The Federal Emergency Management Agency (FEMA), which administers the National Flood Insurance Program (NFIP), is authorized to identify natural hazards throughout the United States and its territories. The geographical diversity of the nation provides a wide range of natural hazards, but one of FEMA's key responsibilities is to map areas that are subject to a 1 percent probability of being flooded in any year (the "100-year flood"). The purpose of this charge is to meet the NFIP's requirement that the burden of paying for blood damage be shifted from the general public to those living at risk. In most riverine environments, where channels change their locations only gradually and where catastrophic alterations in their form and flood conveyance capacity during a single event are rare, the procedures for mapping the depth and velocity of floods are generally agreed on. The technical and regulatory community has developed certain language, procedures, and a way of depicting reality (i.e., a paradigm) that allows the identification, delineation, and mitigation of flood hazards (see, e.g., Hydrologic Engineering Center, 1976, volume 6; and Bedient and Huber, 1992, Chapter 7). Although all floods behave, in detail, differently from the paradigm, once an estimate of the 1 percent peak flood discharge is agreed on, institutionalized procedures make the calculation of the extent, depth, and velocity of the flood hazard relatively straightforward and reproducible by different analysts. This report uses the term riverine flooding to represent those cases where application of this standard paradigm allows one to successfully assess and manage flood risk.
However, where catastrophic changes in river channel form and position can occur during a single flood, the traditional paradigm and associated hydraulic procedures cannot be relied on. For example, if a flood deposits large quantities of sediment on the channel bed in a reach, the conveyance capacity of the channel could be reduced drastically and the flow forced overbank at a lower discharge than would be predicted from prestorm surveys of the channel geometry. If overbank flooding causes erosion of a new channel or the reoccupation of an old channel, flood risk assessments based on the historical flow path would misrepresent the location and intensity of flooding downstream of the change. Both of these types of channel changes (form and position) can occur with great frequency and intensity on a type of landform called an alluvial fan. An alluvial fan, as defined by this committee, is " a sedimentary deposit located at a topographic break, such as the base of a mountain front, escarpment, or valley side, that is composed of
fluvial and/or debris flow sediments and which has the shape of a fan either fully or partially extended."
Despite the fact that to geologists alluvial refers strictly to features and materials deposited by streams of water (American Geological Institute, 1976), in the case of fans the term has been used more loosely in the scientific, engineering, and planning literature to refer to the products of streams or debris flows. This usage is too well established for this committee to reverse. However, to a person actually analyzing flood risk on the ground, the distinction is important, and so the committee will recognize the needs of these people in Chapter 2, which is concerned with the physical processes associated with flooding rather than policy and regulatory aspects. Chapter 2 explains that alluvial fans can be deposited by streams or by debris flows or by some combination of the two, and that recognition of the difference between these situations can be crucial for correctly identifying the hazard potential in certain areas. The chapter thus distinguishes between streamflow fans, debris flow fans, and composite fans. However, the committee has elected to follow common usage and use alluvial fan as the generic term for any of these categories.
In NFIP Regulations, CRF 44, §59.1, when the form and position of the flow paths is so radically uncertain that the risk of flooding at a place cannot be estimated through traditional procedures, a characteristic that frequently is associated with alluvial fans, this type of flooding is called alluvial fan flooding. Deviation from the traditional flood paradigm is further compounded on alluvial fans subject to debris flow hazard, that is, where the base flood is not caused by runoff but by a debris flow with triggering mechanisms, flow characteristics, and probability of occurrence that are completely different from those assumed in hydrological models of flood behavior.
Because the designation that an area is subject to alluvial fan flooding sets in motion specific, restrictive federal regulations, the determination by FEMA that an area is subject to alluvial fan flooding rather than ordinary flooding during the 100-year flood can be controversial. Thus, this report is in part an attempt to clarify what alluvial fan flooding means by providing a more precise definition and by describing how to apply the definition through the use of field indicators. The current chapter discusses this essential attribute of alluvial fan flooding in the section entitled "Origin of the Problem."
FEMA developed a procedure for estimating flood risk in environments subject to alluvial fan flooding (Dawdy, 1979; FEMA, 1990). In particular, the method predicts the extent of a fan-shaped area subject to a 1 percent chance of flooding in any year, as well as the average speed and depth of such a flood. Application of the alluvial fan flooding definition, the associated regulations, and the procedure for risk estimation together aroused considerable opposition from floodplain managers and other interested parties in some (but not all) communities that participate in the NFIP. This conflict led, eventually, to confusion and mutual suspicion.
Recognizing the need to resolve the conflict, FEMA requested the appointment of a National Research Council committee to study the issue of alluvial fan flooding. In particular, the committee was asked (1) to clarify, as necessary, the definition of alluvial fan flooding contained in section 59.1 of the NFIP regulations, (2) to specify criteria that can be used to determine whether an area is subject to alluvial fan flooding; and (3) to provide examples of applying the revised definition and criteria to real situations.
The Committee on Alluvial Fan Flooding met four times to study these issues and conduct a series of field visits during which it consulted with FEMA staff, floodplain management experts,
and local public officials (see Box 1-1). The committee's approach involved examining the hydrologic and geomorphologic processes that characterize flooding on alluvial fans in a range of varied environments. Understanding these processes from a natural science perspective provides the committee's basis for evaluating how current NFIP practice might be improved to better characterize flood hazards and to more accurately delineate zones of flood risk. The committee's revised definition of alluvial fan flooding is presented in this chapter. Later chapters provide an overview of flood and sedimentation processes on alluvial fans (Chapter 2), describe field indicators and methods to delineate hazard boundaries based on the revised definition (Chapter 3), and give examples of applying the field indicators to specific sites (Chapter 4). Chapter 5 presents the committee's conclusions and recommendations.
