Modernizing Probable Maximum Precipitation Estimation (2024)
Chapter: Summary
Summary
For more than 75 years, high-hazard structures in the United States, including dams and nuclear power plants, have been engineered to withstand floods resulting from the most unlikely but possible precipitation, termed Probable Maximum Precipitation (PMP). More than 16,000 high-hazard dams and 50 nuclear power plants are located in the United States, many of which are approaching or exceeding their design lifetime. Failure of any one of these structures will likely result in loss of life and could impose significant economic losses and widespread environmental damage. The pressures of climate change on flood hazards further highlight the urgent need to re-assess the safety of and flood protection provided by structures designed decades ago.
The scientific and engineering foundations of PMP are old. The key ideas underlying PMP were developed by the Miami Conservancy District more than a century ago to address the catastrophic impacts of the Great Flood of 1913 in the Upper Ohio River. The rapidly accelerating pace of dam building in the United States led to the standardization of PMP procedures by federal agencies in the 1940s. PMP informed rational engineering solutions for the U.S. water and power infrastructure to diminish risks of flood hazards. However, weaknesses in the scientific foundations of PMP, combined with advances in understanding, observing, and modeling extreme storms, call for fundamental changes to the definition of PMP and the methods used to estimate it.
Although they have changed over time, definitions of PMP have always been based on the assumption that rainfall is bounded. The National Oceanic and Atmospheric Administration (NOAA) has defined PMP as “theoretically, the greatest depth of precipitation for a given duration that is physically possible over a given size storm area at a particular geographical location at a certain time of year.” A compelling case for the existence of upper bounds on rainfall, however, has yet to emerge, either through physical arguments or statistical analyses. PMP is defined as an upper bound on rainfall, and thus as a value that cannot be exceeded. Yet, PMP estimates are based
on limited observations and subjective estimation procedures—they can be and have been exceeded in the past.
THE NEED TO MODERNIZE PMP AND ITS ESTIMATION
Current PMP estimation methodologies are based on observations contained in storm catalogs and are grounded on three fundamental components: storm transposition, moisture maximization, and orographic adjustment. Each component has significant limitations as outlined below.
Storm catalogs containing observations of rainfall from extreme storms provide the data used to estimate PMP. Rainfall observations from non-standard rain gauges obtained from bucket surveys play a critical role in PMP estimation. U.S. and world record rainfall accumulations from bucket surveys include 305 mm in 42 minutes from the Holt, Missouri, storm on 22 June 1947; 560 mm in 2.75 hours from the D’Hanis, Texas, storm on 31 May 1935; and 780 mm in 4.5 hours from the Smethport, Pennsylvania, storm on 18−19 July 1942. A cattle trough was the instrument used for the D’Hanis measurement, and a mason jar was the instrument used for the Smethport measurement. Bucket surveys have also provided the observations needed to effectively assess the “given storm area” requirement of the PMP definition. For the Smethport storm, more than 400 bucket survey rainfall observations were obtained around the area of peak rainfall; the region had no conventional rain gauges.
It is not entirely fortuitous that world record rainfall accumulations were obtained in 1935, 1942 and 1947. The era of flood studies with carefully developed bucket surveys of extreme rainfall waxed and waned as the priorities of the federal dam building program changed. A fundamental limitation of storm catalogs used for PMP estimation is the incomplete temporal and spatial sampling of storms. Storm catalog records are incomplete, and it is not possible to detail what is missing, either in time or space. This limitation precludes statistical characterization of uncertainty of current PMP estimates.
The most important component of PMP estimation based on storm catalogs is storm transposition, which aims to specify—for each storm—the geographic region over which it could be transported and thus be used to estimate PMP. Specification of storm transposition regions relies on the scientific judgment of PMP practitioners and is therefore inherently subjective. Storm transposition, more than any other component of PMP estimation, depends on the effective application of scientific understanding of extreme storms. Previous studies have shown that PMP estimates are generally more sensitive to storm transposition than to any other component of PMP estimation. For example, PMP estimates for much of the eastern United States are strongly dependent on decisions specifying the storm transposition region for the July 1942 Smethport storm. Similar patterns of sensitivity to storm transposition hold for regions across the United States.
