Paleoflood Data Research Articles

Holocene Thermal Maximum paleofloods improve flood frequency analyses in the lower Tennessee River Basin (USA)

Extreme floods are underrepresented in instrumented flow records. Consequently, flood frequency model estimates of extreme floods contain large uncertainty. The lower Tennessee River (LTR) Basin is located within the southeastern United States and is a large, regulated, and socio-economically significant North American river. The flood-of-record for the LTR occurred in 1867 CE. The 1867 flood-of-record is a design flow for the dams built and managed by the Tennessee Valley Authority. Here we develop and apply an extreme paleoflood chronology and use it to improve flood frequency model estimates of large, rare floods. We created an extreme paleoflood chronology that contained paleodischarge estimates for six extreme paleofloods, dating back to 5900 years before present, for a ∼2 km segment of the LTR. We obtained paleoflood hydrologic data from two sources. These included field sampling and analysis of slackwater deposits found in a rock shelter located on a limestone bluff and published extreme paleoflood data from a terrace located within the study segment.Placed within the context of the extreme paleoflood chronology, the 1867 flood-of-record was the largest flood to occur within the last 2000 years and coincided with the conclusion of the cooler, wetter Little Ice Age. The three highest magnitude extreme floods, however, occurred during a peak in Holocene temperatures 6000 to 5000 years ago, during the mid-Holocene Thermal Maximum. Paleodischarge estimates for the three mid-Holocene extreme floods were three times larger than the estimated discharge of the 1867 CE flood-of-record. Different iterations of a Bayesian flood frequency model showed removing the three most extreme, mid-Holocene paleofloods resulted in a rarer annual exceedance probability for the 1867 CE flood-of-record, changing it from an 0.002 AEP (500-year event) to an 0.0007 AEP (1300-year event) and underestimation of annual exceedance probabilities for design flows commonly used in river management and infrastructure design, including the 100-yr, 500-yr, 1000-yr, and 10,000-yr floods. These results suggest that flood frequency models may underestimate design flows if historic-floods-of-record are assumed to be the most extreme floods within these analyses. This finding has implications for regulated rivers worldwide, which now includes most large rivers. The previously unknown, extreme mid-Holocene paleofloods, which were found to be larger than all known modern or paleofloods, significantly altered the estimates of the final flood frequency model. This finding makes a compelling case for seeking out more extreme flood paleodischarge estimates from the mid-Holocene in rivers of the Northern Hemisphere, where the mid-Holocene Thermal Maximum's occurrence is well substantiated. Flood data from the mid-Holocene Thermal Maximum may help improve flood frequency model-based estimates of extreme floods by minimizing risk associated with flood “unknown unknowns.”

Paleoflood evidence for an upper limit of the maximum flood magnitudes along the Gold Gorge, the upper Hanjiang River, China

This paper provides an important archive of the largest paleoflood event along the Gold Gorge of the upper Hanjiang River, with a review of regional extraordinary flood magnitudes using historical, gauged, and paleoflood datasets. Paleoflood slackwater deposit (SWD) was identified using the sedimentological criteria and high-resolution physicochemical indexes. End-member model demonstrates the paleoflood SWD was dominated by sand component. Optical chronology indicates the paleoflood event was dated to 1000–900 a BP, corresponding to the North Song Dynasty (AD 960–1127). Paleoflood magnitudes derived from one-dimensional unsteady model vary between 23,000 and 34,000 m3/s. The paleoflood data provide a good representation of extraordinary flood event with a 10,000-years return period and greatly improve current flood frequency estimations. Moreover, the envelope curve for extraordinary floods was first established as Q = 50D0.65 and the paleoflood peak discharges are approximately 2–2.5 times of the largest gauged floods within the catchment. These results provide an upper limit of the maximum flood magnitudes along the upper Hanjiang River and also expand our understanding of rare and large magnitude floods in the middle Yangtze River valley.

