Search

We analyze the diurnal cycle of maximum rainfall from ∼300 TCs from March 2014 through February 2017, by cross-referencing the path of tropical cyclones (TCs) and high-resolution rainfall estimates from IMERG (Integrated Multi-satellitE Rainfall from GPM - Global Precipitation Measurement mission). IMERG is a gridded satellite product that offers high-resolution rainfall estimates at a spatiotemporal resolution of 0.1° × 0.1° every 30 min, which are particularly suitable for these analyses. Because of the nature of the data, we use circular statistics. Circular statistics allows us to account for the natural periodicity of a random variable such as the time of the day at which maximum rainfall from TCs occurs. We follow the non-parametric approach of Mixtures of Von Mises-Fisher distribution (MvMF), which enables an easy-to-interpret parameter identification of multimodal and anisotropic distributions of the TC-rainfall. We stratify our analysis by storm duration, maturity, and intensity, basin of origin, radial proximity to the center of the storm, and whether the storm is over the ocean or land. In general, and across all scales, we find that there are mainly two cycles of maximum TC-rainfall: one diurnal cycle with peaks at ∼10 and ∼22 h (local time), and one semi-diurnal cycle with peaks at ∼2 and ∼5 h (local time). Although in a smaller proportion, the latter exhibits a weak afternoon alternative, i.e., ∼14 and ∼18 h (local time).

The aim of this study is to examine the contribution of North Atlantic tropical cyclones (TCs) to flooding and heavy rainfall across the continental United States. Analyses highlight the spatial variability in these hazards, their temporal changes in terms of frequency and magnitude, and their connection to large-scale climate, in particular to the North Atlantic Oscillation (NAO) and El Nino-Southern Oscillation (ENSO). We use long-term stream and rain gage measurements, and our analyses are based on annual maxima (AMs) and peaks-over-threshold (POTs). TCs contribute to ∼20–30% of AMs and POTs over Florida and coastal areas of the eastern United States, and the contribution decreases as we move inland. We do not detect statistically significant trends in the magnitude or frequency of TC floods. Regarding the role of climate, NAO and ENSO do not play a large role in controlling the frequency and magnitude of TC flooding. The connection between heavy rainfall and TCs is comparable to what observed in terms of flooding. Unlike flooding, NAO plays a significant role in TC-related extreme rainfall along the U.S. East Coast, while ENSO is most strongly linked to the TC precipitation in Texas.

The goal of this study is the evaluation of the skill of five state-of-the-art numerical weather prediction (NWP) systems [European Centre for Medium-Range Weather Forecasts (ECMWF), UK Met Office (UKMO), National Centers for Environmental Prediction (NCEP), China Meteorological Administration (CMA), and Canadian Meteorological Center (CMC)] in forecasting rainfall from North Atlantic tropical cyclones (TCs). Analyses focus on 15 North Atlantic TCs that made landfall along the U.S. coast over the 2007–2012 period. As reference data we use gridded rainfall provided by the Climate Prediction Center (CPC). We consider forecast lead-times up to five days. To benchmark the skill of these models, we consider rainfall estimates from one radar-based (Stage IV) and four satellite-based [Tropical Rainfall Measuring Mission - Multi-satellite Precipitation Analysis (TMPA, both real-time and research version); Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN); the CPC MORPHing Technique (CMORPH)] rainfall products. Daily and storm total rainfall fields from each of these remote sensing products are compared to the reference data to obtain information about the range of errors we can expect from “observational data.” The skill of the NWP models is quantified: (1) by visual examination of the distribution of the errors in storm total rainfall for the different lead-times, and numerical examination of the first three moments of the error distribution; (2) relative to climatology at the daily scale. Considering these skill metrics, we conclude that the NWP models can provide skillful forecasts of TC rainfall with lead-times up to 48 h, without a consistently best or worst NWP model.

Atmospheric rivers (ARs) play a central role in the hydrology and hydroclimatology of the central United States. More than 25% of the annual rainfall is associated with ARs over much of this region, with many large flood events tied to their occurrence. Despite the relevance of these storms for flood hydrology and water budget, the characteristics of rainfall associated with ARs over the central United has not been investigated thus far. This study fills this major scientific gap by describing the rainfall during ARs over the central United States using five remote sensing-based precipitation products over a 12-year study period. The products we consider are: Stage IV, Tropical Rainfall Measuring Mission – Multi-satellite Precipitation Analysis (TMPA, both real-time and research version); Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN); the CPC MORPHing Technique (CMORPH). As part of the study, we evaluate these products against a rain gauge-based dataset using both graphical- and metrics-based diagnostics. Based on our analyses, Stage IV is found to better reproduce the reference data. Hence, we use it for the characterization of rainfall in ARs. Most of the AR-rainfall is located in a narrow region within ∼150 km on both sides of the AR major axis. In this region, rainfall has a pronounced positive relationship with the magnitude of the water vapor transport. Moreover, we have also identified a consistent increase in rainfall intensity with duration (or persistence) of AR conditions. However, there is not a strong indication of diurnal variability in AR rainfall. These results can be directly used in developing flood protection strategies during ARs. Further, weather prediction agencies can benefit from the results of this study to achieve higher skill of resolving precipitation processes in their models.