Moisture maximization is a procedure used to amplify rainfall observations from the storm catalog events in an attempt to reflect the maximum rainfall that could occur from similar storms but under even more extreme moisture conditions. An underlying assumption is that rainfall varies linearly with precipitable water, which is the total water
vapor content in an atmospheric column. The moisture maximization factor used to amplify storm rainfall is the ratio of maximum precipitable water at the storm location to the actual precipitable water for the storm. Modeling and observational studies have not provided support for the assumptions underlying moisture maximization. Significant challenges in developing the data needed to implement moisture maximization have also been noted. Moisture maximization provides a plausible engineering safety factor for PMP estimation, but it lacks a solid scientific and observational foundation.
The challenges to estimating PMP in mountainous terrain were highlighted in the 1994 National Academies study Estimating Bounds on Extreme Precipitation Events: A Brief Assessment (NRC, 1994) and remain largely unresolved. The principal tools used to address terrain effects on extreme rainfall center are orographic transposition factors, which amplify or decrease observed storm rainfall based on precipitation frequency products. In transposing a storm from point A to point B, an orographic transposition factor is computed as the ratio of T-year rainfall at point B to T-year rainfall at point A (the return interval T is often taken to be 100 years). Orographic transposition factors nudge PMP estimates toward the spatial patterns of precipitation frequency maps. They address the long-recognized challenge of estimating PMP in mountainous terrain where observations are often severely limited, but they do not adequately address the scientific challenges imposed by orographic precipitation mechanisms. Observational, modeling, and theoretical advances are required to effectively estimate PMP in mountainous regions.
The principal weaknesses of current PMP methods are listed below. They reflect many of the issues identified in the National Academies 1994 study (NRC, 2014).
- The assumption that rainfall is bounded
- The absence of procedures to account for the effects of climate change on rainfall extremes
- The incomplete temporal and spatial sampling of extreme rainfall events in storm catalogs
- The inherently subjective implementation of storm transposition procedures
- The absence of a sound scientific foundation for moisture maximization
- The empirical correction factors used to account for the effects of complex terrain on extreme rainfall
- The absence of procedures to account for statistical uncertainty of PMP estimates
A VISION FOR PMP
Given these limitations and the importance of ensuring the safety of our critical infrastructure, the concept of PMP and its estimation methodology must be revisited. This report presents the committee’s conclusions and recommendations, including a revised definition of PMP, near-term enhancements to PMP estimation, and a transition to a long-term approach to PMP estimation that is based on computer simulations using physics-based climate models, which would facilitate the effective treatment of climate change effects on extreme precipitation and the characterization of uncertainty of PMP estimates. The committee’s vision is as follows:
Model-based probabilistic estimates of extremely low exceedance probability precipitation depths under current and future climates will be attainable at space and time scales relevant for design and safety analysis of critical infrastructure within the next decade.
Toward this vision, this report presents a set of recommendations that are summarized below and expanded in the chapters that follow.
A NEW DEFINITION OF PMP
The committee recommends revising the definition of PMP to become “the depth of precipitation for a particular duration, location and areal extent, such as a drainage basin, with an extremely low annual probability of being exceeded, for a specified climate period.” Federal and state agencies, in partnership with state dam safety officials, would develop national guidelines for specifying the annual exceedance probability (AEP) as detailed in recommendations below. The proposed long-term methodology for PMP estimation is based on statistical analysis of long-term simulated rainfall fields from high-fidelity and high-resolution storm-resolving climate models (model-based PMP estimates). This model-based approach permits incorporation of advances in physical understanding and numerical modeling of extreme storms, the effects of climate change, and uncertainty characterization of PMP estimates.
The revised definition of PMP differs from the previous one in two primary ways: (1) it replaces an “upper bound” on rainfall with an “extremely low exceedance probability” and (2) adds “for a specified climate period” so that PMP estimates can change with climate. The revised definition addresses the two most critical weaknesses of current PMP methods: the assumption that rainfall is bounded does not provide a tenable foundation for estimation of PMP, and climate change has resulted in historical changes in extreme rainfall and will likely cause even greater changes over the coming decades. These changes are essential for developing scientifically grounded methods for estimating PMP. “Time of year” is omitted from the revised definition because the model-based approach can readily provide seasonally varying estimates in settings where they are useful.
Specification of the AEPs that define PMP presents a challenging societal question regarding the level of risk judged to be acceptable for high-hazard dams and nuclear power plants that still assures their safety. Rough assessments of the AEPs corresponding to current PMP estimates are on the order of 10-4 to 10-7. However, if the AEP were set to 10-4 for all high-hazard dams in the United States, roughly two dams per year on average would be subject to catastrophic failure, a rate that would likely prove societally unacceptable. The committee recommends that federal and state agencies, in partnership with the Association of State Dam Safety Officials (ASDSO), develop national guidance for specifying AEPs used for PMP estimation.
