Hydrologic Analysis for Mad River near the City of Blue Lake, Humboldt County, California Prepared for: Humboldt County Public Works Department 8 February 2013 Prepared by: Northern Hydrology & Engineering Manhard Consulting P.O. Box 2515 611 I Street, Suite A McKinleyville, CA 95519 Eureka, CA 95501 Introduction/Background This document describes the hydrologic analysis conducted for the Mad River in the vicinity of the Mad River levee (a federal levee project) near the City of Blue Lake, Humboldt County, California. The hydrologic analysis was prepared by Northern Hydrology & Engineering and Manhard Consulting for the Humboldt County Public Works Department (County). Humboldt County is currently in the process of conducting hydrologic and hydraulic analysis and floodplain mapping (Project) in the vicinity of the Mad River levee with funding provided by the Federal Emergency Management Agency (FEMA). The overall purpose of this Project is to provide accurate, up-to-date information regarding flood risk for areas formally protected by levees and adjacent areas in the study area in conformance with current FEMA standards. The work products from this Project will be used by FEMA to update the Digital Flood Insurance Rate Map (DFIRM) and Flood Insurance Study (FIS) report for Humboldt County. Detailed Study Area/Reach The Mad River detailed study area/reach is located in Humboldt County and consists of approximately 3 miles of the Mad River and approximately 1.1 miles of the North Fork (NF) Mad River (Figure 1), near the City of Blue Lake. The study area includes the approximate 1.5- mile long Mad River levee located along the right bank (facing downstream) of the Mad River and NF Mad River near their confluence (Figure 1). Basin Description The Mad River watershed is located approximately 80 to 120 miles south of the California- Oregon border, and spans both Humboldt and Trinity Counties (Figure 2). The Mad River drains approximately 497 square miles (mi2) of the North Coast Ranges of California, above its confluence with the Pacific Ocean. The basin is elongated in the south-southeast direction, and is approximately 100 miles long and 6 miles wide (Mad River Watershed Assessment, 2010). Basin elevations range from sea level to 3,000 feet along the western ridge, and 6,000 feet in the headwaters. Vegetation in the watershed is composed of early to late seral coniferous forests, hardwoods and grasslands (Mad River Watershed Assessment, 2010). Prior to entering the ocean, Mad River flows through a small estuary located west of City of Arcata and the unincorporated community of McKinleyville. The NF Mad River tributary flows into the Mad River south of the City of Blue Lake (Figure 2). The confluence is currently located near the middle of the Mad River levee, upstream of the Hatchery Road Bridge. The NF Mad River sub-basin is approximately 48.9 mi2 in size. Rainfall averages approximately 40 inches along the coast to over 80 inches at the higher elevations (Mad River Watershed Assessment, 2010). As with many north coast watersheds, 1 precipitation generally falls as rain, with snow more common at the higher basin elevations. Generally, snow melt has only a minor impact on runoff, however, rain-on-snow events can increase peak flows as occurred during the major flood of 1964. Matthews Dam (or Ruth Dam) was completed in July 1961 and impounds Ruth Reservoir on the upper Mad River situated within Trinity County. The reservoir is owned and operated by the Humboldt Bay Municipal Water District (HBMWD) to provide domestic, municipal and industrial water service. Ruth Reservoir captures runoff from approximately 121 mi2 of the upper Mad River watershed, and is relatively small with a storage capacity of 48,030 acre-feet (HBMWD website; http://www.hbmwd.com/water_supply). Ruth Reservoir is not operated for flood control purposes. The reservoir generally fills rapidly in the early winter and remains full throughout the winter/spring period, with no storage allocated for flood control. Since the reservoir has an overflow-type ungated spillway, minor peak discharge reduction occurs due to spillway surcharge storage (COE, 1968). It has been reported that peak discharge near Arcata can be reduced 5 to 15 percent depending on initial reservoir storage, while volume and duration of overbank flow are essentially unaltered (COE, 1968). Previous Peak Flows from the Effective FEMA Flood Insurance Study The effective Flood Insurance Study (FEMA, 1999) contains results of a flood-frequency analysis for the U.S. Geological Survey (USGS) Mad River near Arcata, CA gaging station (station no. 1148100). The 10-, 2-, 1-, and 0.2-percent annual chance peak discharges were determined by fitting a log-Pearson Type III distribution following Bulletin 17A (WRC, 1977) procedures to 30-years of annual-peak discharge data for the Mad River near the Arcata station. The period of record for the annual-peak data was water year (WY) 1911-1913, and 1951-1977. Annual-peak flows prior to 1962 (14-years) were reduced to account for the storage effects of Ruth Dam by the U.S. Army