Daily temperature values are generally computed as the average of the daily minimum and maximum observations, which can lead to biases in the estimation of daily averaged values. This study examines the impacts of these biases on the calculation of climatology and trends in temperature extremes at 409 sites in North America with at least 25 years of complete hourly records. Our results show that the calculation of daily temperature based on the average of minimum and maximum daily readings leads to an overestimation of the daily values of ~ 10+ % when focusing on extremes and values above (below) high (low) thresholds. Moreover, the effects of the data processing method on trend estimation are generally small, even though the use of the daily minimum and maximum readings reduces the power of trend detection (~ 5–10% fewer trends detected in comparison with the reference data).

The unseasonal winter heavy rainfall and flooding that occurred during December 2015–January 2016 had large socio-economic impacts for the central United States. Here we examine the climatic conditions that led to the observed extreme precipitation, and compare and contrast them with the 1982/1983 and 2011/2012 winters. The large precipitation amounts associated with the 1982/1983 and 2015/2016 winter flooding were linked to the strongly positive North Atlantic Oscillation (NAO), with large moisture transported from the Gulf of Mexico. The anomalous upper-level trough in the 1982- and 2015- Decembers over the western United States was also favorable for strong precipitation by leading the cold front over the central United States. In contrast, the extremely positive NAO in December 2011 did not lead to heavy rainfall and flooding because the Azores High center shifted too far westward (like a blocking high) preventing moisture from moving towards the central and southeastern United States.

Unprecedented alterations in precipitation characteristics over the last century and especially in the last two decades have posed serious socio-economic problems to society in terms of hydro-meteorological extremes, in particular flooding and droughts. The origin of these alterations has its roots in changing climatic conditions; however, its threatening implications can only be dealt with through meticulous planning that is based on realistic and skillful decadal precipitation predictions (DPPs). Skillful DPPs represent a very challenging prospect because of the complexities associated with precipitation predictions. Because of the limited skill and coarse spatial resolution, the DPPs provided by General Circulation Models (GCMs) fail to be directly applicable for impact assessment. Here, we focus on nine GCMs and quantify the seasonally and regionally averaged skill in DPPs over the continental United States. We address the problems pertaining to the limited skill and resolution by applying linear and kernel regression-based statistical downscaling approaches. For both the approaches, statistical relationships established over the calibration period (1961–1990) are applied to the retrospective and near future decadal predictions by GCMs to obtain DPPs at ∼4 km resolution. The skill is quantified across different metrics that evaluate potential skill, biases, long-term statistical properties, and uncertainty. Both the statistical approaches show improvements with respect to the raw GCM data, particularly in terms of the long-term statistical properties and uncertainty, irrespective of lead time. The outcome of the study is monthly DPPs from nine GCMs with 4-km spatial resolution, which can be used as a key input for impacts assessments.

The central United States is plagued by frequent catastrophic flooding, such as the flood events of 1993, 2008, 2011, 2013, 2014 and 2016. The goal of this study is to examine whether it is possible to describe the occurrence of flood and heavy precipitation events at the sub-seasonal scale in terms of variations in the climate system. Daily streamflow and precipitation time series over the central United States (defined here to include North Dakota, South Dakota, Nebraska, Kansas, Missouri, Iowa, Minnesota, Wisconsin, Illinois, West Virginia, Kentucky, Ohio, Indiana, and Michigan) are used in this study. We model the occurrence/non-occurrence of a flood and heavy precipitation event over time using regression models based on Cox processes, which can be viewed as a generalization of Poisson processes. Rather than assuming that an event (i.e., flooding or precipitation) occurs independently of the occurrence of the previous one (as in Poisson processes), Cox processes allow us to account for the potential presence of temporal clustering, which manifests itself in an alternation of quiet and active periods. Here we model the occurrence/non-occurrence of flood and heavy precipitation events using two climate indices as time-varying covariates: the Arctic Oscillation (AO) and the Pacific-North American pattern (PNA). We find that AO and/or PNA are important predictors in explaining the temporal clustering in flood occurrences in over 78% of the stream gages we considered. Similar results are obtained when working with heavy precipitation events. Analyses of the sensitivity of the results to different thresholds used to identify events lead to the same conclusions. The findings of this work highlight that variations in the climate system play a critical role in explaining the occurrence of flood and heavy precipitation events at the sub-seasonal scale over the central United States.