As challenging as it is, specifying the AEPs of PMP is only one step in modern dam and nuclear safety programs. Another crucial step and a key element of Risk-Informed Decision Making (RIDM) is quantitative assessment of the uncertainty in PMP estimates. The recommended model-based method for PMP estimation provides a natural
path for developing statistical estimates of uncertainty, enabling the development of objective, robust, and site-specific approaches for risk analysis and decision making. The dam and nuclear safety communities have developed procedures for integrating uncertainty characterization into RIDM-based safety programs.
Another consideration in dam and nuclear safety is the central role of the Probable Maximum Flood (PMF), which is largely, but not entirely, dependent on PMP. The PMF provides a design hydrograph at the outlet of a drainage basin specified by the location of the dam or nuclear power plant. Tailoring PMP estimates to a specific drainage basin has long been recognized as significantly challenging. The proposed model-based approach provides the capability to develop PMP estimates that are naturally linked to PMF estimation over drainage basins. Temporal and spatial patterns of rainfall over drainage basins, as well as antecedent basin conditions, can be readily provided. The detailed methods used to compute PMF are beyond the scope of this study, but the linkages between PMP and PMF are important to consider in pursuing methods for modernizing PMP estimation.
Use of an extremely low rainfall AEP poses distinctive scientific challenges. One of the most daunting challenges arises from the contrast between record lengths of historical observations (~101 to 102 years) and the return intervals of PMP storms (~104 to 107 years). This problem is amplified by the rapid pace of climate change over the period of historical observations. Modernization of PMP estimation will require innovative and synergistic development of observational, statistical, and modeling tools that focus on the rainfall extremes that define PMP.
The observational and modeling challenges for PMP estimation are most pronounced for small-area, short-duration convective rainfall, as noted in the 1994 PMP study. The practical importance of this is the fact that half of the high-hazard dams in the United States have watersheds with drainage areas less than 20 km2. Furthermore, the impacts of climate change on extreme precipitation are arguably most difficult to assess for small-area, short-duration storms.
PATH TO MODERNIZING PMP ESTIMATION
The path toward implementation of model-based PMP estimation is impeded by two significant challenges to the development of kilometer-scale or finer resolution models necessary to resolve storms that produce PMP-magnitude precipitation. First, increased model resolution is not a sufficient guarantee that models that will be fit-for-purpose, because storm-resolving simulations are sensitive to parameterized processes such as cloud microphysics and boundary layer turbulence. Second, significant computational resources are needed to produce large ensembles of storm-resolving simulations to address model uncertainty and internal variability. The committee proposes a phased approach to addressing these challenges, whereby near-term enhancements to current PMP methods based on observations will transition to the long-term model-based approach (Figure S-1). An important component of this proposed process is a Model Evaluation Project (MEP), which will provide scientific grounding for model-based PMP estimation, inform development of the necessary modeling infrastructure, and
provide the foundation for determining when the transition should occur. Results from the MEP will also provide key tools for enhancing PMP estimation in the near term.
Near-term enhancements to current methods can be applied to update PMP estimates for the United States over the next several years. These enhancements can be grounded in improved data for storm catalogs, integration of model-based analyses of PMP-magnitude storms into PMP estimation procedures, and synthesis of advances in scientific understanding of extreme rainfall into the approaches used to implement storm transposition, moisture maximization, and transposition factors. Building on recent advances in PMP studies, improved rainfall data for PMP estimation can be developed from radar and surface rainfall observations. Model-based reconstruction of storm catalog events that control historical PMP estimates can refine rainfall analyses for these storms and provide scientific grounding for subjective decisions used to implement PMP methods. Reconstructions of major historical storms also contribute to development of model-based PMP estimation procedures and are an important component of the MEP. For near-term PMP estimation, the effects of climate change can be incorporated through climate change adjustment factors developed from model-based temperature scaling relationships.
