SEARCH

SEARCH BY CITATION

Keywords:

  • extreme value theory;
  • global climate models;
  • global warming;
  • precipitation extremes;
  • reanalysis;
  • uncertainty

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] Recent research on the projection of precipitation extremes has either focused on conceptual physical mechanisms that generate heavy precipitation or rigorous statistical methods that extrapolate tail behavior. However, informing both climate prediction and impact assessment requires concurrent physically and statistically oriented analysis. A combined examination of climate model simulations and observation-based reanalysis data sets suggests more intense and frequent precipitation extremes under 21st-century warming scenarios. Utilization of statistical extreme value theory and resampling-based uncertainty quantification combined with consideration of the Clausius-Clapeyron relationship reveals consistently intensifying trends for precipitation extremes at a global-average scale. However, regional and decadal analyses reveal specific discrepancies in the physical mechanisms governing precipitation extremes, as well as their statistical trends, especially in the tropics. The intensifying trend of precipitation extremes has quantifiable impacts on intensity-duration-frequency curves, which in turn have direct implications for hydraulic engineering design and water-resources management. The larger uncertainties at regional and decadal scales suggest the need for caution during regional-scale adaptation or preparedness decisions. Future research needs to explore the possibility of uncertainty reduction through higher resolution global climate models, statistical or dynamical downscaling, as well as improved understanding of precipitation extremes processes.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] Current climate change mitigation policies, including national resource allocations and international emissions negotiations, are influenced by vulnerabilities to natural hazards (e.g., precipitation extremes) at relatively aggregate scales. Adaptation strategies ranging from engineering decisions such as the design or reinforcement of hydraulic infrastructures to water-resources management, are constructed partially based on probabilistic assessments of extreme hydro-meteorological processes such as severe precipitation events. Thus, there is a need to better understand the potential change of global and regional frequencies of precipitation extremes. However, the extent to which climate model projected precipitation can be translated to decision-relevant metrics for hydraulic infrastructures and water resources management [Milly et al., 2008] has not been explored in detail.

[3] Recent studies of precipitation extremes [Sugiyama et al., 2010; Allan and Soden, 2008; Lenderink and van Meijgaard, 2008; Sillmann and Roeckner, 2008; Kharin et al., 2007; Liu et al., 2009; O'Gorman and Schneider, 2009a, 2009b; Pall et al., 2007; Diffenbaugh et al., 2005; Kunkel et al., 2003] from climate models and observations have either focused on the validity of governing physical mechanisms (e.g., O'Gorman and Schneider [2009a] and Sugiyama et al. [2010]) or sophisticated statistical analysis (e.g., Kharin et al. [2007]). While the Clausius-Clapeyron (CC) relationship provides a physical basis for quantifying increased precipitation extremes in a warming environment, several studies [Liu et al., 2009; O'Gorman and Schneider, 2009a, 2009b; Pall et al., 2007; Diffenbaugh et al., 2005] have pointed to more involved mechanisms, especially for shorter duration extremes. The inability of current generation climate models to adequately resolve cloud microphysics [O'Gorman and Schneider, 2009a, 2009b; Pall et al., 2007; Diffenbaugh et al., 2005], upward velocity in the tropics [O'Gorman and Schneider, 2009a], and oceanic influences [Alexander et al., 2009], are considered among the main factors driving uncertainties in model-simulated regional precipitation extremes [Tebaldi et al., 2006; Tomassini and Jacob, 2009; Wilcox and Donner, 2007; Boberg et al., 2009]. O'Gorman and Schneider [2009a] and Sugiyama et al. [2010] suggest a discrepancy between the credibility of climate model projected precipitation extremes in the tropics versus the extra-tropics, as well as the influence of updraft velocity and the moist-adiabatic temperature lapse rates. A comparison of observed and climate model simulated precipitation extremes appears to suggest that models may underestimate the expected severity of precipitation extremes under climate change [Liu et al., 2009], while the intensification of shorter duration (e.g., hourly) precipitation extremes may exceed expectations [Lenderink and van Meijgaard, 2008]. Pall et al. [2007] suggests that use of the CC relation can aid in predicting changes in extreme precipitation, possibly making them more detectable than mean changes. Thus, understanding of the relationship between extremes and saturation vapor pressure in the atmospheric column may make prediction of precipitation extremes more viable, especially at regional scales of interest to water-resources managers [e.g., Rosenberg et al., 2010].

