Laserfiche WebLink
<br />001447 <br /> <br />INADEQUACY OF SAMPLING <br /> <br />The n?tural range of most hydrologic variables is large and some <br />classes of records sample the range most inadequately. CClllIII10nly this <br />inadequacy of sampling handicaps the hydrologist much more seriously <br />than errors of measurement at the points sampled. <br /> <br />For example, there is ample evidence that instantaneous rates of <br />precipitation at a given station vary considerably from moment to <br />moment. At the particular station, this variability probably compensates <br />in some degree over the term of a single storm, and more so over a <br />season or year. Similarly, among the stations ofa network, substantial <br />compensation occurs within the geographic reach of a particular storm or <br />over any extensive area. Nonetheless, even when spaced more closely than <br />is ordinarr, a network of precipitation stations takes only a woefully <br />small sample of the water precipitated. Thus, although the conventional <br />statistical records are a usable index to relative amounts (volumes) of <br />water precipitated on an area, they afford only a rough measure of that <br />volume. If a hydrologic study involves mean depth or volume of water <br />precipitated on an area, the most probable value is that derived from <br />an isohytal map constructed with due regard to all recognized parameters <br />of precipitation. Any value derived by arithmetical. procedures alone <br />would be less defendable. (It should be noted that these li1l1itations of <br />preCipitation records are inherent; they are independent of the accuracy <br />of measurihg devices and procedures.) <br /> <br />In contrast, a stream-gaging station measures theiritegrated volume of <br />water running off from the drainage area. Thus, the conventional record <br />of streamflow is li1l1ited inherently not by inadequacy of sampling, but <br />by accuracY of techniques for measurement, which are considered later in <br />this manual. <br /> <br />Among the hydrologic variables listed above, and in addition to <br />streamflow, only storage in reservoirs and lakes is measured directly <br />in conventional records. All the other kinds of records produce data <br />that are inherently index values rather than absolute measures of the <br />hydrologic variable being sampled. Probably few samples are integrated <br />in the sense that they reflect, in true proportion, all conditions or <br />aspects of, the particular water body for a particular moment or interval. <br />of time. Thus, in any hydrologic study that involves correlation among <br />two or more of the variables, and in which the available data is inadequate <br />for computing a sound correlation coefficient by statistical procedures, <br />conclusions should be drawn cautiously. <br /> <br />2 <br /> <br /> <br />';';',i.. <br />