Reanalysis datasets have been under critical scrutiny due to their widespread use in various climatic and hydrological modeling applications, in particular over many areas of the globe with limited or absent reliable observational data. Nevertheless, reanalysis products are in the process of continuous improvements reflecting the improved system knowledge, model physics and assimilation techniques. In addition, several internal model adjustments have also been adopted to minimize the bias in reanalysis datasets. Considering these factors, it is necessary to investigate the inherent chaotic dynamics of reanalyses and the possible discrepancies, if any, with respect to the observational data. Here we compare and contrast the chaotic dynamics of daily precipitation and daily mean surface temperature simulated by the reanalysis against observed data over the continental United States. Our focus is on four reanalysis products: the National Aeronautics and Space Administration's Modern Era Retrospective-Analysis for Research and Applications (MERRA), European Centre for Medium-Range Weather Forecasts' ERA-Interim, Japanese Meteorological Agency's Japanese 55-year Reanalysis (JRA-55), and National Center for Environmental Prediction/National Center for Atmospheric Research's Reanalysis I. The inherent chaotic dynamics measured in terms of three statistics (i.e., maximum predictability, predictive error and predictive instability) reveal the inconsistency among the four reanalysis products. ERA-Interim is capable of simulating the precipitation's chaotic dynamics over much of the study region, while MERRA is found to be superior in capturing the temperature's chaotic dynamics. Analyses on various aspects of daily precipitation and temperature indicate that the biases in precipitation's chaotic dynamics may be attributed to the inconsistencies in simulating the signal-to-noise ratio and non-rainy days, while biases in temperature's chaotic dynamics could be due to the failure in replicating the abrupt trends in the recent decades by the reanalyses products.

This study examines the potential improvement in seasonal predictability of monthly precipitation and temperature using a novel machine learning approach, the convolutional Gaussian process (CGP). This approach allows us to take into account multiple quantities and their interdependencies simultaneously. We use one global climate model (FLORb01) part of the North American Multi-Model Ensemble (NMME) project and quantify its skill in reproducing precipitation and temperature in March and July across Iowa (central United States) for lead times from one month to one year. As a first step we train the CGP over the 1985–2005 period, and then apply it out of sample from 2006 to 2019. Over the validation period, our results indicate that the CGP is able to increase the skill (i.e., increased correlation coefficient and reduced root mean squared error) in predicting precipitation and temperature compared to both the raw outputs and after standard bias correction. These statements are consistent across different lead times and target month (i.e., March or July). These encouraging findings provide a new potential path towards improved predictability of the regional climate at the seasonal scale.

Baseflow, or the groundwater component of streamflow, is an important source of water for several applications, from increasing demands on freshwater resources to ecosystem health. Despite its relevance, our understanding of the processes driving baseflow and its interannual variability is limited. In this study, we focus on 458 U.S. Geological Survey streamflow gauges that have at least 50 years of daily data. We use a statistical modeling framework to select a set of predictors that represent the role of climate (i.e., precipitation, temperature and antecedent wetness) and land use (harvested acres of corn and soybeans). The models are able to describe well the variability in monthly baseflow across the region, with an average correlation coefficient between the observational records and the median of the fitted distribution of 0.70 among all months. Our results indicate that precipitation and antecedent wetness are the strongest predictors, where the latter was selected the most often. Temperature is an important predictor during the spring when snow-related processes are the most relevant. Agriculture was frequently selected in the Cornbelt region during the growing season (from March to July). The results of this study can inform future watershed management that sustains low flows and improves water quality.

The aim of this study is to examine whether the climatic driving forces can describe the observed variability in the frequency of flooding over the central United States. Results are based on daily streamflow records from 774 U.S. Geological Survey (USGS) stations with at least 50 years of data and ending no earlier than 2011. Five climate indices related to both the Atlantic and Pacific Oceans are used in this study: the North Atlantic Oscillation (NAO), the Southern Oscillation Index (SOI), the Pacific Decadal Oscillation (PDO), the Atlantic Multidecadal Oscillation (AMO), and the Pacific-North American pattern (PNA). A peak-over-threshold approach is used to identify flood peaks, and the relationship between the frequency of flood events and climate indices is investigated using Poisson regression. The results of this work indicate that climate variability can play a significant role in explaining the variations in the frequency of flooding over the central United States. Different climate modes are related to the frequency of flood events over different parts of the domain and for different seasons, with PNA playing an overall dominant role. Analyses related to flood events are extended to examine climate controls on heavy precipitation over the same area. We find that the variability of the Atlantic and Pacific Oceans can influence the frequency of heavy precipitation days in a manner similar to what was found for flooding. Therefore, these results suggest that the recent observed variability in the frequency of flood events and heavy precipitation over the central United States can be largely attributed to the variability in the climate system.

Summary Hydrologic time series are often characterized by temporal changes that give rise to non-stationarity. When the distribution describing the data changes over time, it is important to detect these changes so that correct inferences can be drawn from the data. The Lombard test, a non-parametric rank-based test to detect change points in the moments of a time series, has been recently used in the hydrologic literature to detect change points in the mean and variance. Little is known, however, about the performance of this test in detecting changes in variance, despite the potentially large impacts that these changes (shifts) could have when dealing with extremes. Here we address this issue in a Monte Carlo simulation framework. We consider a number of different situations that can manifest themselves in hydrologic time series, including the dependence of the results on the magnitude of the shift, significance level, sample size and location of the change point within the series. Analyses are performed considering abrupt changes in variance occurring with and without shifts in the mean. The results show that the power of the test in detecting change points in variance is small when the changes are small. It is large when the change point occurs close to the middle of the time series, and it increases nonlinearly with increasing sample size. Moreover, the power of the test is greatly reduced by the presence of change points in mean. We propose removing the change in the mean before testing for change points in variance. Simulation results demonstrate that this strategy effectively increases the power of the test. Finally, the Lombard test is applied to annual peak discharge records from 3686 U.S. Geological Survey stream-gaging stations across the conterminous United States, and the results are discussed in light of the insights from the simulations’ results.