[4] Studies of mechanisms that generate precipitation extremes have typically relied on percentile based definitions and have been somewhat divorced from statistical extreme value approaches that model low probability events. Extreme value theory (EVT) [Reiss and Thomas, 2007; Coles, 2001] is analogous to the central limit theorem but applies to large deviations or extremes. Corresponding statistical distributions attempt to extrapolate tail behavior from maximum values within a temporal window (e.g., annual maxima) or values exceeding high thresholds (e.g., high percentiles). EVT has been widely used to study climate and hydrologic extremes [Min et al., 2011; Buishand et al., 2008; Kuhn et al., 2007; Katz et al., 2002; Khan et al., 2007; Katz and Brown, 1992]. Kharin et al. [2007] were among the first to develop comprehensive EVT characterizations of temperature and precipitation extremes based on climate model simulations and reanalysis data. EVT (and fitting of other probability density functions) is also an accepted practical approach to translate the quantity of extremes in terms of recurrence interval for water resources management. However, given that the conventional EVT framework is only valid given a stationary environment, one specific challenge to date is how to quantify the frequency of extremes given an unknown pattern of nonstationarity as modeled by various climate models and meteorological reanalyses. For climate impacts and adaption it is also important to understand and translate the trend of changing frequency in terms that are familiar to water resources managers.

[5] In this study, we approach model-simulated precipitation extremes from a hydrological and water-resources perspective: our goal is to identify future direction for hydro-climatic adaption and mitigation. We begin by investigating variability among extreme rainfall frequency from multiple climate models and reanalysis data sets. Given that the EVT-based rainfall frequency is derived from annual maxima, the contribution of convective versus large scale rainfall among different models is also examined. Based on a 30 year moving window, the gradual change of precipitation extremes is examined. In addition to historical reconstructions represented by reanalyses data sets and model hindcasts, the analysis is performed for several climate model simulations forced with several IPCC-SRES greenhouse-gas emissions scenarios. The CC ratio, which can be expressed simply as a function of surface air temperature, is plotted and compared to the depth change rate of precipitation extremes (i.e., the rate of change of extreme rainfall magnitudes that correspond to the same return period but estimated at different temporal windows). To gain further insights for hydrologic impacts, the intensity-duration-frequency (IDF) curves, which can extract attributes of design storms used for hydraulic design and water-resources management, are also built for each model. The methodologies described here are geared toward responding to the challenge posed by Milly et al. [2008]: how can we quantify the frequencies of hydrologic extremes when the assumption of stationarity is no longer hold? The results in this paper are expected to demonstrate the potential of these approaches to characterize the nonstationary behavior of precipitation extremes under projected climate change.

2. Methodology and Data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

2.1. Extreme Value Statistics

[6] Following prior literature [Kharin et al., 2007; Min et al., 2011; Kuhn et al., 2007; Reiss and Thomas, 2007; Coles, 2001; Katz et al., 2002; Khan et al., 2007], the Generalized Extreme Value (GEV) distribution, which is based on the block maxima theory, is adopted to quantify the intensity of extreme precipitation. Given a duration of interest (e.g., 6 hourly data), the annual maximum precipitation (AMP) series are computed from data; these are then used to estimate parameters of a fitted GEV distribution (see Wehner et al. [2010] and Min et al. [2011] for mathematical details). We employ the AMP approach rather than the peak-over-threshold approach [Kuhn et al., 2007; Coles, 2001; Khan et al., 2007], because the i.i.d. assumptions can be better satisfied using annual, likely independent, data points. The AMP approach is also utilized by the U.S. National Weather Service for the development of regional design rainfall thresholds for hydraulic structures [Bonnin et al., 2004], so our approach may be appropriate for such stakeholders. The parameters are used to derive extreme rainfall thresholds corresponding to various levels of exceedance probabilities. In many studies, extreme rainfall statistics are expressed in terms of T year rainfall depth, which represents that, statistically, one annual maximum precipitation event exceeding a given threshold is expected to occur within every nonoverlapping T year length window, given that the system is stationary [Chow et al., 1988]. While the AMP approach is deemed useful and reasonable to estimate the rainfall return period from historic observations, the assumption of stationarity may not hold if the system is in fact nonstationary, as may be the case under climate change. Since we are examining the extreme rainfall statistics from both model projections and meteorological reanalyses in this study, such a limitation must be taken into consideration. To address the potential nonstationary effect, rather than using all temporal data to fit one GEV distribution, GEV parameters are estimated separately for a succession of 30 year moving windows. In doing so, we are able to illustrate the continuous change in extreme rainfall thresholds over time. (As an alternative one could also consider assuming the GEV parameters to be functions of time [see Kharin and Zwiers, 2005]. However, the types of temporal functions need to be specified a priori, and such information may not be available and consistent given the complexity of climate system.) The 30 year standard is selected, because (1) it is expected to smooth out the effects of most multidecadal climate oscillators, and (2) it provides more confidence for low frequency extremes compared to the original 20 year standard of the World Climate Research Program's (WCRP's) Coupled Model Intercomparison Project phase 3 (CMIP3) [see Kharin et al., 2007]. Current data availability also supports the use of 30 year moving windows for most models and reanalysis data sets.

[7] Rather than focusing exclusively on subdaily, daily, or multiday duration precipitation extremes, it is of interest to see variation in extreme rainfall statistics across multiple durations. Thus, the 6, 12, 18, 24, 36, 48, 72, 120, and 240 h AMP series are computed and used to derive thresholds (for daily data only 24, 48, 72, 120, and 240 h AMP series are computed). These extreme rainfall estimates are then illustrated in terms of IDF curves [e.g., Langousis and Veneziano, 2007], which are commonly used in practice for the design of hydraulic structures.

