Rainfall Depths and Intensities in Harris County (Revised 05/31/2019)
Total Page:16
File Type:pdf, Size:1020Kb
Recommendation for: Rainfall Depths and Intensities in Harris County (Revised 05/31/2019) Goal: To determine how to apply NOAA Atlas 14 1 rainfall to replace the standard rainfall amounts for Harris County from the U.S. Geological Survey. 2 This will serve Harris County for all future analysis and design. Recommended Procedure: It is recommended that NOAA Atlas 14 rainfall data be used across Harris County. As a result, it is further recommended Harris County be divided into three hydrologic regions. 3 Table 1 identifies each region by watershed and the recommended 1% AEP 24-hour rainfall from Atlas 14. Figure 2 provides a map of the three regions. Table 1: 3-Region (1% AEP/24-Hour) Depth Region Watersheds (inches) Spring Creek, Willow Creek, Little Cypress Creek, Cypress 1 16.3 Creek, Addicks Reservoir, Barker Reservoir Luce Bayou, San Jacinto River (U/S L.Houston), Greens 2 Bayou, Hunting Bayou, White Oak Bayou, Buffalo Bayou, 16.9 Brays Bayou, Sims Bayou Cedar Bayou, Jackson Bayou, Goose Creek & Spring Gully, San Jacinto River (D/S of L.Houston), Carpenters Bayou, 3 18.0 Vince Bayou, Galveston Bay Tribs, Armand Bayou, Clear Creek Assumptions: It is assumed that the NOAA Atlas 14 data is representative of rainfall across Harris County and supersedes previous data associated with TP-40 (published by the U.S. Weather Bureau in 1961), Hydro-35 (published by NOAA in 1977), WRIR 98-4044 (published by the USGS in 1998), and SIR 2004-5041 (published by the USGS in 2004). Testing Procedure: Atlas 14 is provided online and in a gridded format. This allows the user to identify various depth-duration-frequency data for specific locations. However, this would be cumbersome for HCFCD to use for the MAAPnext Study and for other agencies (Harris County, City of Houston, etc.) to administer in their floodplain management programs. Therefore, an effort was undertaken to determine whether rainfall regions could be established across Harris County with representative rainfall amounts for each region, similar to current criteria. The following approach was used to identify various regions. 1. Map Partial Duration Series rainfall contours (isohyets) across the county for the 50%, 10%, and 1% AEP/24-hour storms against the centroid of each watershed to determine the rainfall depths from the isohyets that cross each centroid. 1 https://hdsc.nws.noaa.gov/hdsc/pfds/pfds_map_cont.html?bkmrk=tx 2 Asquith, W.H., Depth-Duration Frequency of Precipitation for Texas (Water-Resources Investigation Report 98-4044), U.S. Geological Survey, Austin, Texas 1998. 3 Note that this configuration is slightly different than TSARP recommendations as the Sims Bayou watershed has been moved to Region 2. 2. Using the gridded data for the 50%, 10%, and 1% AEP/24-hour storms directly, determine the weighted average rainfall totals for each watershed and each storm. 3. Compare the results of Step #1 and #2 above (see example in Figure 1; note that first/top number listed is based on the isohyets and the second number, labeled “Wtrshd Avg” is based on the weighted average from the gridded data). Figure 1: 1% AEP/24-Hour Rainfall Totals (Isohyets vs. Weighted Avg. from Grids) 4. Determine if certain watersheds can be combined together to define regions where the rainfall is generally consistent among watersheds in each region. 5. Use the isohyet method (based on the centroid of the region) or the gridded method (based on the weighted rainfall of the region) to determine the representative rainfall amounts for an entire region. 6. Compare differences in rainfall depth between each watershed and its representative region. 7. Determine the latitude and longitude of the centroid of each region. 8. Input the latitude and longitude into NOAA Atlas 14 to determine the final Partial Duration Series depth- duration-frequency data for each location. Test Results: From Step #3 above, it was determined that rainfall results were nearly identical, whether estimated from isohyets passing through the centroid or based on weighted averages from gridded data. From Steps #4, #5, and #6, it was determined that three hydrologic regions (as described in Table 1 and illustrated in Figure 2 ) would be best representative of the NOAA Atlas 14 rainfall across Harris County. Under this grouping, the difference between weighted average rainfall for each watershed and the given region within which a watershed is located not more than 0.6 inches. In addition, the difference between each region is not more than 1.1 inches. Figure 2: Harris County Hydrologic Regions Spring Creek San Jacinto River Luce (u/s of L. Houston) Bayou Willow Creek Little Cypress Creek Cypress Cree k 2 Jackson 1 Bayou Greens Bayou Addicks Cedar Bayou Reservoir White Oak Bayou Hunting Carpenters Bayou Bayou Goose Creek Barker Buffalo Bayou & Spring Gully Reservoir San Jacinto River (d/s of L. Houston) Vince Brays Bayou Bayou Sims Bayou Armand Galveston Bayou Bay Tribs Clear Creek 3 From Steps #7 and #8, Table 2 provides coordinates used for the centroid of each region for input into NOAA Atlas 14 to determine final Partial Duration Series depth-duration-frequency data. Table 2: Centroid Coordinates by Region Region Latitude Longitude 1 30.02745567 -95.72049946 2 29.92937387 -95.33891697 3 29.74077711 -95.09958119 Exhibits 1 through 3 provide the final depth-duration-frequency and intensity-duration-frequency data 4 for each region across Harris County (please note that the regions refer to those regions identified in Figure 2 ). 