This study examines the seasonality of flooding across the continental United States using circular statistics. Analyses are based on 7506 USGS stream gage stations with a record of least 30 years of annual maximum instantaneous peak discharge. Overall, there is a very strong seasonality in flooding across the United States, reflecting differences in flood generating mechanisms. Most of the flood events along the western and eastern United States tend to occur during the October–March period and are associated with extratropical cyclones. The average seasonality of flood events shifts to April–May in regions where snowmelt is the dominant flood agent, and later in the spring–summer across the central United States. The strength of the seasonal cycle also varies considerably, with the weakest seasonality in the Appalachian Mountains and the strongest in the northern Great Plains. The seasonal distribution of flooding is described in terms of circular uniform, reflective symmetric and asymmetric distributions. There are marked differences in the shape of the distribution across the continental United States, with the majority of the stations exhibiting a reflective symmetric distribution. Finally, nonstationarities in the seasonality of flooding are examined. Analyses are performed to detect changes over time, and to examine changes that are due to urbanization and regulation. Overall, there is not a strong signal of temporal changes. The strongest impact of urbanization and regulation is on the strength of the seasonal cycle, with indications that the signal weakens (i.e., the seasonal distribution becomes wider) under the effects of regulation.

There is growing empirical evidence that many river basins across the U.S. Midwest have been experiencing an increase in the frequency of flood events over the most recent decades. Albeit these detected changes are important to understand what happened in our recent past, they cannot be directly extrapolated to obtain information about possible future changes in the frequency of flood events. Building on recent statistically-based attribution studies, we project seasonal changes in the frequency of flood events at 286 U.S. Geological Survey gauging stations across the U.S. Midwest using projections of precipitation, antecedent wetness conditions and temperature as drivers. The projections of the covariates are obtained from two datasets obtained by downscaling global circulation models from the Fifth Coupled Model Intercomparison Project (CMIP5). We focus on the representative concentration pathway (RCP) 8.5 and on four different flood thresholds (i.e., from more common to less frequent flood events). We find that the frequency of flood events during the 21st century increases during spring at most of the analyzed gauging stations, with larger changes in the Northern Great Plains and regardless of the flood threshold value. Our findings also point to a projected increasing number of flood events during the winter, especially in the stations in the southern and western part of the domain (Iowa, Missouri, Illinois, Ohio, Indiana and Michigan). A marked change in the frequency of flood events is not projected for the summer and fall.

This study analyzes daily mean streamflow records from 5,311 U.S. Geological Survey stream gages in the continental United States and develops a Metastatistical Extreme Value Distribution (MEVD) tailored for flood frequency analysis. We compare the new tool with the Generalized Extreme Value (GEV) and Log-Pearson Type III (LP3) distributions and investigate the role of El Nino Southern Oscillation (ENSO) in the generation of floods. Hence, we formulate the MEVD in terms of mixture of distributions to describe the occurrence of flood peaks generated under different ENSO phases. We find that the MEVD outperforms GEV and LP3 distributions respectively in about 76% and 86% of the stations, with a significant improvement in the accuracy of quantiles corresponding to return periods much larger than the calibration sample size. The ENSO signature detected in the distributions of the daily peak flows does not necessarily improve the estimation of high return period flow values.

Summary Atmospheric rivers (ARs) are narrow corridors within the warm conveyor belt of extratropical cyclones in which the majority of the poleward water vapour transport occurs. These filamentary synoptic features are responsible for extreme precipitation and flooding in Europe and the central and western United States, and also play an essential role for water resources in these areas. Using gridded precipitation products across Europe and the continental United States and the ERA-Interim reanalysis, we investigate the fraction of precipitation from 1979 to 2012 that is related to ARs in these regions. The results are region- and month-dependent, with the largest contribution generally occurring during the winter season and being on the order of 30–50%. This is particularly true for Western Europe, the U.S. West Coast, and the central and northeastern United States. Our results suggest that ARs are important agents for water supply in Europe and the United States. We have also examined whether there have been changes over time in the fractional contribution of ARs to seasonal rainfall using zero-inflated beta regression. We find that there has been a decrease in the average AR-contribution over the Mediterranean region and over the central United States.