[8] Three possible sources of uncertainty that may be relevant for the application of EVT to climate model simulations are evaluated: (1) applicability of the EVT, in this case the goodness of fit of the GEV distribution; (2) uncertainties inherent in the estimation of GEV parameters from relatively small samples, in this case computed by 1000 member bootstrapping [Efron and Tibshirani, 1994]; and (3) uncertainties caused by disagreement in extreme trends between models and observations.

[9] The GEV parameters are estimated using the maximum likelihood method. We test for the goodness of fit of the extreme value distributions, and if they fail to pass the corresponding statistical tests (Kolmogorov-Smirnov (KS) and Cramer-von Mises (CM)) at a 5% significant level, they are replaced with the empirically based kernel density function. Less than 1% of the total cases fail to pass both goodness-of-fit tests, suggesting the appropriateness of the GEV distribution.

2.2. Conceptual Physical Relationship

[10] The CC relationship operating at relatively fine scales is thought to be the predominant relationship governing precipitation extremes [Liu et al., 2009; O'Gorman and Schneider, 2009a, 2009b; Pall et al., 2007]. While the CC relationship is typically expressed as a function of saturation vapor pressure of water vapor and atmospheric temperature, the ratio of water vapor mass (or density, since the volume does not change) under different temperatures can be approximated based on the ideal gas law.

[11] Let ρv be the density of water vapor, e the vapor pressure, Rv the gas constant for water vapor, and T the temperature (Celsius). From the ideal gas law we have:

  • equation image

For fully saturated air, the vapor pressure e equals saturated vapor pressure es. From the CC relationship, es becomes [Chow et al., 1988]:

  • equation image

By incorporating (1) and (2), the density of saturated water vapor ρsv can be expressed as:

  • equation image

For a given location, if the temperature changes from T1 to T2, then the ratio ρsv2/ρsv1 represents the rate change of saturated vapor density:

  • equation image

[12] If the precipitation extremes are assumed to occur when the water vapor is close to fully saturated, and the amount of precipitation extremes is proportional to es, then the ratio ρsv2/ρsv1 represents the change of precipitation mass (amount) from T1 to T2. Since most of the water vapor is located near the earth's surface, we utilize the annual average surface temperature from each model (climate model or reanalysis) to compute the CC ratio at each grid cell. The relation between saturated surface water vapor and precipitation extremes is also suggested by O'Gorman and Schneider [2009a, 2009b].

2.3. Data Sets

[13] As one major objective of this study is to examine how precipitation extremes change gradually with time and across different durations, qualified data sets must be archived continuously for the whole study period and should be available at the finest temporal resolution (6 hourly). Given this requirement, the National Center for Atmospheric Research (NCAR) Community Climate System Model version 3 (CCSM3, Collins et al. [2006]) is selected, in which the complete 6 hourly precipitation data set is available from the Earth System Grid [Bernholdt et al., 2005] from 1900–1999 for the twentieth century control runs (20C3M), and from 2000–2099 for various projection scenarios. In addition, the Australian Commonwealth Scientific and Industrial Research Organisation Global Climate Model Mk 3.5 (CSIRO, Gordon et al. [2002]) is selected, in which the daily based continuous data is available from the WCRP's CMIP3 archive (1901–2000 for control runs and 2001–2100 for projection runs). Five 21st-century emission scenario runs are used for comparison, including Commit (commitment runs with forcings fixed at year 2000 values), B1 (low CO2 emissions), A1B (moderate CO2 emissions), A2 (high CO2 emissions), and A1FI (fossil fuel intensive emissions, available for CCSM3 only). The fossil fuel intensive A1FI scenario was originally considered unrealistically high, but recent observed emissions were reported to exceed the A1FI trajectories [Raupach et al., 2007]. We note that since it is not required for modeling groups to submit complete finer temporal-scale results for both twentieth and 21st centuries, suitable model outputs at CMIP3 archive are limited to these two models for our stated purposes. Those interested in the comparison of extreme rainfall statistics among a comprehensive set of models for specific time periods (1981–2000, 2046–2065, and 2081–2000) are referred to Kharin et al. [2007].