Improvements to Flood Frequency Analysis on Alluvial Rivers Using Paleoflood Data

AbstractHydrologists and engineers routinely use flood frequency analyses to compute flood probabilities for mitigation, infrastructure planning, and emergency management. Conventional flood frequency analyses—in which annual discharge maxima from a stream gage are fit to a statistical probability distribution—often encounter large uncertainties when estimating extreme flood levels. Most gage records span relatively short periods of time (<100 years), and thus the most extreme and infrequently occurring flood events tend to be poorly represented in instrumental data sets. Here, we demonstrate how a new generation of paleoflood records derived from floodplain sediments can be used to improve the accuracy and precision of extreme flood probability estimates along alluvial rivers. We use a series of simulation experiments to show that incorporating large numbers of paleoflood events in flood frequency analyses can significantly reduce the uncertainty of extreme flood estimates when the paleoflood data are sufficiently accurate and precise. Our results illustrate that robust paleoflood records can improve the shape parameter of flood frequency distributions, which determine the thickness of distribution tails, when as many as 50 paleoflood events are incorporated. We conclude by demonstrating how an alluvial paleoflood data set reduces uncertainty in a flood frequency analysis for a gage on the lower Mississippi River. Finally, we provide recommendations for how to incorporate paleoflood information into flood frequency analysis to improve the accuracy of extreme flood probabilities.

Paleoflood hydrology on the lower Green River, upper Colorado River Basin, USA: An example of a naturalist approach to flood-risk analysis

Through a comprehensive paleoflood hydrological investigation we document natural evidence for at least 27 high-magnitude paleofloods at six sites on the Lower Green River, Utah. Hydraulic analysis, using the Sedimentation and River Hydraulic-2D model (SRH-2D), shows that the responsible peak paleoflood discharges ranged between 500 and 7500 m3/s. At least 14 of these paleoflood discharge peaks exceed a level twice that of the maximum systematic record of gauged flows: 1929 m3/s. Geochronological analyses, employing optically stimulated luminescence (OSL) and radiocarbon dating techniques, demonstrate that these 14 largest paleoflood peaks occurred during the past 700 years. Integration of the paleoflood data into flood frequency analyses (FFA) reveals considerably higher values for the upper tails of the flood distribution than does a FFA based solely on the systematic gauged record, indicating that extreme floods are larger and more frequent than implied by the relatively short gauged record. Through examination of three approaches to extreme flood estimation – conventional FFA, probable maximum flood estimation (PMF), and paleoflood hydrology (PFH) – we show the significance of the natural evidence for advancing scientific understanding of extreme floods that naturally occur in the Colorado River system. We argue that this kind of scientific understanding is absolutely essential for achieving a credible evaluation of extreme flood risk in a watershed of immense importance to economic prosperity of the southwestern U.S.

Pluridisciplinary analysis and multi-archive reconstruction of paleofloods: Societal demand, challenges and progress