Because the establishment of this committee was requested by FEMA, that agency and its consultants are the primary audience for this report. However, the committee hopes that communities participating in the NFIP, other agencies, and floodplain management professionals in general will also appreciate this effort to better manage natural hazards in alluvial fan environments.
ORIGIN OF THE PROBLEM
Following a series of damaging floods in the southwestern part of the United States during the 1970s, FEMA sought a new approach to flood risk assessment in areas where flow paths are difficult to predict. Pictures of these floods are shown in FEMA Document 165 (FEMA, 1989), and the images are memorable (see Figure 1-1): water and debris flows along new paths not anticipated by planners and residents of the normally dry landscape, automobiles crushed by boulders, a house full of sand. These pictures depict a type of natural hazard different from ordinary riverine flooding. The hazard on active alluvial fans is less foreseeable, more difficult to control or resist, and more dangerous.
Many of the pictures in FEMA Document 165 are of flooding on alluvial fans. The term applied by the NFIP to this image was alluvial fan flooding, and the purpose of Document 165 was to explain how such floods occur. As our understanding changes of how floods occur, regulators develop new policies that eventually become formalized by the writing of regulations. Special rules were thus promulgated to regulate development and to set insurance premiums in areas subject to alluvial fan flooding. Conflict arose when Flood Insurance Rate Maps (FIRMs) prepared by contractors to FEMA were criticized by some participating communities. Most criticism focused on the underlying assumption (or "default assumption") in the FEMA procedure that flooding on alluvial fans is completely unpredictable, an assumption that is not always appropriate (French et al., 1993).
The Problem of Delineating Flood Hazards on Alluvial Fans
Faced with an increased need to map flood risk on alluvial fan areas in the 1970s and 1980s, and needing a method that can be applied at reasonable expense, FEMA adopted an analytical technique proposed by Dawdy (1979). The procedure uses a general mathematical formula (known as the conditional or total probability equation) to describe the probability of an
BOX 1-1 SELECTING THE SITE VISIT LOCATIONS
To respond to FEMA's charge, the committee gave a great deal of consideration to the geographic locations of its site visits. The committee had support to meet four times during the study process and with those four meetings decided to focus on geographic areas where the implementation of the NFIP to alluvial fans has been either particularly challenging or has met with moderate success. Many sites were considered and with only four opportunities to meet the committee certainly could not visit every location of interest. The experience of the committee's members and extensive use of the published literature of course expanded our knowledge base considerably, but there is special value to site visits because they allow an opportunity to talk directly with the people involved.
The first meeting was held in Arizona where FEMA has encountered notable resistance to its regulatory approach for alluvial fan areas. The second meeting was held in southern California where there is a wider degree of acceptance of FEMA's approach, at least as far as the mapping methods are concerned. The third meeting was held in southern Utah where FEMA was endeavoring to identify hazard areas on debris flow fans, an important issue. The fourth meeting was devoted primarily to writing this report but the committee brought in guests to talk about the difficulties of identifying flood hazards in the arroyos near Albuquerque, New Mexico.
Had there been more time, the committee of course would have liked to talk with experts carrying on important work at a number of other locations. These include the immense flood control facilities being constructed in Clark County, Nevada; the technical research being conducted by the Department of Energy at the Nevada Test Site and other federal lands; the hydrologic studies being performed as part of the proposed Yucca Mountain nuclear waste repository; the arroyos of western Texas; the composite fans in the Death Valley National Monument; the alluvial fan flood hazards in the Navajo Reservation; the large-scale alluvial fans along the Tahachapee Mountains in Kern County, California; urbanized fans such as in Wenatchee, Washington; and the effects of water-well induced ground subsidence on channel incision in the alluvial fans of central California.
event given a knowledge of some associated event. The procedure calculates the depth and velocity of the flood that has a 1 percent chance of occurrence at any point on a fan-shaped region, given a knowledge of the peak flow-frequency relationship at the apex of the fan. The procedure is based on only a few postulates and is attractive in the simplicity of both its conception and its implementation.
In its initial form (Dawdy, 1979), the procedure assumes:
That the peak flow-frequency has been estimated for the apex of the fan.
That the alluvial fan is shaped like a sector of a cone, all of which is subject to flooding, and that channels move across its surface at random during floods or from flood to flood (i.e., that their behavior is completely uncertain). The procedure ignores possible incision or stability of channels on alluvial fans. At any distance down the fan, the channel has an equal probability, over the long-term, of intersecting any part of a contour in a flood, and this probability is proportional to the ratio between the widths of the channel and the total width of the fan at that radial distance.
That individual floods remain in single channels in which the flow occurs at critical depth and velocity (i.e., velocity is proportional to the square root of depth) and has a width-to-depth ratio of 200.
Although the original outline of the procedure was based on the assumption of complete uncertainty about the behavior of channels, the recommendations for application were quite flexible (Dawdy, 1979) and left open the importation of other concepts and data to constrain the generally applicable probability theory (e.g., Mifflin, 1990). FEMA adopted this method in an appendix to its Guidelines and Specifications for Study Contractors, calling it, in early versions, ''Alluvial Fan Studies" (FEMA, 1985), and, in the latest version, "Studies of Alluvial Fan Flooding" (FEMA, 1995). The simplest form of this method was eventually codified into a computer program called FAN (FEMA, 1990), which likened alluvial fan flooding to rolling balls down a cone. In this form, the procedure can be followed by anyone, even with little or no knowledge of alluvial fans and their flooding characteristics. The FAN manual leads the practitioner through the procedure required to map zones of flood risk on the basis of only (1) a cursory identification that the site of interest lies on an alluvial fan, (2) measurement of the apical angle of the fan for computing its width at any radial distance, and (3) choice of a peak flow-frequency curve from regional data or a similar source.