[14] Several reanalysis data sets are also used in this study, including the National Center for Environmental Prediction (NCEP)-NCAR reanalysis (NCEP1, data available from 1948–2008 [Kalnay et al., 1996]), NCEP-Department of Energy (DOE) AMIP-II reanalysis (NCEP2, data available from 1979–2008 [Kanamitsu et al., 2002]), and the European Centre for Medium-Range Weather Forecasting (ECMWF) Reanalysis (ERA-40, data available from 1958–2001 [Simmons and Gibson, 2000]). The data length is again an essential criterion for selecting suitable data sets. While NCEP1 and ERA-40 are sufficiently long for the computation of multiple 30 year return periods, NCEP2 is only 40 years in length. Reanalysis data sets were chosen as surrogates for rainfall observations as per several considerations: (1) A fair comparison between climate projection and observation must be performed in a spatially uniform fashion, or results will be biased toward regions with more gauge stations; (2) The observational data sets must have a 6 hourly minimum temporal resolution to support the investigation of precipitation extremes under different rainfall durations; and (3) The observational data sets should have consistent temporal and spatial data coverage with minimum missing values. Though reanalysis data sets seem to be the most appropriate for our purpose, it should be noted that reanalysis precipitation is not constrained by gauge observations. Therefore, considerable difference may exist between various reanalysis models and gauge observations (e.g., Zolina et al. [2004] performed a comparative assessment of precipitation extremes over Europe from different reanalyses). For the purpose of climate impact assessment, a greater emphasis should hence be placed on the relative trend from time to time and at a larger spatial scale rather than limiting the attention to the absolute values at each individual grid cell.

[15] The spatial resolution of selected models and reanalyses vary. The CCSM3 adopts a 256 × 128 T-85 Gaussian grid (∼1.4° in space) and the CSIRO has a 192 × 96 T-63 Gaussian grid (∼1.9°). For reanalysis, both NCEP1 and NCEP2 use 192 × 94 grids (∼1.9°). The ERA-40 data has been preinterpolated from a finer 320 × 160 grid and are available on 144 × 73 grids (∼2.5°). Without introducing extra assumptions of spatial correlation, the extreme precipitation statistics and CC ratios (equation (4)) are computed on each data set's original grid and summarized over large regions for comparison.

3. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

3.1. Difference Between Models and Reanalyses

[16] We start by computing the 3, 5, 10, 30, 50, and 100 year return level precipitation extremes for 6, 12, 18, 24, 36, 48, 72, 120, and 240 h duration at each grid cell using a 30 year moving window. It should be noted that since the five 21st-century simulations are continued from the twentieth century control runs, the 1970–2000 20C3M annual maxima are adopted as the initial values to support the moving window analysis into the 21st century. For clarity, the extreme rainfall estimates are labeled by the ending year of the window (e.g., year 2039 estimate was computed from 2010–2039 annual precipitation maxima).

[17] The year 1999 estimates (1970–1999 window) are illustrated in Figure 1 to provide a general understanding of precipitation extremes. The 24 h, 30 year precipitation extremes from NCEP1, ERA-40, CCSM3, and CSIRO, associated with their corresponding 10% and 90% bootstrapping uncertainty bounds are illustrated. The 10% and 90% bounds are computed based on a 1000 member bootstrap [Efron and Tibshirani, 1994; Kharin et al., 2007]. The largest difference occurs near the tropics (30S ∼ 30N), and the bootstrapping uncertainty is not as large as the difference across models and reanalyses.

image

Figure 1. The 24 h duration, 30 year return period precipitation extremes estimated from the 1970–1999 annual maxima. The 10% and 90% uncertainty bounds are computed based on a 1000 member bootstrap.

Download figure to PowerPoint

[18] To assist in further interpretation, the ERA-40, CCSM3, and CSIRO estimates shown in Figure 1 are bilinearly interpolated to the NCEP1 resolution for direct comparison. (We note that the interpolated data sets were used only in generating Figure 2, discussed next.) Percentage difference, which is defined as 100*(A − B)/[(A + B)/2] between variables A and B, is calculated for each grid cell and illustrated in Figure 2. The latitudinal averages are also shown. As expected [Kharin et al., 2007; O'Gorman and Schneider, 2009a, 2009b], the relative agreement of the climate model simulations with reanalysis data sets, as well as between the climate models at the extratropics (higher than 30 degree latitudes), contrast with the disagreement in the tropics. The disagreement in the tropics between model-model, model-reanalysis, and reanalysis-reanalysis pairs in Figure 2 suggests that future research may be needed to understand and improve modeling of the physics that drive tropical precipitation extremes. The largest difference in tropical extremes appears to occur between the two reanalysis data sets (this verified in Table 1, which is discussed in the next paragraph), while the two climate models show relatively better agreement with each other. The difference of spatial resolution among data sets may introduce another source of bias in extreme precipitation statistics; however, Chen and Knutson [2008] and Kharin et al. [2007] examined the difference between NCEP2 and ERA-40 and concluded that return values were reduced only by a few percentage points. The majority of variation in the tropics, then, may not be a result of the differences in spatial resolution.

image

Figure 2. Percentage difference of 24 h duration, 30 year return period precipitation estimated from 1970 to 1999 extremes from model and reanalysis data sets.

Download figure to PowerPoint

Table 1. Summary of Precipitation Extremes and Their Uncertainty Bounds Compared Across Pairs of Climate Models and Reanalysisa
ModelsRange of Percentage Difference
<−50%−50% ∼ −20%−20% ∼ 20%20% ∼ 50%>50%
  • a

    The comparison between 10% lower bounds is reported in subscripts, while the comparison between 90% upper bounds is in superscripts.