Abstract Floods are one of the gravest natural hazards for societies, worsened by population growth, unchecked development, and climate change. From a Global Change perspective, past extreme events merit particular interest because they can be linked to wider climate and environmental changes, introduce perturbations. During the last decade, knowledge of long-term flood frequency and magnitude has been improved by extracting data from different types of archive. But, despite advances in dating methods, proxies and statistical techniques and efforts to identify atmospheric drivers, some fundamental questions remain unresolved. The Special Issue entitled “Pluridisciplinary analysis and multi-archive reconstruction of paleofloods” in the journal Global and Planetary Change addresses these uncertainties and complexities by assembling a selection of studies, which were first presented at the Past Climate Changes (PAGES) Open Scientific Meeting held at Zaragoza in 2017. In this introductory paper, the guest editors outline the 17 research contributions and meta-data from the 17 paleoflood studies were systematically analyzed in terms of i) geographical distribution; ii) methodologies applied; iii) types of archives; iii) numbers of flood series compiled and iv) spatial and temporal resolution of paleoflood data. The data indicate that paleoflood studies focused on fluvial depositional environments show a higher rate of integration with other types of paleoflood archive (mean of 4.5 types of archive) than studies focused on documentary sources (mean of 3.5) and lake sediments (mean of 2.4). We suggest that this strategy of archive integration has been adapted to effectively compensate for the higher uncertainties of fluvial deposition in floodplains. Statistical processing of the meta-data shows quantitative associations between specific types of flood archive and offers a solid platform for designing the optimal approach for multi-archive paleoflood research. A qualitative review and visual comparison of the 17 paleoflood series shows some consistent trends and breaks but also notable differences within and between regions. While a trend of increased flooding since 4-5 ka BP is evident, the lack of synchronicity between breaks and the coeval increases and decreases in fluvial activity is manifest. The majority of studies in the Special Issue do denote the 19th century – including the youngest cool climate pulses during the Little Ice Age – as a particularly flood-rich period. It is more difficult to assess the 20th century because of social changes, population growth and extensive river modification. Despite the mentioned uncertainties, 10 of 14 papers do not record the 20th century as an exceptional flood period. Assessing the effects of human impact on paleoflood calendars and disentangling anthropogenic from natural drivers are major challenges in integrated paleoflood analysis. It is concluded that the interpretation of flood series is complex as landscapes and flood drivers are heterogeneous and systems show different sensitivities to flood control and drivers. Thus, the study of past floods, from historical and natural archives, is challenging but also offers unparalleled opportunities to document low-frequency, large-magnitude flood events, which occurred under a broad range of climate and/or environmental scenarios, and, probably, the only way to reconstruct robust paleoflood series.

Weather Typing‐Based Flood Frequency Analysis Verified for Exceptional Historical Events of Past 500 Years Along the Meuse River

AbstractGovernments, policy makers, and water managers are pushed by recent socioeconomic developments such as population growth and increased urbanization inclusive of occupation of floodplains to impose very stringent regulations on the design of hydrological structures. These structures need to withstand storms with return periods typically ranging between 1,250 and 10,000 years. Such quantification involves extrapolations of systematically measured instrumental data, possibly complemented by quantitative and/or qualitative historical data and paleoflood data. The accuracy of the extrapolations is, however, highly unclear in practice. In order to evaluate extreme river peak flow extrapolation and accuracy, we studied historical and instrumental data of the past 500 years along the Meuse River. We moreover propose an alternative method for the estimation of the extreme value distribution of river peak flows, based on weather types derived by sea level pressure reconstructions. This approach results in a more accurate estimation of the tail of the distribution, where current methods are underestimating the design levels related to extreme high return periods. The design flood for a 1,250 year return period is estimated at 4,800 m3 s−1 for the proposed method, compared with 3,450 and 3,900 m3 s−1 for a traditional method and a previous study.

Reducing uncertainty with flood frequency analysis: The contribution of paleoflood and historical flood information

AbstractUsing a combination of stream gauge, historical, and paleoflood records to extend extreme flood records has proven to be useful in improving flood frequency analysis (FFA). The approach has typically been applied in localities with long historical records and/or suitable river settings for paleoflood reconstruction from slack‐water deposits (SWDs). However, many regions around the world have neither extensive historical information nor bedrock gorges suitable for SWDs preservation and paleoflood reconstruction. This study from subtropical Australia demonstrates that confined, semialluvial channels such as macrochannels provide relatively stable boundaries over the 1000–2000 year time period and the preserved SWDs enabled paleoflood reconstruction and their incorporation into FFA. FFA for three sites in subtropical Australia with the integration of historical and paleoflood data using Bayesian Inference methods showed a significant reduction in uncertainty associated with the estimated discharge of a flood quantile. Uncertainty associated with estimated discharge for the 1% Annual Exceedance Probability (AEP) flood is reduced by more than 50%. In addition, sensitivity analysis of possible within‐channel boundary changes shows that FFA is not significantly affected by any associated changes in channel capacity. Therefore, a greater range of channel types may be used for reliable paleoflood reconstruction by evaluating the stability of inset alluvial units, thereby increasing the quantity of temporal data available for FFA. The reduction in uncertainty, particularly in the prediction of the ≤1% AEP design flood, will improve flood risk planning and management in regions with limited temporal flood data.