River discharge represents a vital resource for many human activities. The improved understanding of the physical processes controlling its regime can lead to large economic and societal benefits, such as improved flood warning and mitigation, and improved water management during droughts. This is particularly true for the agricultural U.S. Midwest and Iowa more specifically. Iowa is relentlessly plagued by catastrophic flooding, with the spring and summer river floods of 1993, 2008, and 2013 and the drought of 2012 being the most recent widespread events affecting the state. These natural disasters also come with a very large price tag, both in terms of economic damage and fatalities. During the 20th and 21st centuries, discharge over this area has been changing on a number of temporal scales, from annual to decadal. An outstanding question is related to the contribution of changes in the climate system and in land use/land cover and agricultural practices in explaining changes in discharge. We address this question by developing statistical models to describe the changes in different parts of the discharge distribution. We use rainfall and harvested corn and soybean acreage to explain the observed stream flow variability. We focus on the Raccoon River at Van Meter, which is a 9000-km2 watershed with daily discharge measurements covering most of the 20th century up to the present. Our results indicate that rainfall variability is responsible for the majority of the changes observed in the discharge record, with changes in cultivated area affecting the discharge responses in different ways, depending on which part of the discharge distribution is considered. In particular, land use change exacerbates high discharge during heavy precipitation and low discharge during low precipitation.

It is widely acknowledged that radar-based estimates of rainfall are affected by uncertainties (e.g., mis-calibration, beam blockage, anomalous propagation, and ground clutter) which are both systematic and random in nature. Improving the characterization of these errors would yield better understanding and interpretations of results from studies in which these estimates are used as inputs (e.g., hydrologic modeling) or initial conditions (e.g., rainfall forecasting). Building on earlier efforts, the authors apply a data-driven multiplicative model in which the relationship between true rainfall and radar rainfall can be described in terms of the product of a systematic and random component. The systematic component accounts for conditional biases. The conditional bias is approximated by a power-law function. The random component, which represents the random fluctuations remaining after correcting for systematic uncertainties, is characterized in terms of its probability distribution as well as its spatial and temporal dependencies. The space–time dependencies are computed using the non-parametric Kendall's τ measure. For the first time, the authors present a methodology based on conditional copulas to generate ensembles of random error fields with the prescribed marginal probability distribution and spatio-temporal dependencies. The methodology is illustrated using data from Clear Creek, which is a densely instrumented experimental watershed in eastern Iowa. Results are based on three years of radar data from the Davenport Weather Surveillance Radar 88 Doppler (WSR-88D) radar that were processed through the Hydro-NEXRAD system. The spatial and temporal resolutions are 0.5 km and hourly, respectively, and the radar data are complemented by rainfall measurements from 11 rain gages, located within the catchment, which are used to approximate true ground rainfall.

Significant progress has been made in relation to extreme precipitation and climate change. Overall, the tendency for dry areas to get drier and wet areas to get wetter has been identified. However, much of the work has focused on the daily time scale, and much less is known about sub-daily precipitation, despite indications that climate change could have more of an impact on sub-daily (e.g., 3-hourly) rather than daily precipitation. To complicate the matter, there is still a need to evaluate the performance of global climate models in reproducing the precipitation statistics at the sub-daily time scales. Therefore, the goal of this study is to explore the projected changes in sub-daily precipitation compared to the daily scale, and understand model accuracy at these finer time scales. We found that model accuracy is low for sub-daily precipitation for most of the models and model performance increases as the temporal resolution becomes coarser. In addition, remarkable differences exist in the accuracy of different GCMs in simulating sub-daily precipitation. However, there are several models that stand out comparatively for both time scales. Ultimately, the greatest changes in extreme precipitation, both increases and decreases, are generally in the tropics. There is a clear connection between greenhouse gas concentrations and extreme precipitation, with the greatest changes occurring towards the end of the 21st century when greenhouse gas concentrations are the greatest. At both time scales, models generally show a large increase in precipitation in the tropics, and global averages indicate increases in extreme precipitation will be more dominant than decreases. Changes are projected to be stronger at the sub-daily than daily scale, especially between 30° N and 30° S.

We use 16 years (1995–2010) of data from the National Lightning Detection Network (NLDN) to examine the spatial and temporal variability of major cloud-to-ground (CG) lightning days (defined as the 80 days with the largest lightning activity) over the continental United States. Extreme lightning activity is concentrated over the central U.S. and west of the Appalachian Mountains. The largest frequency of major lightning days is concentrated during the summertime, with a tendency for these major days to have occurred in recent years. We also examine the presence of monotonic patterns over time in CG lightning flashes over the continental United States. Analyses are performed at the monthly scale (from April to September) and for total, negative-only, and positive-only flashes. The non-parametric Mann–Kendall test is used to examine the presence of monotonic patterns. The upgrades in NLDN during the study period complicate the separation between cloud-to-cloud flashes (CC) and the targeted CG lightning flashes. The results of the trend analyses are sensitive to the threshold used to discriminate between CC and CG flashes, in particular for positive-only flashes. The central U.S. is an area that exhibits statistically significant increasing trends independently of the selected threshold, while there is a general tendency towards decreasing trends over the Rocky Mountains. These results raise the question of whether the observed changes in lightning activity during the recent years are related to natural or human-induced changes in the climate system, and/or to inhomogeneities in the observational network.