ERA-40∼NCEP15.85.17.816.415.317.441.544.335.610.610.812.025.724.527.2
Extratropics7.86.810.724.022.025.157.661.449.38.06.811.72.63.03.2
CCSM3∼NCEP15.94.98.617.216.217.648.251.441.415.914.617.812.812.914.6
Extratropics5.24.58.220.919.521.658.061.849.712.09.715.93.84.54.6
CSIRO∼NCEP17.06.39.717.516.218.442.544.936.714.613.815.818.418.819.4
Extratropics8.16.911.823.922.224.655.058.946.19.78.013.13.44.04.4
CCSM3∼ERA-4023.121.025.211.410.812.745.949.138.815.614.817.64.04.35.7
Extratropics1.20.73.210.48.013.762.767.352.521.820.524.43.83.56.3
CSIRO∼ERA-4018.115.322.516.915.018.251.056.941.811.710.313.52.22.44.0
Extratropics1.30.83.614.611.218.456.673.255.215.412.918.42.11.84.3
CSIRO∼CCSM34.74.47.217.415.119.553.156.945.616.616.117.68.27.510.1
Extratropics2.92.45.821.018.023.665.171.254.510.08.113.60.90.32.6

[19] Following Figure 2, the percentage difference of precipitation extremes between NCEP1, ERA-40, CCSM3, and CSIRO is summarized in Table 1. We divide the percentage difference into five ranges (<−50%, −50% ∼ −20%, −20% ∼ 20%, 20% ∼ 50%, and >50%) and report the percentage of grid cells that lie in each range. The percentage difference is also computed for the 10% and 90% uncertainty bounds shown in Figure 1. Around 40% ∼ 50% of the total grid cells fall in the −20% ∼ 20% range and around 70% ∼ 80% fall in the −50% ∼ 50% range. Perhaps the most interesting insight from Table 1 is that the maximum similarity is found between CSIRO and CCSM3, while the least is between NCEP1 and ERA-40. It may be nonintuitive that the largest disagreement is not between models and reanalysis. This is interesting and may suggest the need for deeper exploration of precipitation physics and parameterization schemes, as well as improved quantitative precipitation estimates for generating more accurate and more consistent observations. Studies on differences between the two families of reanalysis data sets as compared to observational estimates of variables of interest can be found in the works of Trenberth et al. [2001] (tropical temperature) and Zolina et al. [2004] (European precipitation extremes). Further exploration of reanalysis quality, specifically with regard to tropical precipitation extremes, may be of value. With the tropical region excluded (30S ∼ 30N), approximately 55% ∼ 65% of the total grid cells fall in the −20% ∼ 20% range, and 85% ∼ 95% of the total grid cells fall in the −50% ∼ 50% range. The largest observation/model inconsistency for precipitation extremes is in the tropical regions. Though the above illustrations are only based on the 24 h, 30 year precipitation extremes, we note that similar patterns are observed for other durations, recurrence intervals, and temporal windows as well. The annual maxima and the corresponding derived rainfall estimates are used in the following analysis.

3.2. Contribution of Convective Precipitation in Annual Maximum Rainfall

[20] The reanalysis data sets and climate model outputs provide a unique opportunity to examine the influence of modeled convective precipitation on rainfall extremes, even though they are not direct observations. The percentage contribution of convective rainfall in the AMP series is examined next. These percentage contributions are illustrated in Figure 3, in which the average contribution of convective precipitation in each of 1979–1999 NCEP1, NCEP2, ERA-40, and CCSM3 annual maximum is computed at all latitudinal bands. We note that the CSIRO convective rainfall outputs are not achieved continuously at a daily scale within WCRP's CMIP3 and hence are excluded from this part of analysis.

image

Figure 3. Contribution of convective precipitation to the annual maxima according to climate model simulations and reanalysis data sets.

Download figure to PowerPoint

[21] In Figure 3a, the average percentage contribution of convective rainfall within 6 h annual maxima along latitudinal bands is shown to vary substantially. Convection is generally considered to be the predominant generative mechanism of precipitation extremes, especially for shorter subdaily duration (with exceptions; see Scinocca and McFarlane [2004]). While convective contributions appear to be more dominant in the tropics, the significant differences between reanalysis and models may diminish the value of any generic insights. CCSM3-simulated convective contributions to 6 h extremes are in between NCEP1, NCEP2 (nearly 90%), and ERA-40 (barely 50%). When combined with the uncertainty in the tropics, these results suggest that significant improvements are needed in our understanding of precipitation processes beyond what is suggested in the literature [Allan and Soden, 2008; Lenderink and van Meijgaard, 2008; O'Gorman and Schneider, 2009a, 2009b; Pall et al., 2007; Diffenbaugh et al., 2005; Alexander et al., 2009].

