Laserfiche WebLink
<br />T860 <br /> <br />methods yield results not significantly different, they can <br />be substituted. <br /> <br />t = D/SD <br /> <br />(1) <br /> <br />SampUng Design <br />In planning a study to verify the salt-load reduction result- <br />ing from a salinity control project, several critical decisions <br />about documentation and sample size are necessary. Sam- <br />pling locations and frequency must be selected and the <br />total number of samples must be determined. Develop- <br />ment of written documentation of the sampling design in- <br />cludinglocations, parameters, frequencies, program ration- <br />ale, and reporting procedures is critical attbis stage of the <br />work. <br /> <br />Because SD is.a function of sample size, equation 1 can be <br />used to estimate the number of samples necessary to deter- <br />mine the significance of a salt-load reduction at a given <br />level of confidence. Assuming the standard error of mean <br />load is 10 percent of the preproject mean, the number of <br />samples required to detect significant (90 percent con- <br />fidence level) decreases in salt load are shown in the <br />following table: <br /> <br />Detectable decrease <br />in salt load <br />(percent) <br /> <br />Required number of samples <br /> <br />Tbe choice of sampling location is usually made based on <br />reconnaissanCe data collected during the identification <br />stage of the project. Computation of dissolved-solids load <br />at the selected locations requires sampling frequently <br />enough to include the major variations in load at the site. <br />For wells, springs, and larger streams where variations in <br />flow and dissolved-solids concentration occur gradually, a <br />monthly sampling frequency may be adequa~e. For small <br />streams, which respond rapidly to storm and snowmelt <br />events, sampling should be frequent enough to determine <br />the impact of these events on dissolved-solids loading. <br />During the base-flow period, concentration and <br />streamflow may be almost constant and sampling frequen- <br />cy can be decreased. However, base-flow data are often <br />most important in salinity studies; therefore, sampling , <br />during base flow should not be neglected. <br /> <br />Tbe required number of samples should be based on the <br />statistical tests used to determine the significaiice of <br />, project effects. The most common test is a 'comparison <br />between preproject and postproject mean salt load <br />downstream from the project area. If a student's t-test is <br />used to compare the means, the test statistic (I) is com- <br />puted as the ratio of the difference between preproject <br />and postproject mean salt load (D) and the standard error <br />of this difference (SD): <br /> <br />Preproject <br /> <br />Postproject <br /> <br />15 <br />10 <br />5 <br />4 <br /> <br />3 <br />5 <br />15 <br />25 <br /> <br />3 <br />5 <br />15 <br />25 <br /> <br />This analysis assumes the samples are ~~ded evenly be- <br />tween preproject and postproJect. A llUI1lIllum of three <br />samples are required from each pe~od to compute SD: If <br />three preproject and three postproJect ~amples are av~- <br />able the minimum significant decrease m salt load which <br />can be' detected is 15 percent of the preproject load. If <br />more samples are taken, the minimum dete,ctable signifi- <br />cant decrease can be reduced as shown. <br /> <br />If annual means are being compared, each annual value is <br />considered, one sample. Therefore, a minimum of 3 years <br />each for preproject andpostproject sampling is required. <br />If the expected change in salt load is less than 15 per~nt, <br />additional years of sampling will be necessary. To mirli- <br />mire data collection, sampling sites should be selected at <br /> <br />28 <br /> <br />29 <br />