J5.2 PROBLEMS OF CLIMATE VARIABILITY AND UNCERTAINTY IN FLOOD HAZARD PLANNING FOR THE COLORADO FRONT RANGE Mary W. Downton*, Heidi Cullen, Rebecca Morss, Olga Wilhelmi National Center for Atmospheric Research, Boulder, CO Balaji Rajagopalan University of Colorado, Boulder, CO * Corresponding author address: Mary W. Downton, NCAR, P.O. Box 3000, Boulder, CO 80307; email: down- ton@ucar.edu. 1. INTRODUCTION Since the creation of the U.S. National Flood Insurance Program (NFIP) by Congress in 1968, ex- tensive efforts have been made to reduce the impacts of flooding through floodplain mapping and improved community planning. In the last two dec- ades, the Federal Emergency Management Agency (FEMA) has made substantial progress in motivating states and local governments to participate in NFIP. By 2001, over 19,000 communities were involved. In each NFIP community, a Flood Insurance Study is undertaken to analyze flood hazards in the commu- nity and prepare a Flood Insurance Rate Map (FIRM). The FIRM is intended to be the basis for floodplain management, mitigation, and insurance activities under NFIP. Many of the existing floodplain maps have become seriously outdated as communities have grown and floodplain conditions have changed. FEMA acknowledges that, nationwide, nearly two- thirds of its flood maps are over 10 years old and “re- flect outdated flood hazard data because watersheds and floodplains have changed faster than FEMA could afford to prepare updated maps” (FEMA 2002a). FEMA has initiated a Map Modernization Program to update the FIRMs and convert them to a digital format that will simplify their distribution, us- age, and future revisions. In Colorado, goals of map modernization are to reduce the average age of the state’s FIRMs from over 13.6 years to 6 years or less and to develop flood hazard maps for many un- mapped flood prone communities (Browning et al. 2002). FEMA prescribes extensive requirements and guidelines that their “Flood Hazard Mapping Partners” (state and local agencies and study con- tractors) must follow in mapping and regulating floodplains. Acceptable methods and models are strictly specified, designed to create uniform levels of protection throughout the U.S. Assessments of flood risk are based on estimates of discharge, ve- locity, and depth in a "design flood", which is usually defined as the 100-year (1% probability) flood, al- though individual communities may choose to fund a higher level of protection for themselves. Many of the methods were developed by the U.S. Army Corps of Engineers and the U.S. Geological Survey in the 1970s and early 1980s and have received only minor additions and revisions since that time. Flood discharge in the 1% annual chance flood is the basis for the hydrologic and hydraulic analyses that determine base flood elevation, flood- plain boundaries, and other risk-related variables such as depth and velocity. Thus, uncertainty in the flood discharge estimates used as input to hydraulic models is a fundamental source of uncertainty in the resulting floodplain maps and vulnerability as- sessments. Methods of riverine floodplain mapping rec- ommended by FEMA indicate strong preference for use of historical streamgage measurements to es- timate flood discharge (FEMA 2002b). Even on gaged streams, however, the existing streamflow data often violate the assumptions of flood fre- quency analysis (Thomas 2002). For ungaged, unregulated streams, regression equations devel- oped by USGS are recommended to estimate discharge. However, standard errors of the regres- sion estimates for the Colorado Plains region, which includes the Front Range, vary from 204% to 306% (Vaill 2000), partly because of insufficient data. Rainfall-runoff models frequently are used to esti- mate peak stream discharge because of the high spatial and temporal variability in Colorado precipi- tation. There is considerable evidence that the meteorology of extreme storms in Colorado is not 1