operations require the issuance of forecasts during <br />low and moderate flow periods as well (referred to hereafter as "continuous"). As will be seen later. limited further <br />flexibility was provided through modest shifts in station locations to rellect subjective estimates of relative service <br />value (monetary :J.nd loss of life). <br /> <br />The remaining factors to be considered are those that affect the reliability, or error, of the forecast. If we <br />wish to achieve the same accuracy at each forecast p~int. then the required number and placement of reporting <br />precipitation stations depends on numerouS factors, such 3S: <br /> <br />I. Area of catchment; <br />2. Spatial variability of storm precipitation and other hydrometeorological elements; <br /> <br />3. Climatic considerations; <br /> <br />4. Catchment characteristics (soil types, topographic factors, channel capacity, etc.); <br /> <br /> <br />S. Relative contribution of elemental area to storm runoff. <br /> <br />The area of the catchment is readily determined and a number of studies have been reported (1-5) which <br />relate error in mean catchment rainfall to network density. It was decided to take this factor into account in an <br />objective manner as discussed subsequently. A number of possible indexes were considered for taking into account <br />the spatial variability of hydrometeorological elements and other effects of climatic and catchment characteristics. <br />The relationships involved are extremely complex and in most cases there seemed insufficient justification to compli. <br />cate the design in view of the limited knowledge and information available. In the. fUlal analysis, all indexes of these <br />factors were eliminated from objective consideration except thunderstorm frequency and mean arulUal runoff. Thun- <br />derstorm frequency is related to the relative importance of convective-type precipitation in the region and is thus <br />an index to the sp3tial variability of the most important climatic element. Channel capacities tend to increase with <br />increasing mean annual runoff and it was assumed that isolated thunderstorms are less likely to produce serious <br />flooding where channel capacities are relatively large. The storm runoff from some catchments, particularly in more <br />arid regions, originates largely in selected portions of the drainage. There is seldom sufficient information available <br />to evaluate the relative contributions from different pans of catchments under consideration and no attempt was <br />made to treat th.is factor in determining the number of reporting stations within a catchment. It was, however, taken <br />into account in the placement of stations to the extent considered appropriate. <br /> <br />Desigll criteria <br /> <br />The objective aspects of the design sludy are embodied in Figure IV - 4.3 (I) which fixes network density <br />on lhe basis of drainage area above the headwater forecast point (or the local between forecast points), thunder- <br />storm frequency (3 classes), and mean annual runoff (2 classes). The curves are applicable in those cases where <br />only flood forecasts are required. The number of stations indicated by the curves is increased by one if daily fore- <br />casts are required also during periods of low and moderate flow. This increase was believed justified because of the <br />increased value of the forecasts and to reflect the relatively greater effect of small storms on low flows. <br /> <br />Figure IV - 4.3 (I) is based on a study of rainfall variability In a region experiencing 30-45 thunderstorm <br />dlys per year. 1lle results of this study indicate that the standard error (0/0) of average storm rainfall over a catch- <br />mCTI! can bc estirn:ttcd from the equation <br /> <br />~:",,?-,,>>,,~ <br />'::ir"", <br />,"oW <br /> <br />@''''' <br />. ~"."~K~ <br />~'.~~~A~l <br /> <br />l <br />