When delineating flood hazard boundaries, the assumption of complete randomness in channel behavior may be relaxed if some field information is available that will allow the conditional probability equation to be solved with other constraints. But there does not appear to be any practical process, other than the review by consultants to FEMA, for deciding whether or which modifications should be applied in a particular circumstance.
Despite the elegance of its formulation, the FEMA procedure has been resisted in an important number of locations where it has been applied, particularly in those communities with the financial and technical resources to mount a challenge. The resistance arose for a number of reasons:
Misapplication of the procedure to locations that were not alluvial fans or to fans or portions of fans that are not subject to flooding.
Disparities between how floods occur on a particular fan and the assumptions of complete randomness.
Assumption that the hazard is dominated by rainfall-runoff without recognition of the debris flow hazard in flood insurance studies.
Too little investment being made in field identification of the conditions of flooding, leading to later huge expenditures for litigation and field surveys.
The mismatch between the rather inclusive actuarial goals of the original procedure and other traditional uses of the resulting FIRM, such as floodplain management or hazard mitigation.
A number of professionals in the field have taken exception to the formulation of FEMA's risk delineation procedure (e.g., French et al., 1993; Zhao and Mays, 1993), asserting, for example, that alluvial fans do not necessarily behave randomly and that floods are more likely to follow previous flow paths; that flows do not occur at critical depth; that the width-to-depth ratio is different from the assumed value of 200; or that the method, as adopted, is unrealistically simple and therefore easy to misapply. Although specific applications of the method may be deserving of such criticism, the original formulation of the problem (Dawdy, 1979, equation 6) is quite general and sound for calculation of a conditional risk. This same mathematical approach has been used by the U.S. Army Corps of Engineers to analyze the flood risk of a levee failure while dealing directly with the uncertainties inherent in such an occurrence (USACE, 1992, 1994). If the conditional probability equation is to be applied to alluvial fans, however, there is need for a more flexible and realistic approach to the definition of flow path uncertainty, best obtained perhaps, by field evidence for the nature and spatial distribution of processes. A more realistic, process-based approach to flood hazard delineation on alluvial fans remains a challenge to the technical community. However, a mathematical framework for moving from process to the delineation of risk zones has already been correctly set forth by FEMA. Further discussion of methods to delineate flood hazards on alluvial fans, a secondary charge of this committee, is discussed in Chapter 3.
If it is determined that a particular flooding source does not match with the default assumptions of FEMA's alluvial fan method (for example, when the potential for channel movement is not random) the default risk delineation method is inapplicable. However, this does not mean that the area is necessarily free from alluvial fan flooding as discussed next.
The Problem of Defining Alluvial Fan Flooding
The following definition is published in the NFIP regulation (section 59.1):
Alluvial fan flooding means flooding occurring on the surface of an alluvial fan or similar landform which originates at the apex and is characterized by high-velocity flows; active processes of erosion, sediment transport, and deposition; and unpredictable flow paths.
The primary purpose of this definition was to identify the existence of and extend jurisdiction over flooding situations that may have been excluded or improperly dealt with under prior regulations. According to the NFIP rules, alluvial fan flooding is a type of flooding that is recognized by characteristics that distinguish it from ordinary flooding. Almost everyone that the committee heard from held the view that there is a strong correlation, or even an exclusive relationship, between the term alluvial fan flooding and flooding that occurs on an alluvial fan. By this reasoning, if one determines that an area is not an alluvial fan, then it is not subject to alluvial fan flooding. The current definition, however, states that alluvial fan flooding can be present not only on alluvial fans but also on the rather ambiguous category of "similar landform[s]." Such reasoning leads to conflicts. For example, people who believe that alluvial fan flooding is flooding that occurs only on alluvial fans and who obtain technical advice that their community contains no alluvial fans will therefore conclude that there are no areas subject to
alluvial fan flooding in their community. FEMA, from its different perspective, might respond that within the community there are landforms similar to alluvial fans and that these experience alluvial fan flooding as explained in the regulatory definition. The community may counter that the similar landforms are inactive pediments and not alluvial fans; to which the response might be, "Just because you call something a different name, does not mean it is not subject to the severe hazards associated with alluvial fan flooding." The community's reply may then be that the amount of peril and the degree of uncertainty posed by flooding in its case are less than in other regions of the country that clearly have alluvial fans on which processes envisaged by FEMA are active and extensive. This hypothetical interchange typifies many of the discussions heard by the committee during the open, consultative portions of its meetings. There appeared to be a willful lack of reflection on the meaning of words, a general confusion about the diversity of flooding and sedimentation processes that occur on alluvial fans in a range of alluvial fan environments, and a lack of knowledge about information that could be quickly and inexpensively obtained through field examination of particular sites before hazard identification and delineation are carried out.
