7812 - Laserfiche WebLink

Home Browse Search

all but two gages; the two methods pro- duced results that deviate from each other from 1.73 to 46.84 percent. The New England ABF method intro- duced an additional bias by utilizing a mean when combining the 48 gage data points without regard to data variability caused by regional factors. In a data base such as discharge on unregulated streams, which is populated mostly with low values, a few high values can raise the mean substan- tially above the point where most contrib- uting data points lie. The flow duration method reduces bias caused by unusually wet or dry periods in the period of record by using all daily val- ues for all dates on record for a given month as individual samples to determine an in- dividual stream's discharge. Data are not grouped by year or averaged at any level. By drawing from a large sample size (using all daily flow values versus a single month- ly value for each year of record), the sta- tistical influence of abnormal data events is reduced. The median, which corresponds to the point exceeded 50 percent of the time on a flow duration curve, is a more represen- tative descriptive statistic than the mean in any skewed data distribution, such as streamflow data. The combined use of a median statistic with a relatively large sample size reduces bias due to a skewed discharge frequency distribution. The flow duration curve, which generates a median value, is a widely accepted technique used to derive the likelihood of occurrence for a given flow value for an individual gage and period of record. However, it was not used as the basis for creating the New En- gland ABF recommendation value. Although an ABF recommendation val- ue could be calculated by averaging the 48 median flow values from flow duration data, it would not be appropriate for in- dividual streams because environmental factors cause distinctly regional variations in August flow throughout New England. Therefore, the ABF policy should be re- fined by providing region-specific recom- mendations, based on existing flow dura- tion data. REGIONALIZATION OF THE ABF POLICY When applied across a large and diverse geographic area, a policy-based ABF rec- ommendation should, if possible, take into account statistically significant differences among hydrologic regions and assign a re- gion-specific cfsm value to any given water diversion project, based on known basin characteristics. This section of this paper identifies hydrologic regions of New En- gland, based on physiographic basin fea- tures, rainfall, and a statistical analysis of discharge data; these issues were not ad- dressed in the New England ABF policy. The New England ABF policy value does not consider whether flow patterns among the 48 gaged streams used in its analysis are affected by differences in basin char- acteristics such as climate, geology, topog- raphy and other environmental factors (G. Russell, personal communication). How- ever, the wide range of August cfsm values (0.16-0.78, based on flow duration curve data) (Table 1), and poor relationship with the mean (0.40 cfsm) among the gages used to develop the policy suggest that August flows in unregulated New England streams are a reflection of individual physical river basin factors. Further statistical analysis of the cfsm values from these gages shows a poor relationship between the standard descriptive statistics of central tendency (mean and median). Factors that influence precipitation and runoff are generally recognized as impor- tant determinants of str amflow in unreg- ulated streams (R. Milh us, personal com- munication; Morhardt and Altouney 1985). Dingman (1979), for example, indicates that flow prediction methods developed for un- gaged streams in New Hampshire should only be applied in regions with similar el- evations and climatic factors. Factors that most affect August discharge catalogued by the USGS in basin characteristics files in- clude rainfall, slope, land use, and topog- raphy (groundwater geology becomes a base flow factor only during drought pe- riods (W. Bartlett, personal communica- tion). As part of this analysis, these basin char- acteristics were used to classify the New England unregulated stream gage data into logical hydrologic regions. The validity of the regional grouping was then tested 'N_ 14 Rivers • Volume 1, Number 1 January 1990