Physically-based extreme flood frequency with stochastic storm transposition and paleoflood data on large watersheds

Summary Traditionally, deterministic flood procedures such as the Probable Maximum Flood have been used for critical infrastructure design. Some Federal agencies now use hydrologic risk analysis to assess potential impacts of extreme events on existing structures such as large dams. Extreme flood hazard estimates and distributions are needed for these efforts, with very low annual exceedance probabilities ( ⩽ 10 - 4 ) (return periods >10,000 years). An integrated data-modeling hydrologic hazard framework for physically-based extreme flood hazard estimation is presented. Key elements include: (1) a physically-based runoff model (TREX) coupled with a stochastic storm transposition technique; (2) hydrometeorological information from radar and an extreme storm catalog; and (3) streamflow and paleoflood data for independently testing and refining runoff model predictions at internal locations. This new approach requires full integration of collaborative work in hydrometeorology, flood hydrology and paleoflood hydrology. An application on the 12,000 km2 Arkansas River watershed in Colorado demonstrates that the size and location of extreme storms are critical factors in the analysis of basin-average rainfall frequency and flood peak distributions. Runoff model results are substantially improved by the availability and use of paleoflood nonexceedance data spanning the past 1000 years at critical watershed locations.

Regional flood frequency analyses involving extraordinary flood events at ungauged sites: further developments and validations

Flood frequency analyses are often based on recorded series at gauging stations. However, the length of the available data sets is usually too short to provide reliable estimates of extreme design floods. Hence, hydrologists have tried to make use of alternative sources of information to enrich the datasets used for the statistical inferences. Two main approaches were therefore proposed. The first consists in extending the information in time, making use of historical and paleoflood data. The second, spatial extension, consists in merging statistically homogeneous data to build large regional data samples. Recently, a combination of the two techniques aiming at including estimated extreme discharges at ungauged sites of a region in the regional flood frequency analyses has been proposed. This paper presents a consolidation of this approach and its comparison with the standard regional flood frequency approach proposed by Hosking & Wallis. A modification of the likelihood function is introduced to enable the simultaneous calibration of a regional index flood relation and of the parameters of the regional growth curve. Moreover, the efficiency of the proposed method is evaluated based on a large number of Monte Carlo simulated data sets. This work confirms that extreme peak discharges estimated at ungauged sites may be of great value for the evaluation of large return period (typically over 100 years) flood quantiles. They should therefore not be neglected despite the uncertainties associated to these estimates.

Paleohydrologic bounds and extreme flood frequency of the Upper Arkansas River, Colorado, USA

The Upper Arkansas River basin has experienced notable large floods, including the event of 2–6 June 1921 that devastated the city of Pueblo, Colorado. We investigated flood and paleoflood hydrology at strategic sites to determine the frequency and geographic extent of extreme floods within the basin for a dam safety application. Streamgage, historical, and paleoflood data were utilized to develop frequency curves at sites near Salida, Cotopaxi, Parkdale, and Pueblo. Soil/stratigraphic descriptions, radiocarbon dating, and hydraulic modeling were used to estimate paleoflood nonexceedance bounds at the four sites, which ranged from 400 to 2200 YBP for late Holocene surfaces to late Pleistocene surfaces near Cotopaxi. Peak-flow data are from lower-magnitude snowmelt runoff in May and June in the upper basin and from high-magnitude rainfall runoff from June to August in the lower basin. Flood frequency curves reflect this transition near Parkdale from snowmelt to extreme rainfall-runoff. For similar return periods, paleoflood peak discharges increase from about 480 m 3/s upstream at Loma Linda to about 4250 m 3/s downstream near Pueblo. This increase is attributed to the larger rainfall component derived from lower elevations between Loma Linda and Pueblo. Return periods for design floods at Pueblo Dam exceeded 10,000 years based on paleoflood frequency curve extrapolations.