THE COMMITTEE'S RESPONSE
As a result of the committee's site inspections, the members' own field experience, consultations with many experienced individuals, including FEMA staff and its consultants, surveys of the literature, and extensive discussions at four meetings, the committee produced the following five products, which are presented in this report:
A revised definition of alluvial fan flooding, including criteria that can be applied by FEMA to identify those parts of alluvial fans that require special regulatory oversight to deal adequately with the uncertainty in flood processes. The committee also provides discussion of the constraints within which the alluvial fan flooding concept could be extended to other landforms. This definition is presented and analyzed in the latter part of Chapter 1. Although alluvial fan flooding is a general term that can involve flooding over an entire fan surface, the FEMA mandate is to determine the extent of flooding associated with a flood having a 100-year recurrence interval (i.e., a 1 percent probability in a given year). Hence, the term alluvial fan flooding is used in two ways. In the geomorphic sense, it can be any flood on an alluvial fan. But in the FEMA sense, it is the distribution of 100-year floodwater on the fan. The reader is cautioned that the term is used in both ways, including in this report.
A description of the flood and sedimentation processes that build alluvial fans and contribute to the existence of alluvial fan flooding (Chapter 2). The chapter also illustrates why the recognition of how floods occur is significant for risk assessment. An understanding of process issues is necessary for realistic and flexible regulatory practice and flood risk assessment in a range of environmental conditions. For example, the process approach to understanding what is distinctive about alluvial fan flooding at a particular location and the conclusion that both the process and the channel form on alluvial fans respond to the environmental history of the site allow a quick and simple mapping of process zones on alluvial fans in the manner demonstrated in Figure 1-2. Such a map, which can be made quickly and inexpensively using methods described in Chapter 3, outlines areas that require various forms of attention from the point of view of flood
risk assessment and provides guidance about how to deal with uncertainty in methods such as those of Dawdy (1979) and Mifflin (1990) and the procedures discussed at the end of Chapter 3.
A description of indicators that are used to assess whether criteria indicative of alluvial fan flooding are present and which allow an investigator to discriminate between those zones of a fan where flow paths are uncertain and other zones where flow path uncertainty is unimaginable in the current range of environmental conditions (Chapter 3). This type of evidence suggests ways of dealing directly with uncertainty in flood risk assessment by (a) indicating realistically those areas that are truly subject to flooding, and (b) showing the nature of the flooding phenomenon that is being assessed. This chapter also illustrates how the alluvial fan flooding problem can be broken down into a set of questions for which there is already an established body of scientific material (e.g., channel bank stability) (USACE, 1992).
A demonstration of the use of the committee's definition of alluvial fan flooding for specific locations (Chapter 4), either field sites that were visited by committee members or sites described in detail in the scientific literature. The examples illustrate that various levels of effort yield answers of varying detail, but that even a one-day field examination can yield valuable insights for assessing flood hazard zones.
A framework that suggests the appropriate direction to advance our ability to delineate more accurately those parts of an alluvial fan that are subject to flooding by 100-year flood by dealing directly with flood process uncertainty (Chapter 3).
THE NFIP DEFINITION OF ALLUVIAL FAN FLOODING
As noted earlier, the NFIP defines alluvial fan flooding as "flooding occurring on the surface of an alluvial fan or similar landform which originates at the apex and is characterized by high-velocity flows; active processes of erosion, sediment transport, and deposition; and unpredictable flow paths."
This definition emphasizes a type of flooding, not a landform, and thus is inherently difficult to translate into the regulatory setting. Defining the hazard more explicitly in process terms emphasizes that a variety of flooding processes with varying distributions and levels of intensity occur on alluvial fans; because of the range of environmental conditions in which such floods occur, a degree of flexibility is needed in defining and quantifying them. For emphasis and elaboration, the primary elements of the current NFIP definition are paraphrased here:
Alluvial fan flooding has an unpredictable flow path. The perceived channel, if there is one, may not be the actual conveyance route for water during a flood.
Alluvial fan flooding occurs on the surface of an alluvial fan or similar landform for which the spatial domain that is subject to flooding may extend over a larger area than the floodplain as determined by the traditional hydrologic paradigm.
Alluvial fan flooding has velocities high enough to erode new channels for floodwaters. Similarly, such erosion may undermine adjacent buildings and destroy them even though the water does not get deep enough to cause inundation.
Alluvial fan flooding transports large volumes of sediment, the deposition of which may influence both the location and the direction of flowing water during a single flood event. This phenomenon is part of the reason for the "unpredictability" of the flow path.
The word "unpredictable" is troublesome in this context, and should be read only in a relative manner. There is always uncertainty associated with the prediction of how floods occur. Thus, in a sense, all flooding is "unpredictable." In the case of riverine flooding, which presumes a stable-bed condition, however, we can set aside this uncertainty because there are established procedures to predict how such floods occur. In the case of the alluvial fan to which the FAN computer program (FEMA, 1990) is applicable, we simplify the uncertainty by assuming that flow path behavior is random and a flood has no greater or less chance of following an established flow path than it has of cutting through the neighbor's backyard. The implication of the definition is that for alluvial fan flooding the flow path behavior is so indeterminate that we cannot set aside the uncertainty and still achieve a realistic assessment of the flood risk.
The existing regulatory definition does not describe an alluvial fan but rather a type of flooding that may also occur in nonalluvial fan areas (see Figure 1-3). For example, sediment movement may significantly affect floodflow behavior in river delta areas, and flow velocities high enough to cause erosion and deposition are common in alluvial river floodplains. After the 1993 Missouri River flood, sand deposits of 2 feet depth or greater covered 60,000 acres of adjacent farmland, causing damages in excess of more than $100 million (Interagency Floodplain Management Review Committee, 1994). On the other hand, there are alluvial fans that have well-defined channels and may not be subject to alluvial fan flooding as defined above. The choice of the term is responsible, in large measure, for the confusion surrounding the definition issue. As it stands, the existing definition is vague and potentially very inclusive.