[22] In Figure 3b, the global area-weighted average of convective rainfall contribution is computed for various durations (note that duration is plotted in log-scale). The fact that NCEP1, ERA-40, and CCSM3 are nearly parallel to each other may suggest different parameterizations with constant bias, while the different behavior (nearly constant versus almost linear growth) of the subdaily versus daily or greater than daily extremes may point to differences in the underlying mechanisms generating shorter and longer duration extremes. The contribution of convection to subdaily duration precipitation extremes appear relatively constant with changing durations (even though the two reanalyses and –CCSM3 suggest different levels), but for longer (than one day) duration extremes, the contributions from convection appear to increase linearly (in log-scale) with duration. This suggests that the underlying mechanisms for shorter and longer duration extremes may be fundamentally different. While convective rainfall contributes the most in NCEP1, its contribution is least in ERA-40. Once again, the maximum difference occurs between two reanalysis data sets rather than between a reanalysis data set and CCSM3; the exact cause for this discrepancy may need to be investigated. Possible causes appear to be differences either in the precipitation parameterization schemes or the actual observations used to drive the forecasting models, which generate the two reanalysis data sets.

[23] The geographical pattern of convective contribution in 6 h annual maxima is plotted in Figure 4 for four models. The spatial variation is significant, with NCEP2 suggesting predominance of convection in most land areas but ERA-40 and NCEP1 showing convective precipitation mainly over tropical oceans. For NCEP1 (Figure 4a), convective precipitation has the largest contribution in the entire tropical band (30S∼30N, including both land and ocean), while for NCEP2 (Figure 4b), convective precipitation has a larger contribution on the land than on the ocean surface. Compared to NCEP1, the NCEP2 shows more convective activity over land, especially in the extratropics and in the northern hemisphere. The CCSM3 simulations (Figure 4c) appear to exhibit a noticeable discontinuity between land and ocean around continental Africa, south Asia, and northern Australia. The likely causes of this discontinuity could be due to the difference between land and ocean models in the CCSM3. The difference in surface heat capacity between ocean and land may trigger convection in different ways in the CCSM3 cumulus parameterization. Also, there may be a diurnal effect over land that is absent over ocean. ERA-40 (Figure 4d) exhibits the largest precipitation extreme depths, but the smallest contribution from convection. The precipitation extremes in ERA-40 are primarily controlled by a large-scale precipitation mechanism. We note that while Figure 4 shows the 6 h annual maxima results, these patterns are similar for other durations.

image

Figure 4. Regional variability in the average contribution of convective precipitation to the 6 h annual maxima precipitation extremes.

Download figure to PowerPoint

3.3. Trend of Rainfall Extremes Under Warming Scenarios

[24] In order to understand variation of precipitation extremes over time, the previous analysis is performed repeatedly with a 30 year moving window. For each 30 year period, the average temperature (required in equation (4) for computing the CC ratio), GEV parameters, and 30 year rainfall depth are estimated. The year 1999 values (corresponding to the 1970–1999 window) are selected as a baseline for comparison. By setting T1 as the average surface temperature from 1970–1999, the CC ratio is computed for every grid and for all windows. Since the CC ratio represents the theoretical increase/decrease of saturated vapor density due to temperature change, it can be regarded as a theoretical reference of extreme rainfall variation to year 1999. Similarly, the depth ratios, which are defined as extreme rainfall estimates normalized by the corresponding year 1999 baseline values, are computed for comparison. The depth ratios represent the rate of change of extreme rainfall magnitudes that correspond to the same return period but estimated at different temporal windows. In addition, by referring to the year 1999 baseline values, we further compute the corresponding return periods (frequency) for different time periods.

[25] The analysis is performed for the two reanalyses (NCEP1, ERA-40) and two climate models (CCSM3, CSIRO). NCEP2 is not included here since its data coverage is insufficient for continuous analysis. Taking 24 h duration as an example, the global area-weighted medians of CC ratio, depth ratio, and return period are shown in Figure 5, with insets emphasizing the overlapping period from 1987–2008. While the median is suggested as a proper measure by Kharin et al. [2007], the difference in grid sizes along various latitudes must be adjusted; otherwise the median will be biased toward the extratropics. To make a proper correction, the area-weighted median is computed instead. The area-weighted median [see Yin et al., 1996] is a general form of median in which each grid value is assigned a corresponding areal weight. By sorting the values and using the weights as widths, the 50% percentile is identified to be the area-weighted median. If all grids are equal sizes, the weighted median will be equivalent to the common median.

image

Figure 5. Intensification of precipitation extremes at global scales based on climate models and reanalysis data sets and the significance of Clausius-Clapeyron relationship.