Paleofloods and historical floods of the Ardèche River, France

Better understanding of flood occurrences and long‐term, floodplain planning, and flood risk assessment is achieved by integration of gauged, historical, and paleoflood data. The Ardèche River is ideal for this historical flood‐paleoflood study because its historical flood levels record dates back as early as A.D. 587 and useful data date back to A.D. 1522, its systematic gauging record is over 100 years long, and the geologic and geomorphic settings are optimal for paleoflood studies. Three sites provide three different thresholds for flood stages and SWD accumulation. According to our one‐dimensional (1‐D) step‐backwater calculations these three thresholds are 5200–5700 m3 s−1, 4900–5400 m3 s−1, and 3600–4000 m3 s−1 recording 6, 9, and 19 large Holocene floods, respectively. Dating the deposits enabled a correlation with the historical record. These paleoflood studies indicate that there are long gaps in flood occurrences on the Ardèche River; the floods are not randomly distributed in time but are clustered. They also indicate that the recent nineteenth century floods were the largest at the millennial timescale.

Comparisons of two moments‐based estimators that utilize historical and paleoflood data for the log Pearson type III distribution

The expected moments algorithm (EMA) [Cohn et al., 1997] and the Bulletin 17B [Interagency Committee on Water Data, 1982] historical weighting procedure (B17H) for the log Pearson type III distribution are compared by Monte Carlo computer simulation for cases in which historical and/or paleoflood data are available. The relative performance of the estimators was explored for three cases: fixed‐threshold exceedances, a fixed number of large floods, and floods generated from a different parent distribution. EMA can effectively incorporate four types of historical and paleoflood data: floods where the discharge is explicitly known, unknown discharges below a single threshold, floods with unknown discharge that exceed some level, and floods with discharges described in a range. The B17H estimator can utilize only the first two types of historical information. Including historical/paleoflood data in the simulation experiments significantly improved the quantile estimates in terms of mean square error and bias relative to using gage data alone. EMA performed significantly better than B17H in nearly all cases considered. B17H performed as well as EMA for estimating X100 in some limited fixed‐threshold exceedance cases. EMA performed comparatively much better in other fixed‐threshold situations, for the single large flood case, and in cases when estimating extreme floods equal to or greater than X500. B17H did not fully utilize historical information when the historical period exceeded 200 years. Robustness studies using GEV‐simulated data confirmed that EMA performed better than B17H. Overall, EMA is preferred to B17H when historical and paleoflood data are available for flood frequency analysis.

Data-based comparisons of moments estimators using historical and paleoflood data

This paper presents the first systematic comparison, using historical and paleoflood data, of moments-based flood frequency methods. Peak flow estimates were compiled from streamflow-gaging stations with historical and/or paleoflood data at 36 sites located in the United States, Argentina, United Kingdom and China, covering a diverse range of hydrologic conditions. The Expected Moments Algorithm (EMA) and the Bulletin 17B historical weighting procedure (B17H) were compared in terms of goodness of fit using 25 of the data sets. Results from this comparison indicate that EMA is a viable alternative to current B17H procedures from an operational perspective, and performed equal to or better than B17H for the data analyzed. We demonstrate satisfactory EMA performance for the remaining 11 sites with multiple thresholds and binomial censoring, which B17H cannot accommodate. It is shown that the EMA estimator readily incorporates these types of information and the LP-III distribution provided an adequate fit to the data in most cases. The results shown here are consistent with Monte Carlo simulation studies, and demonstrate that EMA is preferred overall to B17H. The Bulletin 17B document could be revised to include an option for EMA as an alternative to the existing historical weighting approach. These results are of practical relevance to hydrologists and water resources managers for applications in floodplain management, design of hydraulic structures, and risk analysis for dams.