Corps of Engineers (COE, 1968). The flood-frequency estimates determined at the gaged site were transposed to the Mad River detailed study area near Blue Lake using Eq. 7 (described later in this report), with the exponent (b) set to 0.9. Summarized in Table 1 are the flood-frequency estimates from the effective FIS (FEMA, 1999) for the Mad River near Arcata USGS station and the Mad River detailed study area (Mad River downstream of confluence with NF Mad River). Table 1. Flood-frequency estimates for Mad River near Arcata and Mad River below Confluence with NF Mad River from Table 1 in the effective FIS (FEMA, 1999). Peak discharge estimates (cfs) for percent chance exceedance Drainage Station Name Area (mi2) 10 2 1 0.2 Mad River near Arcata, CA (station no. 485 58,360 81,270 90,960 113,480 11481000) Mad River downstream of confluence 443 53,790 74,910 83,840 104,600 with North Fork Mad River 2 Figure 1. Mad River detailed study area (source: Humboldt County Public Works, 2012). 3 Figure 2. Mad River, North Fork Mad River and Little River watershed boundaries. 4 Analytical Methods The following section describes the various methods used to analyze annual-peak discharge data for conducting flood-frequency analysis. Flood-Frequency Based on LP3 Distribution Flood-frequency estimates were computed at the gaged sites by fitting a probability distribution to the series of annual-peak discharge data following the guidelines described in Bulletin 17B (Interagency Advisory Committee on Water Data [IACWD], 1982). The T-year peak discharges (e.g. 2-, 10- and 100-year), where T is the recurrence interval in years, were estimated using the log-Pearson Type III (LP3) distribution as described by Bulletin 17B: (1) where is the peak-flood estimate at a T-year recurrence interval from the fitted LP3 distribution, is the mean of the log10 of the annual-peak discharge data, is a factor based on the weighted skew coefficient and exceedance probability and is obtained from Bulletin 17B, and is the standard deviation of the log10 of the annual-peak discharge data. The flood-frequency analysis conducted for each gaged site followed the guidelines and standards presented in Bulletin 17B. Specific Bulletin 17B procedures used in this study to refine the LP3 frequency curve included: 1. Adjusting and weighting the skew coefficient using a regional skew coefficient. 2. Low outlier treatment using Bulletin 17B defaults and visual assessment. 3. The NF Mad River annual-peak discharge record was extended using the maintenance of variance extension Type 1 technique. Software developed by the U.S. Corps of Engineers (COE) to conduct flood-frequency analysis, HEC-SSP 2.0, was used for these computations (COE, 2010). HEC-SSP fits the LP3 distribution to the annual-peak discharge data, and conducts some of the frequency curve refinements recommended in Bulletin 17B and outlined above (specifically steps 1 and 3). 5 Regional Skew Coefficient The regional skew coefficient adopted in this study is based on the nonlinear relation between skew and mean basin elevation for California (NL-Elev) developed by Parrett et al. (2011). The NL-Elev regional skew model for California is described by: (2) where is the mean basin elevation in feet. The variance of prediction (VPnew) for the NL-Elev model, which corresponds to the mean square error (MSE) used in Bulletin 17B, is provided in Table 8 of Parrett el al. (2011) based on mean basin elevation. Table 2 lists the regional skew coefficient and VPnew used in this study for the Mad River and NF Mad River. Mean basin elevations above the stations were determined using the USGS StreamStats program (http://water.usgs.gov/osw/streamstats). Table 2. Mad River near Arcata and NF Mad River near Korbel regional skew coefficient and VPnew estimates. Mean Basin NL-Elev Model Elevation (ELEV) Regional Skew NL-Elev Model Station Name above Station (ft) Coefficient VPnew Mad River near Arcata, CA (station 2,623 -0.425 0.134 no. 11481000) NF Mad River near Korbel, CA 1,527 -0.550 0.140 (station no. 11480800) Record Extension Record extension consists of extending the data in a short-record based on values in a highly correlated long-record station. The maintenance of variance extension Type 1 (MOVE1) (Hirsch, 1982) technique provides an extended peak-flow record with a variance comparable to that of the short-record. This is different from ordinary least square techniques, such as a linear regression, which can produce estimates with a smaller variance than exhibited by the short- record (Maidment, 1993). The MOVE1 approach is described by the following equation (Hirsch, 1982): 6 (3) where is the log10 estimated flow for year i at the short-record site, is the mean of the log10 values of the short-record site, is the standard deviation of the log10 values of the short-record site, is the log10 flow for year i at the long-record site, is the mean of the log10 values of the long-record site concurrent with the short record site, and is the standard deviation of the log10 values of the long-record site concurrent with the short record site.