During the hazard identification process, the question is asked whether an area is subject to alluvial fan flooding. The answer affects both how flood zones are delineated (FEMA, 1995) and the rules that apply. Although the way Special Flood Hazard Areas (SFHAs) are delineated for alluvial fan flooding may differ from that for ordinary flooding, the regulatory importance of alluvial fan flooding is realized during mitigation. Section 65.13 of the NFIP regulations precludes the removal of a SFHA based solely on the elevation of a structure above the estimated flood stage or the placement of fill that creates such a condition. It also sets forth standards for structural flood control measures to remove the zone designation by mitigating the flood hazard.
Through discussion with representatives of FEMA and the review of various documents, the committee has identified several regulatory difficulties that originate from using the riverine flooding paradigm in alluvial fan flooding situations:
Issues related to Letters of Map Amendment (LOMAs). A LOMA is issued by FEMA when a parcel of land is inadvertently included in the SFHA because of the limitations of map scale or topographic data. By certifying that the finish floor of a structure is higher than the adjacent base flood elevation (BFE), the parcel can be removed from the SFHA. For areas subject to debris flows, extreme deposition, or shifting of flow channels, such an approach may not be appropriate however, because the hazard still exists. Because LOMA applicants are allowed to contact FEMA directly, the agency has no way of determining whether a LOMA should be granted or not apart from the alluvial fan flooding designation.
Issues related to Letters of Map Revision (LOMRs) based on fill. A LOMR based on fill is issued when a parcel of land is raised by adding layers of soil so that the finish floor of a building is higher than the adjacent BFE. Since applicants for this type of LOMR can also contact FEMA directly, the same concerns exist as for a LOMA, that is, the processes of erosion and deposition and the impact force of debris, may still result in damage to the building.
The perceived flow path issue. Because of the convex cross-slope of many alluvial fans, floodwaters leaving the historical or perceived channel may follow a new direction and inundate areas distant from the channel. The alluvial fan flooding designation enables the inclusion of these
distant areas by consideration of the scenario where all or part of the flow leaves the main channel.
The sandbar issue. For a wide, shallow wash, a base flood elevation determined using backwater conveyance calculations such as those performed by computer program HEC-2 (Hydrologic Engineering Center, 1990) might indicate local areas within the wash that are higher than the computed water surface and could therefore be shown outside the SFHA. This, according to FEMA, could allow someone to build a house on a sandbar in the middle of a braided alluvial wash. The alluvial fan flooding designation discourages this by including the sandbar within the area subject to flooding and disallowing the removal of the designation until some type of reliable mitigation is implemented.
The split flow issue. Because of variations in channel cross section shape during an event or because of blockage by sediment, a network of channels that progressively splits into smaller channels may not reliably distribute the flow of water and sediment in such a way that all of the channels remain stable and protect the higher interchannel areas from flooding. For the case where the channel network fails, the alluvial fan flooding designation enables one to identify on the FIRM that the interchannel areas are at risk of being flooded even though they might be a meter or more higher than an adjacent channel.
Because these regulatory difficulties exemplify specific weaknesses in the riverine flooding paradigm, they portray some of the essential elements that make alluvial fan flooding a distinct type of flooding and they will help formulate the committee's revised definition.
IMPLICATIONS OF ALLUVIAL FAN FLOODING AS A DISTINCT TYPE OF FLOODING
Because of its character, alluvial fan flooding offers particular challenges to floodplain managers and regulators. Figure 1-4 compares the current paradigm that governs the analysis of ordinary riverine flooding, as viewed by the NFIP, to alluvial fan flooding. It contrasts the flood risk on two surfaces, labeled 1 and 2 in each case. The regulations require the identification of areas that have a 1 percent chance of being flooded in any given year. This process starts by the development of a graph showing peak flood discharge from the watershed plotted versus its recurrence interval. For most alluvial fan source areas, this step is tenuous at best and may be completely illusory in the cases where the events of significance are debris flows or where recent or frequent disturbance by fire makes it futile to view the watershed merely as a generator of independent rainfall-runoff events. However, this aspect of the uncertainty is not as controversial, perhaps because the methodology is widely used.
After discharge is specified, the riverine approach and alluvial fan flooding approach to analysis and risk prediction diverge. In the riverine case (Figure 1-4a), location of the flood within the spatial domain is assumed to be along the perceived or historical flow path, which is recognized as the main channel (Figure 1-4a(ii)). Much work is then necessary to identify the flood hazard by determining the relationship between depth and discharge (Figure 1-4 a(iii)). This is usually done using Manning's equation or a similar technique embedded within a step-backwater computer model. Finally, the flood hazard is delineated via the water surface elevation,
thus establishing a relationship between inundation depth above a point (such as the finished floor of a house) and the recurrence interval. Figure 1-4a(iv) shows the cases for the main channel and the higher surface (labeled 2 on the cross section), which is often called the overbank area or the floodway fringe. The stage recurrence graph developed in this manner always has some uncertainty, but it is usually not presented. The riverine approach provides a clear method that allows us to communicate about flooding and to make reproducible calculations of its severity. Over time it has become widely accepted and its weaknesses seldom questioned.