Download figure to PowerPoint

[26] As shown in Figures 5a and 5b, the Clausius-Clapeyron ratio almost exactly mirrors the depth ratio and it provides a theoretical reference of global precipitation extremes intensification owing to atmospheric warming. However, the trends from NCEP1 in the prior decade do not agree with the trends in climate model hindcasts and must be further investigated. The climate model hindcasts suggest that early twentieth century extreme intensities were less intense (about 96% of current) and less frequent (current 30 year intensities, or occurring with a probability of 1/30, were about 40 year then). Climate models project precipitation extremes to be more intense and frequent in the future. The worst case A1FI scenario projects an intensification of 30% with the 30 year rainfall event becoming as frequent as the current 7 year event. We note that the scenarios show considerable difference, suggesting that emissions may heavily influence the intensification and larger frequencies of precipitation extremes in the future. The Commit scenario, which sustains atmospheric concentration at year 1999 levels, shows the intensification in the future owing to both system delay of stabilization and temporal memory in the moving average calculation; the rainfall extreme stabilizes after approximately three decades. The intensity and frequency projections display similar trends but have considerably more uncertainties and variability at regional scales. The intensification and increasing frequencies of precipitation extremes in a warming environment, as well as the correspondence with the CC relation, are clearly illustrated in this approach, which also relies on extreme value statistics.

[27] Given the potential limitation of inferences regarding precipitation extremes at tropical regions (as discussed in section 3.1), Figure 6 illustrates the area weighted median of extratropical regions (90S ∼ 30S and 30N ∼ 90N). The general trends are similar to Figure 5, but the Clausius-Clapeyron provides a higher ratio in the extratropics. This may appear nonintuitive since the reanalysis and model pairs appear to match better in the extratropics (see Figures 1 and 2, as well as O'Gorman and Schneider [2009a, 2009b]). It suggests that equation (4) and the use of surface temperature may be overly simple. Because this may be the case, regional and local scale trends may not be captured, even though the relationship appears reasonable at a global-average scale. Figure 7 shows the results for Europe. We selected Europe as a case study because, visually, there is a good match between the various reanalysis and climate model pairs. We observe that the intensification of precipitation extreme trends does not appear obvious from NCEP1 but is relatively clear from ERA-40. The general trends agree with Figures 5 and 6, while the variability is larger compared to the others, probably because there are considerably fewer grid cells.

image

Figure 6. Intensification of precipitation extremes at the extratropical regions based on climate models and reanalysis data sets as well as the significance of Clausius-Clapeyron relationship.

Download figure to PowerPoint

image

Figure 7. Intensification of precipitation extremes in the continental Europe.

Download figure to PowerPoint

[28] The general temporal trends are observed in Figures 57, and spatial variability is shown to be large. An example is shown in Figure 8, in which the 6 h duration precipitation extremes from NCEP1 and CCSM3 are illustrated in a more detailed fashion for the extratropical land (90S ∼ 30S, 30N ∼ 90N, land). By referring to the year 1999 30 year return level estimates at each grid cell, Figure 8a displays histograms (spanning all grid cells) of the corresponding return levels of the year 1977 (1948–1977 window) and year 2008 (1979–2008 window) NCEP1 precipitation extremes. Less than half (48%) of the grid cells exhibit precipitation extremes that are less than the 30 year level at year 1977, implying that these extremes were less frequent historically. Conversely, more than half (59%) of the grid cells show precipitation extremes less than 30 year levels during year 2008, implying more frequent extremes. However, it should be noticed that spatial variability is large, and the trend may be flat or opposite within some individual grid cells. Similarly, Figure 8b compares the year 2099 CCSM3 precipitation extremes of five scenarios. Spatially empirical probability density functions are estimated. The degree of intensification follows the order of projected CO2 emissions.

image

Figure 8. Increasing frequency of 6 h duration 30 year return periods of precipitation extremes over the last several decades as well as for the rest of the 21st century in the extratropical land.

Download figure to PowerPoint

[29] To understand the spatial variability, the 30 year return period, 24 h duration CCSM3 and CSIRO depth ratios at year 2099 (2070–2099 window) are illustrated for four emission scenarios (Commit, B1, A1B, and A2) in Figures 9 and 10, respectively. A nine-neighbor grid smoothing is performed for easier visualization and identification of regional patterns. The model-projected intensification trends grow stronger with increases in projected CO2 emissions (as defined in the IPCC-SRES scenarios). Increased intensities of design storms, their large variability across geographical regions, and the spotty nature of the visuals (reflecting the large spatial variability of precipitation and their extremes, as well as the estimation sensitivities) are clear from the maps. While the intensification trends projected by CCSM3 and CSIRO agree relatively well at global and continental scales, the projections are inconsistent for relatively finer resolution regional scales, even after smoothing.

image

Figure 9. Intensification of precipitation extremes at the end of the century according to CCSM3.

Download figure to PowerPoint

image

Figure 10. Intensification of precipitation extremes at the end of the century according to CSIRO.

Download figure to PowerPoint

3.4. Intensity-Duration-Frequency Relationships of Rainfall Extremes

[30] Precipitation intensity-duration-frequency (IDF) curves are frequently used in hydraulic design and water-resources management [Houghtalen et al., 2009; Dunne and Bergere, 1978; Koutsoyiannis et al., 1998]. By plotting the average rainfall intensity (total depth divided by duration) versus duration, it is empirically observed that rainfall intensities with the same frequency are negatively correlated to duration on the log-log scale. We note that the duration here refers to the temporal window used to compute annual maxima instead of the actual storm durations. In other words, IDF curves are the empirical relationship of rainfall extremes across different durations from actual observation. Whether this relationship holds for reanalyses and climate projections is seldom discussed, partially due to the challenge of analyzing rainfall extremes across a wide range of rainfall durations. In order to support the construction of IDF curves, temporally higher resolution data sets must be available.