4 Please note that for consistency purposes the intensity-duration-frequency data was derived from the depth-duration-frequency, and not taken directly from NOAA Atlas 14. Therefore, numbers in the intensity-duration-frequency tables may vary slightly from those published in Atlas 14. Final Notes: Use of Partial Duration vs. Annual Maximum Series MAAPnext is driving the use of consistent rainfall across varying agencies within Harris County and its incorporated areas. As a result, there is a need to define depth-duration-frequency values in such a way that they are useful not only for floodplain management purposes, but also for drainage design and analysis of smaller storm water systems. For consistency purposes, it is necessary to define depth-duration-frequency data in terms of either Partial Duration Series or Annual Maximum Series. Annual Maximum Series data has been used by HCFCD since at least the Tropical Storm Allison Recovery Project (TSARP, 2001 to 2007). However, questions have been raised related to the appropriate use of each dataset. Literature research on Partial Duration Series (PDS) and Annual Maximum Series (AMS) was undertaken to determine if some clarity could be provided on the use of one type of series over the other. [It was noted during the research that certain terms add to the confusion. For example, the term “annual maximum exceedance” refers to PDS where there are the same number of values in the dataset as there are years in the period of record; this is not the same as AMS.] The following information outlines the findings of this research. • The following references were searched. There are many other textbooks available that discuss this topic, but these authors provide a reasonable sample of experts in hydrology who have addressed the issue. o Flood Frequency Analysis by Rao and Hamed o Introduction to Hydrology by Viessman, Knapp, Lewis, and Harbaugh o Hydrology for Engineers by Linsley, Kohler, and Paulhus o Statistical Methods in Hydrology by Haan o Applied Hydrology by Chow, Maidment, and Mays • Most of the textbooks use flood frequency (not rainfall frequency) to discuss PDS vs. AMS. However, since MAAPnext (and in general, all hydrologic/hydraulic modeling) is creating flows from rainfall, the discussion should still apply. • Use of both PDS and AMS have disadvantages. o The disadvantage of AMS is that the rainfall/flood dataset contains only the highest annual rainfall/flood for a given year. In other words, there may be peak rainfall/flood events that occur within a given year that are not the highest, but may be higher than the highest peak of other years. o The disadvantage of PDS is that the rainfall/flood dataset contains all peaks above a certain threshold (hence, “partial”). However, it is difficult to establish statistical independence with such a dataset, particularly among floods. It might also be suggested (though this assumption could be incorrect) that this is true of PDS for rainfall, not because it is difficult to establish independence of rainfall events, but because the independence of a resulting flood which is being calculated from PDS rainfall is difficult to determine. o At events greater than a 10-year (10% AEP), the difference between PDS and AMS flood/rainfall estimates are approximately the same. The difference between the two methods (PDS yielding greater rainfall depths) is 12% for the 2-year (50% AEP), 4% for the 5-year (20% AEP), and 1% for the 10-year (10% AEP). • The following quotes were noted in the research. Again, all are related to flood probability, not necessarily to rainfall. o Linsley, et al (Page 359): “If the analysis is concerned with probabilities less than 0.5 [understood to mean events greater than a 50% AEP] a data series of annual floods, the largest event in each year, is the best choice. For more frequent events, the partial duration series (or simply partial series) is better.” o Linsley, et al (Page 373): “When the design problem requires consideration of events with return periods less than 5 years, a partial flood series is preferable to the annual series. Use of such short return periods might be appropriate in urban drainage, where damage is small.” o Rao, et al (Page 9): “Cunnane (1973) developed a statistical method to compare the statistical efficiency of the T-year flood by two different methods.