Regional interdisciplinary paleoflood approach to assess extreme flood potential

In the past decade, there has been a growing interest of dam safety officials to incorporate a risk‐based analysis for design‐flood hydrology. Extreme or rare floods, with probabilities in the range of about 10−3 to 10−7 chance of occurrence per year, are of continuing interest to the hydrologic and engineering communities for purposes of planning and design of structures such as dams [National Research Council, 1988]. The National Research Council stresses that as much information as possible about floods needs to be used for evaluation of the risk and consequences of any decision. A regional interdisciplinary paleoflood approach was developed to assist dam safety officials and floodplain managers in their assessments of the risk of large floods. The interdisciplinary components included documenting maximum paleofloods and a regional analyses of contemporary extreme rainfall and flood data to complement a site‐specific probable maximum precipitation study [Tomlinson and Solak, 1997]. The cost‐effective approach, which can be used in many other hydrometeorologic settings, was applied to Elkhead Reservoir in Elkhead Creek (531 km2) in northwestern Colorado; the regional study area was 10,900 km2. Paleoflood data using bouldery flood deposits and noninundation surfaces for 88 streams were used to document maximum flood discharges that have occurred during the Holocene. Several relative dating methods were used to determine the age of paleoflood deposits and noninundation surfaces. No evidence of substantial flooding was found in the study area. The maximum paleoflood of 135 m3 s−1 for Elkhead Creek is about 13% of the site‐specific probable maximum flood of 1020 m3 s−1. Flood‐frequency relations using the expected moments algorithm, which better incorporates paleoflood data, were developed to assess the risk of extreme floods. Envelope curves encompassing maximum rainfall (181 sites) and floods (218 sites) were developed for northwestern Colorado to help define maximum contemporary and Holocene flooding in Elkhead Creek and in a regional frequency context. Study results for Elkhead Reservoir were accepted by the Colorado State Engineer for dam safety certification.

The 1993 Mississippi River Flood: A One Hundred or a One Thousand Year Event?

Power-law (fractal) extreme-value statistics are applicable to many natural phenomena under a wide variety of circumstances. Data from a hydrologic station in Keokuk, Iowa, shows the great flood of the Mississippi River in 1993 has a recurrence interval on the order of 100 years using power-law statistics applied to partial-duration flood series and on the order of 1,000 years using a log-Pearson type 3 (LP3) distribution applied to annual series. The LP3 analysis is the federally adopted probability distribution for flood-frequency estimation of extreme events. We suggest that power-law statistics are preferable to LP3 analysis. As a further test of the power-law approach we consider paleoflood data from the Colorado River. We compare power-law and LP3 extrapolations of historical data with these paleo-floods. The results are remarkably similar to those obtained for the Mississippi River. Recurrence intervals from power-law statistics applied to Lees Ferry discharge data are generally consistent with inferred 100- and 1,000-year paleofloods, whereas LP3 analysis gives recurrence intervals that are orders of magnitude longer. For both the Keokuk and Lees Ferry gauges, the use of an annual series introduces an artificial curvature in log-log space that leads to an underestimate of severe floods. Power-law statistics are predicting much shorter recurrence intervals than the federally adopted LP3 statistics. We suggest that if power-law behavior is applicable, then the likelihood of severe floods is much higher. More conservative dam designs and land-use restrictions may be required.

Multilevel Model for Gage and Paleoflood Data

A model is developed to estimate the parameters of a lognormal flow‐frequency distribution, and it can include multiple historical or paleoflood periods in which the largest flow within the period is defined and a current gage record. Asymptotic approximations of the variances and covariances of the distribution parameters are presented and used to compute first‐order approximations of the standard deviations of 100‐year and 500‐year floods. The model is applied to gage paleoflood data for Black Bear Creek at Pawnee, Oklahoma. Distribution parameters, their approximate variances and covariance Q100 and Q500, and their standard deviations were computed. A simulation experiment using 300 samples was performed. The Q100 and Q500 estimated using the multiple‐period model had lower standard errors than those estimated using either a single‐threshold or gage‐record‐only model. First‐order approximations of the standard deviations of Q100 and Q500 were compared to the simulation results and the largest error was...