Figure 1-4b shows the analogous components of an analytical approach to alluvial fan flooding. First, the discharge recurrence relationship is estimated for the apex of the fan. Second, based on the knowledge that the perceived or historical flow path may not convey all of the water during a flood, it is assumed that the actual flow path has no greater chance of occupying the perceived channel than it does of straying to any location on the fan. This default assumption is shown in Figure 1-4(ii). The relationship between depth and discharge is then determined using the method proposed by Dawdy (1979) (although a case can be made for altering this step by using the process-based knowledge described in Chapter 3 and an alternative solution to the conditional probability). Finally, the inundation depth and velocity are delineated based on the assumption that the entire fan surface is subject to flooding. The predicted degrees of flood hazard for the surfaces labeled 1 and 2 are identical because the procedure knows nothing of the differences between them. Real floods on alluvial fans are, of course, much more complex than this.
An important implication of this approach to the prediction of flood risk on alluvial fans is illustrated in Figure 1-5, which portrays a flood-prone surface with three distinct elevations, labeled 1, 2 (the main, recently occupied channel), and 3. In the traditional riverine flooding paradigm (Figure 1-5a) the historical channel is the main conveyor of the base flood and the computation of the water surface in "overbank" areas is based on the conveyance capacity of a single cross section that includes surface 2. This approach shows surface 3 as "wet" merely as a consequence of surface 2 being too small to convey the entire flood. Surface 1 is above the computed base flood elevation and could therefore be shown as outside of the 100-year floodplain on the FIRM. In such areas, we imply that the probability density function that describes flow path location within the lateral domain is narrow and strongly peaked, that is, that all of the flow behaves hydraulically as a single channel contained within a relatively narrow zone that does not shift during the event.
Figure 1-5b portrays the case of alluvial fan flooding where, during the base flood event, the channel might separate into two branches upstream of the cross-section, allowing a flow path to develop that invades the higher surface 1. After such a flow split occurs upstream, surface 1 may be flooded even during events smaller than the 100-year event depending on the specific behavior of a real sequence of floods. If the alluvial fan flooding paradigm is applied to the situation (Figure 1-5b), surface 1 is treated as a separate flow path, which it may well become during an actual flood. The corresponding probability density function for this situation shows three peaks indicating that each of surfaces 1, 2, and 3 have a finite chance of conveying water during a flood. Based on this potential, multiple scenarios are analyzed that represent the possible distribution of flows on the three surfaces. From this information, a separate base flood elevation is computed for each surface via the conditional probability equation. The FIRM in this case would show that surface 1 is indeed subject to flooding. (Note: The probability density functions in these figures are shown to illustrate the difference between the two flooding perspectives. Showing the
general case, where a probability density function may be defined at each point of interest rather than for a cross section, would unnecessarily complicate the drawing of Figure 1-5).
The alluvial fan flooding approach (Figure 1-5b) is a more realistic way of depicting the floodplain boundaries of even a river system to a large flood because the riverine flooding approach (Figure 1-5a) relies on maximizing conveyance in a preflood cross section to determine which surfaces are flooded and ignores the possibility that a surface can be flooded due to the redirection of flows at an upstream point. Thus, a key difference between alluvial fan flooding and riverine flooding is the implicit assumption of spatial consistency of the depth discharge relationship between adjacent surfaces. Based on physical processes, the alluvial fan flooding approach is the superior way for viewing how floods occur in general, and the riverine flooding approach is a special case where nothing out of the ordinary is going on. The question raised by Figure 1-5 is whether the presence of uncertainty in flood processes by itself constitutes alluvial fan flooding because failure to deal with channel changes by modeling scour, fill, and lateral movement (as the riverine flooding approach fails to do) results in grossly inaccurate delineation of flood hazard boundaries.
A previous National Research Council committee (NRC, 1983) concerned with the effects of process uncertainty in alluvial rivers was faced with the question of whether flood studies should use riverbed mobility models rather than fixed-bed models. Its conclusion was that the uncertainty introduced by ignoring the effects of sediment degradation/aggradation was no greater than the additional uncertainty introduced by the use of mathematical techniques. That committee's conclusion was that evaluating the effect of parameter uncertainty (i.e., the variation in channel roughness, geometry, and slope) was a suitable way to deal directly with process uncertainty in alluvial rivers. In other words, when executing the riverine flood paradigm it is often better to identify the potential impacts of error upon predicting the behavior of real floods than to strive for a more realistic approach in the hope that this error will go away. Simple approaches often yield satisfactory results. For example, the Flood Insurance Study for the City of Palmdale (FEMA, 1987, p. 8) recognizes this and contains the following counsel:
Average depths of flooding were assigned based on standard hydraulic calculations through irregular cross sections. In many cases, the assigned average depth is not representative of the true degree of flood hazard. This situation occurs where the average depths are based on a wide cross section which encompasses one or more low flow drainage courses. The actual depth of flooding and, consequently, the true flood hazard will be greater adjacent to the drainage course.
The intensity of flood and sedimentation processes on many actively accumulating parts of alluvial fans is much greater and the frequency, magnitude, and suddenness of channel changes are more severe, however, than envisioned by the 1983 committee. In these more complex cases, the problem of channel location and flow distribution can be guided by the kind of process-based understanding, field evidence, and analysis techniques outlined in Chapters 2 and 3.
