Laserfiche WebLink
a 4.8 value would be 20% sand and 80% gravel. Intermediate code values <br />refer to a percentage mixture, not a size gradation. <br />FREQUENCY ANALYSIS <br />Frequency analysis was preferred over all other curve construction <br />techniques because the shape of the curves could be directly determined. <br />Data for frequency analysis consisted of measurements of depth, veloc- <br />ity, and substrate at individual capture or observation locations for <br />individual fish. Additional information included the species observed <br />and its length for each set of hydraulic parameter data, as well as <br />information such as water temperature, use of cover objects, etc. <br />Frequency analysis was occasionally possible using data obtained <br />from laboratory experiments testing the preference or tolerance of a <br />species, or life history phase of a species, to single or multiple <br />hydraulic parameters. An example would be tests on the swimming ability <br />of fry of a certain species at different temperatures. The applica- <br />bility of such experimentally derived data to frequency analysis <br />depended upon the number of replications obtained and the variety of <br />conditions tested by the experiment. Table 3 shows the type of data <br />most commonly used in frequency analysis. <br />A consistent procedure was followed in the definition of optimum <br />conditions and probabilities of use for each species and life stage: <br />1. Each hydraulic parameter was assessed independently by listing <br />a continuum of the parameter, starting at zero, and extending, in equal <br />increments, to or beyond the total range of use. <br />2. The total number of individuals of a certain species and life <br />stage was tallied and assigned to the appropriate increment of each <br />parameter. <br />3. Total numbers (frequency of individuals) in each parameter <br />increment were summed. <br />4. Adjacent increments were then clustered to reduce the variance <br />typical of such a distribution. Clustering was limited to two adjacent <br />increments, but each increment was clustered two ways to determine the <br />pattern giving the least variance (Table 4). <br />6 <br />