Summary Spatial and temporal variability in extreme rainfall, and its interactions with land cover and the drainage network, is an important driver of flood response. “Design storms,” which are commonly used for flood risk assessment, however, are assumed to be uniform in space and either uniform or highly idealized in time. The impacts of these and other commonly-made assumptions are rarely considered, and their impacts on flood risk estimates are poorly understood. This study presents an alternate framework for rainfall frequency analysis that couples stochastic storm transposition (SST) with “storm catalogs” developed from a ten-year high-resolution (15-min, 1-km 2 ) radar rainfall dataset for the region surrounding Charlotte, North Carolina, USA. The SST procedure involves spatial and temporal resampling from these storm catalogs to reconstruct the regional climatology of extreme rainfall. SST-based intensity–duration–frequency (IDF) estimates are driven by the spatial and temporal rainfall variability from weather radar observations, are tailored specifically to the chosen watershed, and do not require simplifying assumptions of storm structure. We are able to use the SST procedure to reproduce IDF estimates from conventional methods for four urban watersheds in Charlotte. We demonstrate that extreme rainfall can vary substantially in time and in space, with potentially important flood risk implications that cannot be assessed using conventional techniques. SST coupled with high-resolution radar rainfall fields represents a useful alternative to conventional design storms for flood risk assessment, the full advantages of which can be realized when the concept is extended to flood frequency analysis using a distributed hydrologic model.

Summary The authors developed an approach for deriving a statistical model of range-dependent error (RDE) in radar-rainfall estimates by parameterizing the structure of the non-uniform vertical profile of radar reflectivity (VPR). The proposed parameterization of the mean VPR and its expected variations are characterized by several climatological parameters that describe dominant atmospheric conditions related to vertical reflectivity variation. We have used four years of radar volume scan data from the Tulsa weather radar WSR-88D (Oklahoma) to illustrate this approach and have estimated the model parameters by minimizing the sum of the squared differences between the modeled and observed VPR influences that were computed using radar data. We evaluated the mean and standard deviation of the modeled RDE against rain gauge data from the Oklahoma Mesonet network. No rain gauge data were used in the model development. The authors used the three lowest antenna elevation angles to demonstrate the model performance for cold (November–April) and warm (May–October) seasons. The RDE derived from the parameterized models shows very good agreement with the observed differences between radar and rain gauge estimates of rainfall. For the third elevation angle and cold season, there are 82% and 42% improvements for the RDE and its standard deviation with respect to the no-VPR case. The results of this study indicate that VPR is a key factor in the characterization of the radar range-dependent bias, and the proposed models can be used to represent the radar RDE in the absence of rain gauge data.

Summary Annual and seasonal maximum daily discharge time series for 55 stations in central Europe (Germany, Switzerland, Czech Republic, and Slovakia) are used to examine flood frequency from a regional perspective. In this study we examine temporal nonstationarities in the flood peak records, and characterize upper tail and scaling properties of the flood peak distributions. There is a marked seasonality in the flood peak record, with a large fraction of annual maximum flood peaks occurring during the winter in the western part of the study domain, and during the summer in the southern portion of this region. The presence of abrupt and slowly varying changes in the flood time series is examined by means of non-parametric tests. Change-points in the mean and variance of the flood peak distributions are examined using the Pettitt test, while the presence of monotonic patterns is examined by means of Spearman and Mann-Kendall tests. Abrupt changes, rather than monotonic trends are responsible for violations of the stationarity assumption. These step changes can often be associated with anthropogenic effects, such as construction of dams and reservoirs and river training. Given the profound changes that these catchments have undergone, it is difficult to detect a possible climate change signal in the flood peak record. The estimates of the location, scale, and shape parameters of the Generalized Extreme Value distribution are used to examine the upper tail and scaling properties of the flood peak distributions. The location and scale parameters exhibit a power law behaviour when plotted as a function of drainage area, while the shape parameter decreases log-linearly for increasing catchment area. The findings of this study suggest that these records exhibit a heavy tail behaviour.

Summary Annual maximum daily rainfall time series from 221 rain gages in the Midwest United States with a record of at least 75 years are used to study extreme rainfall from a regional perspective. The main topics of this study are: (i) seasonality of extreme rainfall; (ii) temporal stationarity and long-term persistence of annual maximum daily rainfall; (iii) frequency analyses of annual maximum daily rainfall based on extreme value theory; and (iv) clustering of heavy rainfall events and impact of climate variables on the frequency of occurrence of heavy rainfall events. Annual maximum daily rainfall in the Midwest US exhibits a marked seasonality, with the largest frequencies concentrated in the period May–August. Non-parametric tests are used to examine the validity of the stationarity assumption in terms of both abrupt and slowly varying temporal changes. About 10% of the stations show a change-point in mean and/or variance. Increasing monotonic patterns are detected at 19 stations. Quantile regression analyses suggest that the number of stations with a significant increasing trend tends to decrease for increasing quantiles. Temporal changes in the annual maximum daily rainfall time series are also examined in terms of long-term persistence. Conclusive statements about the presence of long-term persistence in these records are, however, not possible due to the large uncertainties associated with the estimation of the Hurst exponent from a limited sample. Modeling of annual maximum daily rainfall records with the Generalized Extreme Value (GEV) distribution shows well-defined spatial patterns for the location and scale parameters but not for the shape parameter. Examination of the upper tail properties of the annual maximum daily rainfall records points to a heavy tail behavior for most of the stations considered in this study. The largest values of the 100-year annual maximum daily rainfall are found in the area between eastern Kansas, Iowa, and Missouri. Finally, we use the Poisson regression as a framework for the examination of clustering of heavy rainfall. Our results point to a clustering behavior due to temporal fluctuations in the rate of occurrence of the heavy rainfall events, which is modulated by climatic factors representing the influence of both Atlantic and Pacific Oceans.

