
Reprinted from the Preprint Volume of the Conference on Climate and Water Management-A Critical Era and Conference on the Human Consequences of 1985's Climate, August 4-7, 1986, Asheville, N.C. Published by the American Meteorological Society, Boston, Mus. C4.4 SHORT DURATION RAINFALL RELATIONS FOR THE WKSTEKB UNITED STATES Richard E. Arkell and Frank Richards Office of Hydrology NOAA, National Weather Service Silver Spring, Maryland 1 • IH"l'RRDUCfiON The present study develops short duration precipitation-frequency ratios for the 10 western Long records of short-duration (less than states not included in either Frederick et al. 1 hr) precipitation observations necessary to (1977) or Frederick and Miller (1979): Arizona, estimate precipitation-frequency amounts are only Colorado, Idaho, Montana, Nevada, New Mexico, available for a relatively small number of Oregon, Utah, Washington and Wyoming. The ratios stations. This dearth of data has made the relate 5-, 10-, 15-, and 30-minute precipitation­ development of generalized short-duration esti­ frequency amounts to 1-hour amounts from NOAA mates difficult, especially in the western United Atlas 2. We addressed a number of problems in States where station density is particularly low developing these ratios. ~irst, the station den­ and where significant meteorological variation can sity was lower (17 ,000 mi /station) compared to occur over short distances. The first short th~ eastern and central United States (12,000 duration precipitation-frequency estimates for the mi /station) and California (600 mi2 /station). western United States were based on very limited Second, the rugged topography, ranging from sea data (U.S. Weather Bureau 1953, 1954). Later, level to over 14,000 ft, imposed limitations on Hershfield (1961) developed precipitation­ the data's applicability, especially since most frequency maps for the entire continental United stations tended to represent lower elevations. States and used uniform ratios to relate the Third, there are wide variations in climatology shorter-duration amounts to longer-duration within the study area. amounts. By relating the shorter durations to a longer duration that had significantly greater station density, the detailed depiction of the 2. THE DATA spatial variation of the longer duration could effectively be incorporated into the shorter The data used in this study are the duration estimates. This approach was based on largest annual precipitation amounts for 5-, 10-, the assumption that the variation of the ratio 15-, 30- and 60-minute durations. The amounts for fields was smoother than was the variation of the each duration for a given year were not neces­ absolute values themselves. sarily from the same storm, but rather were the largest amounts for that year, regardless of date Miller et al. (1973), hereafter referred of occurrence. to as NOAA Atlas 2, developed a technique to treat spatial variations in mountainous areas and The locations of the 61 stations included applied it in the western United States. Miller in this study are shown in figure 1. Of these, 55 et al. chose to adopt Hershfield's nationally had at least 15 years of data at all durations. averaged ratios for short durations. Frederick et Six stations had less than 15 years and were used al. (1977) developed isoheytal maps of short­ only on a limited basis; three stations were sig­ duration precipitation-frequency amounts instead nificantly above the surrounding terrain and were of ratios for the eastern and central United used only for comparative purposes. The earliest States. They limited their study to the largely data records go back to 1896 and the most recent nonorographic portions of the United States where data were through 1984. The average number· of meteorological variation was modest and where data years with data for stations with 15 years or density was generally highest. Finally, Frederick more of data was approximately 45 years at all and Miller (1979) studied short-duration durations. precipitation-frequency amounts in the state of California. In spite of the relatively high Each station record was examined to see if stat ion density, they decided to develop regional significant changes in location and elevation ratios rather than maps depicting the spatial occurred. Fifteen stations moved during their variation of the short-duration estimates because periods of record by more than the nominal dis­ of the large meteorological variability within the tance and elevation cutoffs of 5 miles and 200 state. feet. These 15 moves were further examined with 136 125 120 I 15 I I 0 105 FIGURE • SECTIONS WHERE RATIOS ARE NOT APPLICABLE • ~lATIONS WITH I 5 YEARS OR MORE OF OATA ED STATIONS WITH LESS THAN 15 YEARS OF OATA m HIGH ELEVATION STATIONS ~ SECTIONS OUTSIDE STUOY AREA 120 I 15 I 10 105 respect to changes in terrain, local climatology, to vary more from one year to the next at most and urban/rural character. If, for example, a locations than did the longer duration amounts, station