My WebLink
|
Help
|
About
|
Sign Out
Home
Browse
Search
FLOOD01900
CWCB
>
Floodplain Documents
>
Backfile
>
1001-2000
>
FLOOD01900
Metadata
Thumbnails
Annotations
Entry Properties
Last modified
11/23/2009 10:40:54 AM
Creation date
10/4/2006 10:26:47 PM
Metadata
Fields
Template:
Floodplain Documents
County
Statewide
Community
Nationwide
Title
Techniques of Weather-Resources Investigations of the USGS Book 4, Chapter A1
Date
1/1/1968
Floodplain - Doc Type
Educational/Technical/Reference Information
There are no annotations on this page.
Document management portal powered by Laserfiche WebLink 9 © 1998-2015
Laserfiche.
All rights reserved.
/
44
PDF
Print
Pages to print
Enter page numbers and/or page ranges separated by commas. For example, 1,3,5-12.
After downloading, print the document using a PDF reader (e.g. Adobe Reader).
Show annotations
View images
View plain text
<br />4 <br /> <br />TECHNIQUES OF WATER-RESOURCES INVESTIGATIONS <br /> <br />. <br /> <br />Statistical inference <br /> <br />We have 54 years of record on the Rappahan- <br />nock River of Virginia and might ask two ques- <br />tions about the mean flow. First, what is the <br />mean flow for the period of record? This is a <br />unique value which can easily be computed. <br />The second question, what is the mean flow of <br />the stream?, cannot be answered definitely. We <br />Can only assume that the mean of the 54-year <br />sample is an estimate of the true (population) <br />mean. In other words, we infer the population <br />charact~ristics from those of a sample from that <br />population. <br />Statistical inference is based on the theory <br />of sampling. From a population of known char- <br />acteristics many samples are drawn (either <br />actually or conceptually), and the relation of <br />the sample characteristics to the population <br />characteristics is defined. <br />Sampling theory requires use of the concept <br />of a probability distribution. Assume that the <br />distribution of some random variable is normal <br />with mean 1', and standard deviation, q, as <br />shown in figure 4. (The term "random," as <br /> <br /> <br />J.t <br /> <br />Figure 4,-Normal distribution. <br /> <br />used here, means that the probability of drawing <br />anyone item of the population is the same as <br />for any other.) <br />Now suppose we take many samples of size <br />N from tills distribution, compute the mean of <br />each of these samples, and compute the mean <br />and variance of these sample means. The dis- <br />tribution of the means of samples of size N is <br />superposed on the original distribution in figure <br />5. It Can be shown that the distribution of <br />the means is centered at I' and that the stan- <br />dard deviation of the distribution of means is <br /><r/m. Therefore, the mean of the means of <br /> <br /> <br />Distribution of means <br />of size N <br /> <br /> <br /> <br />Original distribution <br /> <br />J.L <br /> <br />Figure 5,-Oistribution of means of samples from a norma- <br />distribution. <br /> <br />samples of size N is an unbiased estimate of 1'. <br />Furthermore, the mean of one sample is an <br />unbia~ed estimate of 1'. Consequently, we infer <br />that the sample mean, X, is an estimate of the <br />population mean. Obviously, if we used other <br />samples we would obtain d:i1ferent estimates <br />of the population mean. <br />From a single sample, we can appraise the <br />reliability of the estimate, X, of the population <br />mean. The distribution of means of values <br />drawn from a normal distribution is normal. <br />Consequently, two-thirds of the values should <br />fall witilln one standard deviation (q/~N) <br />on each side of the mean. However, we do not <br />know q so we have to substitute S for it (where <br />S is the standard deviation computed from the <br />sample). The distribution of X having a <br />standard deviation of S/.,fli.f is known as the <br />Student's t distribution, values of which are <br />tabulated in statistics texts for various sizes <br />of N. <br />Suppose now that we have K samples of <br />size N and have defined K different sampling <br />distributions of the mean of size N. For each <br />sampling distribution, we can define a mean <br />and a range of reliability, and we. are interested <br />in whether such a range includes the true mean <br />1'. Considering the range as a random interval, <br />we may state that the probability (P) that <br />the random interval includes I' is 1-e, where e <br />is the level of significance. Mathematically, for <br />e=0.32, <br /> <br />. <br /> <br />P[(X -o-/../N) <I'<(X +q/~N)]=1-e=0.68. <br /> <br />. <br />
The URL can be used to link to this page
Your browser does not support the video tag.