We examine the warm season (April–September) rainfall climatology of the northeastern US through analyses of high-resolution radar rainfall fields from the Hydro-NEXRAD system and regional climate model simulations using the weather research and forecasting (WRF) model. Analyses center on the 5-year period from 2003 to 2007 and the study area includes the New York–New Jersey metropolitan region covered by radar rainfall fields from the Fort Dix, NJ WSR-88D. The objective of this study is to develop and test tools for examining rainfall climatology, with a special focus on heavy rainfall. An additional emphasis is on rainfall climatology in regions of complex terrain, like the northeastern US, which is characterized by land–water boundaries, large heterogeneity in land use and cover, and mountainous terrain in the western portion of the region. We develop a 5-year record of warm season radar rainfall fields for the study region using the Hydro-NEXRAD system. We perform regional downscaling simulations for the 5-year study period using the WRF model. Radar rainfall fields are used to characterize the interannual, seasonal and diurnal variation of rainfall over the study region and to examine spatial heterogeneity of rainfall. Regional climate model simulations are characterized by a wet bias in the rainfall fields, with the largest bias in the high-elevation regions of the model domain. We show that model simulations capture broad features of the interannual, seasonal, and diurnal variation of rainfall. Model simulations do not capture spatial gradients in radar rainfall fields around the New York metropolitan region and land–water boundaries to the east. The model climatology of convective available potential energy (CAPE) is used to interpret the regional distribution of warm season rainfall and the seasonal and diurnal variability of rainfall. We use hydrologic and meteorological observations from July 2007 to examine the interactions of land surface processes and rainfall from a regional perspective.

Summary The flash flood guidance system (FFGS) is an operational system which assists forecasters to issue flash flood warnings and watches over the conterminous United States. Currently, it is a deterministic system: given a specified precipitation accumulation over a basin and over a time period, issuing of flash flood watches and warnings is considered depending on the exceedance of a certain threshold value (flash flood guidance – FFG). For given channel characteristics and initial soil moisture conditions, FFG values are computed with the use of a hydrologic model. The purpose of this study is to consider the effects of radar-rainfall and flash flood guidance uncertainties on the FFGS. The errors in the FFG are accounted for by quantifying the uncertainties due to the estimation of the hydraulic and terrain characteristics, and the hydrologic model parameters and initial state. To account for the uncertainties in radar-rainfall, the authors use an empirically-based radar-rainfall error model. This model requires calibration for each application region and it accounts for range effects, synoptic conditions, space–time resolutions, and the spatial and temporal dependences of the errors. Thus, the combined effects of uncertainty in both radar-rainfall and FFG can be assessed. The results are exemplified through two cases in a small basin in the Illinois River Basin. The potential of transforming FFGS into a probabilistic system is also discussed.

A long record (1862–2004) of seasonal rainfall and temperature from the Rome observatory of Collegio Romano are modeled in a nonstationary framework by means of the Generalized Additive Models in Location, Scale and Shape (GAMLSS). Modeling analyses are used to characterize nonstationarities in rainfall and related climate variables. It is shown that the GAMLSS models are able to represent the magnitude and spread in the seasonal time series with parameters which are a smooth function of time. Covariate analyses highlight the role of seasonal and interannual variability of large-scale climate forcing, as reflected in three teleconnection indexes (Atlantic Multidecadal Oscillation, North Atlantic Oscillation, and Mediterranean Index), for modeling seasonal rainfall and temperature over Rome. In particular, the North Atlantic Oscillation is a significant predictor during the winter, while the Mediterranean Index is a significant predictor for almost all seasons.

Summary The Charlotte, North Carolina metropolitan area has experienced extensive urban and suburban growth during the past 40 years, resulting in increasing flood hazards in the region. Record flooding in the urban core of Charlotte occurred on 23 July 1997 from a storm that produced rainfall accumulations of more than 250 mm during an 18 h period, more than doubling the 24 h rainfall maximum in Charlotte, and causing $60 million in property damage and three fatalities. Analyses of the 23 July 1997 storm and flood are based on rainfall and discharge observations from dense networks of rain gages and stream gages maintained by the U.S. Geological Survey and rainfall estimates from two WSR-88D weather radars, both located approximately 150 km from the urban core of Charlotte. This wealth of observations provides an opportunity to address hydrometeorological questions concerning: (1) the accuracy of radar rainfall estimates for extreme, flood-producing rainfall; (2) the space-time variability of extreme, flood-producing rainfall in urban environments; and (3) the effects of urbanization on extreme flood response in urban environments. It is shown that bias-corrected radar rainfall estimates for the 23 July 1997 storm are quite accurate and provide the capability for resolving the fundamental rainfall forcing associated with regional variation in extreme flood response in urban landscapes. Extreme flood response in urban watersheds is characterized by pronounced nonlinearities in runoff production for rainfall accumulations exceeding 50 mm. Extreme flood response is also characterized by large spatial heterogeneities that are tied to the history of urban development. Case study analyses of four additional flood events in the Charlotte metropolitan area are used to assess the robustness of conclusions derived from analyses of the most extreme event in the region and to examine the transition to “upper tail” properties of extreme flood response.