The long-term model-based approach will employ kilometer-scale climate models capable of resolving PMP storms and producing PMP-magnitude precipitation. To estimate the depth of precipitation with an extremely low AEP over a particular duration and areal extent, researchers will need initial-condition large ensemble simulations to construct the appropriate probability density functions of precipitation. Large ensemble simulations driven by different external forcings will provide precipitation data for estimating PMP for the present-day and for the future under different socioeconomic scenarios or global warming levels. By capturing natural variability, large ensemble simulations will also enable statistical quantification of the uncertainty of the PMP estimates. PMP uncertainty can be used to improve PMF estimates for risk assessments and designs with RIDM. Furthermore, high-resolution space-time fields can be beneficial to a wide variety of other hydrologic and climatological applications.
In specifying the AEP that defines PMP, the user community must consider the relationship between PMP estimates derived from near-term enhancements and from models. Large changes in PMP estimates due to changes in methods would create major problems for the user community and could undermine confidence in new methods. Assessment of model-based exceedance probabilities of the PMP estimates obtained using near-term enhancements will guide selection of AEPs that define PMP.
KEY RECOMMENDATIONS
A New Definition of PMP
Based on a review and discussion of existing PMP definitions, a review of PMP estimation methods, and assessment of user needs, the committee concludes that a new PMP definition is needed.
Recommendation 5-3: NOAA, federal and state agencies involved in dam safety and nuclear regulation, the American Meteorological Society, the American Society of Civil Engineers, and the Association of State Dam Safety Officials should adopt a revised PMP definition: Probable Maximum Precipitation—The depth of precipitation for a particular duration, location, and areal extent, such as a drainage basin, with an extremely low annual probability of being exceeded, for a specified climate period.
Specification of Annual Exceedance Probabilities for PMP
National guidance for specifying AEPs that define PMP is needed. The AEPs derived from model-based analyses of near-term PMP estimates (to be completed over the next several years) will provide a key tool for developing national guidance.
Recommendation 5-4: Commensurate with the new definition, NOAA and the Federal Emergency Management Agency National Dam Safety Program, in partnership with federal agencies, states, and the Association of State Dam Safety Officials, should develop guidance for specifying annual exceedance probabilities for PMP that are acceptable for infrastructure decisions and society.
Phased Approach to Modernizing PMP Estimation
The committee recommends a phased approach to achieving the vision of model-based PMP estimation conforming to the new definition. The framework for the phased approach is summarized in Figure S-1.
Recommendation 5-1: NOAA should pursue a phased approach to modernizing PMP estimation, with the near-term approach building on enhancements to conventional PMP procedures and leading to a long-term model-based framework that can provide uncertainty characterization of PMP estimates, fully incorporating the effects of climate change.
Enhanced Data for Near-Term PMP Estimation
Weather radar is a key observational resource for enhancing storm catalogs used for near-term PMP estimation. Digitizing and enhancing the historical storm catalogs are also important steps to making near-term enhancements to PMP.
Recommendation 5-5: The U.S. Army Corps of Engineers should make its existing storm catalog publicly available. NOAA should facilitate digitization and enhancement of the existing storm catalog of historical extreme storms used in PMP for the United States to contain gridded rainfall fields and moisture data for each event. NOAA should facilitate development of an expanded storm catalog including high-resolution radar rainfall fields and available surface rainfall measurements for the United States to improve near-term estimation of PMP.
Storm Reconstructions for Near-Term PMP Estimation
Model reconstructions of extreme historical storms can improve the data and scientific understanding incorporated in near-term enhancements of PMP. Such reconstructions also provide critical guidance for simulating PMP-magnitude storms, which is needed to implement model-based estimation methods.
Recommendation 5-7: NOAA should facilitate model simulations of historical storm events that (1) may be added to the expanded storm catalog, (2) enhance scientific understanding of PMP-magnitude storms and their precipitation distributions, and (3) contribute to the Model Evaluation Project.
Scientific Guidance for Near-Term Enhancements to PMP Estimation
Subjective judgment plays an important role in implementation of current PMP procedures. Advances in scientific understanding could significantly improve PMP estimates during the near-term enhancement phase.
Recommendation 5-8: NOAA should include a summary of scientific principles in its national guidance for near-term PMP estimation. Near-term enhancements to storm transposition, moisture maximization, and transposition factors—especially for components involving subjective decisions—should be grounded in advances in scientific understanding, as detailed in this guidance.
Climate Change and Near-Term Enhancements to PMP Estimation
Physical understanding, historical trends, and model simulations and projections all signal an increase in extreme precipitation with warming. Near-term enhancements should address the effects of climate change on PMP.
Recommendation 5-9: For near-term enhancements to PMP estimation, NOAA should adopt climate change adjustment factors based on the model-based scaling relationship between extreme precipitation and temperature.