moved 8 miles, but that move was on flat such as 24-hour observations. In addition, no terrain with no adjacent mountains, then the discernable biases were found that could be relocation was probably not of climatological attributed to urban influences. significance. On this basis, 7 stations made significant moves. We also considered the possibility of secular trends. For example, we examined the question of whether the data from one station for A detailed examination of these 7 stations the period 1900 to 1940 could be compared to the revealed no consistent biases attributable to the data for a second station which covered the period station moves. Any possible biases were apparently 1940 to 1980. Significant long-term secular trends smaller than the natural variability of the data were not evident and it was concluded that non­ themselves. Maximum short-duration amounts tended overlapping records were comparable. 137 3. PRECIPITATION-FREQUENCY STATISTICS regions, the proportion of the total number of Frequency values were determined for all annual events occurring in the most active 3-month durations by fitting the data to the Fisher­ period is lower than for other regions, being only Tippett Type I distribution using the Gumbel 55 and 60 percent, respectively. This contrasts fitting technique (Gumbel 1958). Additional with the Rocky Mountains-South and the Southwest statistics, including skew and standard deviation, Deserts where upwards to 90 percent of the largest were computed for all stations. These statistics l-hour amounts occurred during the most active 3 were useful as guides to understand similarities consecutive months, July through September. and differences in the precipitation frequencies of different.stations and different regions. For The last significant factor in determining example, standard deviations were larger in the the regions was topography. In the general sense, southwest deserts than in the coastal northwest topography is well correlated with the climatology due to the difference between the sporadic sum­ discussed above and thus is not a separate factor. mertime convective character of the first region However, on a more detailed scale, the topography and the more regular wintertime stratiform charac­ helps delineate the regional boundaries. For ter of the second. example, the crest of the Cascades separates the Coastal Northwest from the Interior Northwest in a Ratios of 5-, 10-, 15- and 30-minute well-defined fashion. Other geographic boundaries amounts to 1-hour amounts were computed for all are not as well defined. There is no sharp dis­ 61 stations for the 2- and 100-year return continuity delineating the boundary between the periods. Due to the use of ratios, no correction northern and southern sections of the Front Face was necessary to convert from annual to partial and High Plains. However, the northern boundary of duration series. The next step was to average the South Platte River Basin was chosen because these ratios over geographic regions~ this represents an approximate east-west division between where the Front Face of the Rocky Moun­ tains changes from a north-south orientation in 4. DETERMINATION OF REGIONS New Mexico and Colorado to a northwest-southeast orientation in Wyoming and Montana. This change The study area was divided into the 8 in orientation influences the availability of regions shown in figure 1 and listed in table l. moisture inflow to the two regions. The Front The determination of the number of regions in­ Face and High Plains could have been divided into volved a balance between two opposing factors. three or more regions since the ratios gradually First, the regions had to be large enough to in­ changed from south to north. However, the neces­ clude an adequate number of stations within each sity of having enough stations per region to to provide statistically stable results by virtue obtain stable ratios argued against this decision. of large sample size. Second, the regions had to be small enough so that each region adequately In some cases it was difficult to choose represented a climatologically homogeneous area. exact boundaries because a given station had sta­ The discussion below outlines how the regional tistical, climatological, and topographic similar­ boundaries were determined. ities to two adjoining regions. Such was the case for Flagstaff, Arizona, which sits on top of a rim The ratios for each duration were plotted that separates the Southwest Deserts from the on maps for both the 2- and 100-year return Rocky Mountains-South. Due to the greater simi­ periods. By plotting the ratios and finding the larity in the frequency statistics to the South­ similarities and differences between adjoining west Deserts, it was included in that region, and stations, a first pass was made at determining the the region boundary was drawn just to the north of regions. Regional breakdowns of the western states Flagstaff.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages6 Page
-
File Size-