Flood frequency analysis in urban watersheds is complicated by nonstationarities of annual peak records associated with land use change and evolving urban stormwater infrastructure. In this study, a framework for flood frequency analysis is developed based on the Generalized Additive Models for Location, Scale and Shape parameters (GAMLSS), a tool for modeling time series under nonstationary conditions. GAMLSS is applied to annual maximum peak discharge records for Little Sugar Creek, a highly urbanized watershed which drains the urban core of Charlotte, North Carolina. It is shown that GAMLSS is able to describe the variability in the mean and variance of the annual maximum peak discharge by modeling the parameters of the selected parametric distribution as a smooth function of time via cubic splines. Flood frequency analyses for Little Sugar Creek (at a drainage area of 110 km 2 ) show that the maximum flow with a 0.01-annual probability (corresponding to 100-year flood peak under stationary conditions) over the 83-year record has ranged from a minimum unit discharge of 2.1 m 3 s - 1 km - 2 to a maximum of 5.1 m 3 s - 1 km - 2 . An alternative characterization can be made by examining the estimated return interval of the peak discharge that would have an annual exceedance probability of 0.01 under the assumption of stationarity ( 3.2 m 3 s - 1 km - 2 ) . Under nonstationary conditions, alternative definitions of return period should be adapted. Under the GAMLSS model, the return interval of an annual peak discharge of 3.2 m 3 s - 1 km - 2 ranges from a maximum value of more than 5000 years in 1957 to a minimum value of almost 8 years for the present time (2007). The GAMLSS framework is also used to examine the links between population trends and flood frequency, as well as trends in annual maximum rainfall. These analyses are used to examine evolving flood frequency over future decades.

There are large uncertainties associated with radar estimates of rainfall, including systematic errors as well as the random effects from several sources. This study focuses on the modeling of the systematic error component, which can be described mathematically in terms of a conditional expectation function. The authors present two different approaches: non-parametric (kernel-based) and parametric (copula-based). A large sample (more than six years) of rain gauge measurements from a dense network located in south-west England is used as an approximation of the true ground rainfall. These data are complemented with rainfall estimates by a C-band weather radar located at Wardon Hill, which is about 40 km from the catchment. The authors compare the results obtained using the parametric and non-parametric schemes for four temporal scales of hydrologic interest (5 and 15 min, hourly and three-hourly) by means of several different performance indices and discuss the strengths and weaknesses of each approach.

In the quantitative evaluation of radar-rainfall products (maps), rain gauge data are generally used as a good approximation of the true ground rainfall. However, rain gauges provide accurate measurements for a specific location, while radar estimates represent areal averages. Because these sampling discrepancies could introduce noise into the comparisons between these two sensors, they need to be accounted for. In this study, the spatial sampling error is defined as the ratio between the measurements by a single rain gauge and the true areal rainfall, defined as the value obtained by averaging the measurements by an adequate number of gauges within a pixel. Using a non-parametric scheme, the authors characterize its full statistical distribution for several spatial (4, 16 and 36 km2) and temporal (15 min and hourly) scales. To accomplish this task, a large dataset (more than six years) of rain gauge measurements obtained through a highly dense rain gauge network deployed in the Brue catchment in southwest England is used. The authors show that the standard deviation of the spatial sampling error decreases with increasing rainfall intensity and accumulation time and increases with increasing pixel size. Additionally, the authors show how the Laplace distribution could be used to model the distribution of spatial sampling errors for the spatial and temporal scales considered in this study.

Rainfall is characterized by high variability both in space and time. Despite continuous technological progress, the available instruments that are used to measure rainfall across several spatio-temporal scales remain inaccurate. To remedy this situation, scaling relationships of spatial rainfall offer the potential to link the observed or predicted precipitation quantities at one scale to those of interest at other scales. This paper focuses on the estimation of the spatial rainfall scaling functions. Standard scaling analysis constructed by means of the ordinary least squares method often violates such basic assumptions implicit in its use and interpretation as homoschedasticity, independence, and normality of the errors. Consequently, the authors consider alternative regression frameworks i.e. bootstrapping regression, semi parametric linear model, and multilevel normal linear model to show how these different approaches exert a significant impact on the multifractal analysis of radar rainfall. In addition, the uncertainties associated with the construction of the scaling function due solely to the regression procedure are quantified. The radar data come from the polarimetric C-band weather radar located in Rome, Italy, and the scaling properties are computed for a square domain centred on the radar site with a side length of 128 km and a finest resolution of 1 km 2 .