Model-Based PMP Estimation
Ensembles of long-term simulated rainfall fields over the United States from high-fidelity and high-resolution storm-resolving climate models can provide the foundation for long-term modernized PMP estimation, including statistical characterization of uncertainty and incorporation of climate change effects on rainfall extremes.
Recommendation 5-10: In the long term, NOAA should adopt a model-based approach to PMP estimation that aligns with the revised PMP definition, consisting of multi-model large ensemble kilometer-scale or finer-resolution
modeling to construct the probability distribution of precipitation for PMP estimation under different climates.
Recommendation 5-11: For the long-term approach and in agreement with the recommended PMP definition, NOAA should use statistical approaches to estimate PMP (with associated uncertainty) as the precipitation depth corresponding to an extremely low annual exceedance probabilty from the model-simulated precipitation distribution, with particular consideration of extreme value analysis based on threshold exceedance levels.
Model Evaluation Project
The MEP is a critical step in transitioning from near-term enhancements to PMP estimation to implementation of model-based PMP estimation methods. The advances in modeling capabilities necessary for PMP estimation will be developed and demonstrated, including approaches for incorporating the effects of climate change.
Recommendation 5-12: NOAA should embark on a Model Evaluation Project to assess model skill, identify strengths and limitations relevant to PMP estimation in current and future climate states, and achieve fitness for purpose, which is necessary for community confidence in models for estimating PMP.
CORE PRINCIPLES FOR THE DEVELOPMENT AND USE OF MODERNIZED PMP ESTIMATES
The development and use of modernized PMP estimates should be guided by four principles: transparency, objectivity, accessibility, and reproducibility. Transparency plays a pivotal role in building trust among practitioners, regulators, researchers, and the public and lays the groundwork for independent assessment of PMP products that facilitate evidence-based policymaking. Objectivity aims to minimize the reliance on subjective judgments. Advances in data, tools, and scientific understanding of extreme rainfall will enable practitioners to more objectively implement the near-term enhancements of PMP estimation and to transition to model-based methods. Accessibility of data and methodologies should be emphasized throughout the entire process of PMP development. PMP products should be regarded as public goods readily available to the general public with minimum restrictions, as well as adhering to the FAIR principles (findable, accessible, interoperable, reusable). Reproducibility refers to the expectation that PMP products should be broadly reproducible using the same data and methods. Reproducibility is closely linked to the preceding core standards, because transparency, objectivity, and accessibility are essential for ensuring the reproducibility of PMP products.
In addition to the above core principles, the committee advocates for sustained collaboration between NOAA and stakeholder groups throughout the process of modernizing PMP estimation. Collaborative efforts should focus on developing long-term
relationships between NOAA and end-users, establishing two-way communication pathways between groups, and emphasizing the creation of usable science and products.
Recommendation 5-2: NOAA should deliberately engage the scientific and practitioner communities to enhance understanding of the scientific process, clarify methodological considerations, increase awareness of practitioner needs, and collaboratively shape resulting products in support of modernized PMP estimates.
GOING BEYOND PMP: INFRASTRUCTURE SAFETY UNDER EXTREMES IN A CHANGING CLIMATE
The recommended approach for modernizing PMP estimation is based on the premise that state-of-the-art observations, physical understanding of extreme storms, and the capacity for high-fidelity, high-resolution simulations under different climatic forcings can transform the capabilities for assessing precipitation extremes in a warming climate. Significant research is needed to achieve the vision of model-based PMP estimation, and this endeavor will require scientific and modeling advances that should engage researchers across a broad array of disciplines. It will also require synergistic collaborations between federal agencies, academia, and the private sector. Scientific and modeling advances along this front will contribute not only to modernizing PMP estimation, but more broadly to addressing the societal challenges linked to the changes in extreme storms and precipitation in a warming climate—critical steps to ensuring the safety of our infrastructure and society.
Accurate high-resolution simulations of storms and precipitation in the current and future climates will enable rigorous assessment of how space-time patterns of precipitation for extreme storms will change at different spatial and temporal scales, from sub-hourly and kilometer scales to the scales of large basins upstream of high-hazard dams. The information gained from these assessments is essential for modeling extreme floods, for planning and water management decisions, and for vulnerability assessment of communities and critical infrastructure to extremes. The kilometer-scale simulations will also provide critically needed information for assessing future changes in hazards that are often coupled with extreme rainfall, including coastal storm surge and compound flooding.