IMPLICATIONS OF ALLUVIAL FAN FLOODING FOR FLOODPLAIN MANAGEMENT
The nature of alluvial fan flooding as currently defined by the NFIP has implications for both floodplain management and the mitigation of flood hazards. Consider the situation in Figure 1-5, which has been redrawn in Figure 1-6 to illustrate the implications of using fill to raise the level of surface 1. This surface is already above the computed base flood elevation, but it is clear that if one accounts for some degree of error, surface 1 could be completely inundated because it is protected only by a small berm in the natural topography. If ordinary riverine flooding is the appropriate paradigm to apply to the site (Figure 1-6a), filling surface 1 has no effect whatsoever on the computed bankfull flood elevation for surface 2 (the main channel). This would therefore be an acceptable strategy to reassure the concerned parties that the area is protected from flooding.
Viewed from the perspective of alluvial fan flooding (Figure 1-6b), however, where surfaces 1, 2, and 3 each have a chance of conveying all or part of the flood, protecting surface 1 in this manner would not be an allowable mitigation strategy. The reason is that eliminating surface 1 as a potential flow path, increases (theoretically) the frequency with which surfaces 2 and 3 get flooded. This is equivalent to increasing the base flood elevation for the n-year event and could therefore be an infraction of NFIP regulations (e.g., section 65.12). Similar actions that are generally considered to be good floodplain management practice would also come under question. For example, reinforcing a levee reduces the uncertainty about its failure potential during a flood and better protects areas behind it. But eliminating the flow path through a breach in the previously substandard levee could increase both the computed stage-frequency curve and the chance for other failures further downstream. These conclusions illustrate the disconnect between the alluvial fan flooding and the riverine flooding paradigms in the context of floodplain management.
For risk assessment under alluvial fan flooding, existing channels cannot be relied on to convey the 100-year peak flow, so their role is ignored. For riverine floodplain management, however, the channels are significant. They convey the smaller flood events, they indicate how floods have occurred in the past, and they define where future facilities may be located. The default assumption of a uniform risk (FEMA, 1995) or complete uncertainty across an alluvial fan is a formalized guess that allows one to delineate risk on the Flood Insurance Rate Map using a straightforward technique. A FIRM showing alluvial fan flooding hazards mapped in this manner is an expression of uncertainty or the absence of knowledge about floods, however, rather than an indication of how one might actually occur. Unlike a riverine FIRM, an alluvial fan flooding FIRM is of limited use for mitigation and management of flood hazards. By making a conservative trade-off in favor of what might happen, this type of FIRM ignores the importance of what has happened. If the uniform-risk FIRM is interpreted literally, then it can be argued using formal mathematics of the kind that underlies the existing FEMA procedure that any mitigation effort, short of complete channelization, increases the flood risk on another part of the fan. Thus floodplain managers are left with the peculiar responsibility of preserving uncertainty.
THE COMMITTEE'S DEFINITION OF ALLUVIAL FAN FLOODING
This committee was asked to provide a new definition of alluvial fan flooding. Ideally, this definition could be used to divide flood hazard areas into two categories: those subject to alluvial fan flooding and those not. Of course, not all cases will fall clearly into one category or another. Although it is traditional to be conservative in the interest of safety, wholesale adoption of the alluvial fan flooding paradigm may not be a good thing because this approach brings its own set of weaknesses and regulatory traps. Furthermore, such blind caution could result in hundreds of cases where flooding sources that have been successfully mapped and managed as ordinary rivers will need to be completely reassessed for the sole purpose of consistently applying the definition. This would be unnecessary in many cases and could, ironically, undermine the original motivation for creating a new category of flood hazard for those cases that would otherwise be dealt with inadequately.
Because alluvial fans are the place where there is a strong historical connection to the breakdown of the riverine flooding paradigm, these landforms provide a regulatory partition that allows FEMA to concentrate on the most serious cases. This committee has therefore chosen to restrict the term alluvial fan flooding to apply only for alluvial fans. Floods with characteristics that fit the alluvial fan flooding concept but occur in nonalluvial fan environments are discussed as a separate, broader category of flooding.
The Committee on Alluvial Fan Flooding proposes the following definition, which considers the objectives of the original NFIP definition, the administrative concerns of the NFIP, and the criteria necessary to establish 100-year recurrence interval alluvial fan flooding as a distinct hazard:
Alluvial fan flooding is a type of flood hazard that occurs only on alluvial fans. It is characterized by flow path uncertainty so great that this uncertainty cannot be set aside in realistic assessments of flood risk or in the reliable mitigation of the hazard. An alluvial fan flooding hazard is indicated by three related criteria: (a) flow path uncertainty below the hydrographic apex, (b) abrupt deposition and ensuing erosion of sediment as a stream or debris flow loses its competence to carry material eroded from a steeper, upstream source area, and (c) an environment where the combination of sediment availability,slope, and topography creates an ultrahazardous condition for which elevation on fill will not reliably mitigate the risk (Figure 1-7).
Alluvial fan flooding begins to occur at the hydrographic apex, which is the highest point where flow is last confined, and then spreads out as sheetflood, debris slurries, or in multiple channels along paths that are uncertain. The hydrographic apex may be at or downstream of the topographic apex and may change during a flood event due to deposition or erosion. Such flooding is characterized by sufficient energy to carry coarse sediment at shallow flow depths. The abrupt deposition of this sediment or debris strongly influences hydraulic conditions during the event and may allow higher flows to initiate new, distinct flow paths of uncertain direction. Also, erosion strongly influences hydraulic conditions when floodflows enlarge the area subject to flooding by undermining channel banks or eroding new paths across the unconsolidated sediments of the alluvial fan. Flow path uncertainty is aggravated by the absence of topographic
confinement and by the occurrence of erosion and deposition processes. Together, these characteristics create a flood hazard that can be reliably mitigated only by the use of major structural flood control measures or by complete avoidance of the affected area.