Flood frequency analysis with systematic and historical or paleoflood data based on the two‐parameter general extreme value models

Historical and paleoflood data have become an important source of information for flood frequency analysis. A number of studies have been proposed in the literature regarding the value of historical and paleoflood information for estimating flood quantiles. These studies have been generally based on computer simulation experiments. In this paper the value of using systematic and historical/paleoflood data relative to using systematic records alone is examined analytically by comparing the asymptotic variances of flood quantiles assuming a two‐parameter general extreme value marginal distribution, type 1 and type 2 censored data, and maximum likelihood estimation method. The results of this study indicate that the value of historical and paleoflood data for estimating flood quantiles can be small or large depending on only three factors: the relative magnitudes of the length of the systematic record (N) and the length of the historical period (M); the return period (T) of the flood quantile of interest; and the return period (H) of the threshold level of perception. For instance, for N = 50, M = 50 and T = 500, the statistical gain for type 2 censoring becomes significantly larger than for type 1 censoring as H becomes greater than 100 years. In addition, computer experiments have shown that the results regarding the statistical gain based on asymptotic considerations are valid for the usual sample sizes.

Spatial variability of small‐basin paleoflood magnitudes for a southeastern Arizona mountain range

Paleohydrological procedures were used to analyze spatial variations in flood magnitude in drainage basins of the Santa Catalina Mountains (maximum elevation 2791 m), southeastern Arizona. Paleoflood reconstructions in eight watersheds, ranging from 10 to 110 km2, were obtained for the largest floods in stable canyon reaches, yielding discharges ranging from 50 to 270 m3 s−1. Although there is a high correlation between drainage area and flood magnitude, it is also clear that basins located in the southern half of the Santa Catalina Mountains have larger unit discharges (peak discharge divided by drainage area) and that these values decrease toward the northern part of the range. This spatial pattern is probably the result of local hydroclimatological conditions whereby the largest floods are associated with the northward influx of moist, tropical, and subtropical air. For this atmospheric circulation the northern part of the range experiences a rain shadow effect that decreases the unit discharges generated in those basins. The unusually high unit discharge for one of the south facing basins is explained by its basin morphometry. Inverse problem solutions using a rainfall‐runoff model demonstrate that both this basin and another with a comparable drainage area but a much lower estimated peak paleoflood discharge can have their peak flood outputs produced by the same storm intensity. The results demonstrate the potential of paleoflood data collection for understanding spatial variability in small basin flood hydrology.

Paleoflood evidence for a natural upper bound to flood magnitudes in the Colorado River Basin

The existence of an upper limit to the magnitude of floods in a region is a long‐standing and controversial hypothesis in flood hydrology. Regional envelope curves encompassing maximum flood magnitudes stabilize with progressive increases in the areal coverage and period of observation (Wolman and Costa, 1984). However, the short lengths of conventional gaging records limit substantial advances in testing whether this stabilization is evidence of an upper limit. In the Colorado River basin there are 32,120 station years of gage data, but the average period at a gaging station is only 20 years, with most stations having less than 70 years of observation. Paleoflood magnitudes derived from sediments of large prehistoric floods from 25 sites on rivers in Arizona and Utah provide additional data to extend the records of the largest floods. The paleoflood data identify the maximum flood discharges that have occurred on individual rivers over the last several hundred to several thousand years. Even with this increase in the observational period, the largest paleoflood discharges do not exceed the upper bound of maximum peak discharges delineated by the envelope curve derived from the available gaged and historical records. This result accords with the hypothesis of an upper physical limit for flood magnitudes and suggests that, for the Colorado River basin, the upper limit can be approximated by existing systematic and historical data for large floods. Similar relationships also hold when paleofloods and gaged records are presented for the subregion of southern Arizona.

Nonparametric variable kernel estimation with historical floods and paleoflood information

Extending a data record back in time using historical or paleoflood data has the potential to provide a considerable amount of information on very large floods. Parametric estimation methods are readily applicable to flood frequency analysis when historical or paleoflood estimates are available. However, all parametric approaches need an assumption about the underlying parent distribution, which is never known exactly. In recent years nonparametric methods of estimating probability density functions have been developed. Each of these involves the use of a suitable smoothing function known as a kernel. A new nonparametric variable kernel estimation model is proposed. Results obtained from a limited amount of real data and from simulation experiments show that quantiles estimated by the nonparametric method are better than those estimated by some selected parametric models both in descriptive ability and in predictive ability. The uncertainty in the choice of the threshold value of perception of a historical flood is also discussed.