The potential for erosion and deposition, the related uncertainty in flow path behavior, and the imprudence of elevation on fill as a mitigation measure are joint and separate characteristics shared among many flood hazards on depositional environments other than alluvial fans, although not usually with the same intensity. It stands to reason that some of the same rules that apply to alluvial fan flooding should apply to this more inclusive type of flood hazard, termed uncertain flow path flooding. Flood hazards that meet only one or two of the criteria in the definition make up this third category.
To apply any definition in a regulatory context, the definition must be supported by criteria that can serve as standards, principles, or tests. These criteria, as reflected in indicators such as data, measurements, field evidence, and observations, are what floodplain managers and
regulators use to apply the definition to a given situation. Erosion and deposition processes are essential criteria in judging alluvial fan flooding because they may affect hydraulic conditions. Where localized sediment deposition, bed form translation, and erosion produce changes in streambed elevation during an event that approaches typical depths of flow, the uncertainty introduced by these processes is significant. Consequently, floodwater surface elevations computed using preflood topography are not a comprehensive indicator of the true hazard for alluvial fan flooding situations as they are for riverine flooding.
Flow path uncertainty, which means that the perceived, historical channel or network of channels cannot be relied on to convey the base flood, affects the spatial domain subject to flooding through the creation of new flow paths and/or the subdivision of flows into multiple distinct paths as shown in Figure 1-5b. For a network of channels that exist prior to a flood event, uncertainty in the distribution of flows through the network must be considered in order to rule it out as a factor that might cause flow path uncertainty.
The criteria above frequently create a situation where the traditional formulas for mitigation, such as elevating a structure on fill, do not actually eliminate the hazardous condition. The unsuitability of fill as a hazard reduction strategy, however, is perhaps the most important characteristic distinguishing between riverine flooding and alluvial fan flooding.
In summary, the committee's revised definition limits alluvial fan flooding to flood hazard on alluvial fans. The committee recognizes that alluvial fan flooding is one type of flood hazard under the wider category of uncertain flow path flooding. Such hazards may have considerable uncertainty associated with their behavior and require means other than fill for reliable mitigation. Chapter 4 presents examples that illustrate how the definition applies to specific cases.
American Geological Institute. 1987. Glossary of Geology. 3rd Ed. R. L. Bates and J. A. Jackson, eds. Alexandria, Va.: American Geological Institute.
Bedient, P. B., and W. C. Huber. 1992. Hydrology and Floodplain Analysis. 2nd Ed. Reading, Mass.: Addison Wesley.
Dawdy, D. R. 1979. Flood frequency estimates on alluvial fans. American Society of Civil Engineers Journal of Hydraulics Division, 105(HY11):1407–1413.
Federal Emergency Management Agency (FEMA). 1985. Appendix 4: Alluvial fan studies. Guidelines and Specifications for Study Contractors. Doc. no. 37. Washington, D.C.: FEMA.
Federal Emergency Management Agency (FEMA). 1987. Flood Insurance Study, City of Palmdale, California, Los Angeles County . Washington, D.C.: FEMA.
Federal Emergency Management Agency (FEMA). 1989. Alluvial Fans: Hazards and Management. Doc. no. 165. Washington, D.C.: FEMA.
Federal Emergency Management Agency (FEMA). 1990. FAN: An Alluvial Fan Flooding Computer Program, User's Manual and Program Disk. Washington, D.C.: FEMA.
Federal Emergency Management Agency (FEMA). 1995. Appendix 5: Studies of alluvial fan flooding. Guidelines and Specifications for Study Contractors. Doc. no. 37. Washington, D.C.: FEMA.
French, R. H., J. E. Fuller, and S. Waters. 1993. Alluvial fan: Proposed new process-oriented definitions for arid southwest. American Society of Civil Engineers Journal of Water Resources Planning and Management 119(5):588–598.
Hydrologic Engineering Center (HEC). 1976. Water Surface Profiles: Hydrologic Engineering Methods for Water Resources Development, vol. 6. Davis, Calif.: U.S. Army Corps of Engineers Water Resources Support Center.
Hydrologic Engineering Center (HEC). 1990. HEC-2 Water Surface Profiles, User's Manual. Davis, Calif.: U.S. Army Corps of Engineers Water Resources Support Center.
Interagency Floodplain Management Review Committee. 1994. Sharing the Challenge: Floodplain Management Into the 21st Century. Washington, D.C.
Mifflin, E. R. 1990. Entrenched channels and alluvial fan flooding. Pp. 28–33 in Proceedings of the American Society of Civil Engineers (ASCE) International Symposium of Hydraulics and Hydrology of Arid Lands. New York: ASCE.
National Research Council. 1983. Evaluation of Flood-Level Prediction Using Alluvial-River Models. Washington, D.C.: National Academy Press.
U.S. Army Corps of Engineers (USACE). 1992. Guidelines for Risk and Uncertainty Analysis in Water Resources Planning. Report. 92-R-1. Fort Belvoir, Va.: Water Resources Support Center.
U.S. Army Corps of Engineers (USACE). 1994. Risk Based Analysis for Evaluation of Hydrology/Hydraulics and Economics in Flood Damage Reduction Studies. Engineering Circ. EC-1105-2-205. Washington, D.C.: USACE.
Zhao, B., and L. W. Mays. 1993. Uncertainty analysis of the FEMA method for alluvial fans, Pp. 2098–2103 in Proceedings of the Annual Conference of the Hydraulics Division, Hydraulic Engineering, 1993. New York: American Society of Civil Engineers.