[31] Building on the 6, 12, 18, 24, 36, 48, 72, 120, and 240 h extratropical area-weighted median of year 1999 (1970–1999 window) rainfall average intensities, Figure 11 shows the NCEP1, ERA-40, CCSM3, and CSIRO IDF curves for 3, 5, 10, 30, 50, and 100 year return periods (note that the CSIRO data can only support the computation of extremes at durations of a day or more). The linear patterns are preserved in all cases and the curves are more or less parallel to each other, suggesting that the scales of rainfall extremes across different durations seem reasonable. IDF curves can be utilized when investigating precipitation extremes with an arbitrary duration, which has not been encountered before or computed previously. These curves are potentially helpful to understand the behavior of rainfall extremes from subdaily, daily, to multiday scales.

image

Figure 11. Intensity-Duration-Frequency (IDF) curves of precipitation extremes for various frequencies and durations.

Download figure to PowerPoint

[32] Another comparison is shown in Figure 12. Figure 12a compares the year 1999 30 year IDF curves of NCEP1, ERA-40, CCSM3, and CSIRO. The curves appear fairly linear on the log-log scale. The differences between the two climate models are relatively small compared to the differences between the two reanalysis data sets. NCEP1 rainfall intensity is less than CCSM3 in shorter durations, while it becomes larger than CCSM3 for longer durations. Focusing on CCSM3, Figure 12b compares various IDF curves at year 2099 under all emission scenarios. The 30 year IDF curves from the various emission scenarios in the end of the century are parallel to the twentieth century control runs in the log-log graph. Parallel IDF curves in the log-log plots imply a constant ratio, which in turn may translate to a safety factor for engineering design and water-resources management in the context of climate change adaptation. However, the larger differences among climate models and reanalysis (Figure 12a) point to uncertainty which must be characterized and/or ideally reduced prior to making risk informed decisions. The climate change-influenced evolution of IDF curves directly illustrate that water management can no longer assume stationarity [Milly et al., 2008]. Figures 12c and 12d show two examples of IDF curves generated at regional scales, specifically, North America and Europe. While IDF curves could be developed for any region or locale of interest, the relatively low resolutions of extreme precipitation processes within global climate models and the corresponding increase in uncertainty of model-based precipitation extremes projections at higher resolutions limit the credibility of local or even regional IDF curves. Improving the credibility of higher resolution IDF curves may be possible through higher resolution global climate models, and/or through dynamical or statistical downscaling of the global model outputs and/or through improved understanding of extreme precipitation processes. However, the ability of higher resolution global climate models to reduce uncertainty in precipitation extremes remains a hypothesis to be tested (despite reports of initial success [Wehner et al., 2010]) while downscaling may cause additional cascading uncertainties [Schiermeier, 2010].

image

Figure 12. (a and b) Intensity-Duration-Frequency (IDF) curves for precipitation extremes from climate models and reanalysis data sets. (c and d) IDF curves of precipitation extremes for North America and Europe.

Download figure to PowerPoint

4. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

[33] By examining the EVT based precipitation estimates and the physical based CC ratios concurrently, it is shown that while precipitation extremes are projected to continually intensify in the 21st century, according to known physical mechanisms in a spatially aggregate (e.g., global average) sense, significant uncertainties exist in both models and observations, as well as how the models and reanalysis handle precipitation processes, especially at regional scales. The reanalysis and climate models studied here show large discrepancies over the tropics, not just in terms of the intensities of short duration precipitation extremes, but also in terms of the contribution of convective versus large scale precipitation to extremes. The categorization of precipitation into large-scale versus convective suffers from differences in semantics and tends to have a degree of arbitrariness. In addition, the resolution of processes within climate and reanalysis models would impact the categorization. Convection schemes simulated by climate models typically do not attempt to simulate the growth and decay of convection but provide an equilibrium response. An interpretation of the results presented here needs to be informed by and aware of these issues. The central tendencies of the reanalysis and modeled precipitation extreme trends at aggregate scales tend to agree with each other over the last two decades and follow the CC relation quite well. Uncertainties remain significant at regional and decadal scales, but may be quantifiable through statistical approaches such as resampling techniques or Bayesian methods. Changes in the IDF curves, which are typically used for the design of hydraulic infrastructures and for water resources management, may offer guidance for adaptation in a nonstationary environment. Additionally, while the current (IPCC AR4) generation of climate models may not be able to adequately resolve fine-scale processes and hence may not reliably simulate precipitation extremes or generate credible IDF curves at regional scales, higher-resolution CMIP5 simulations may offer new opportunities. In addition, higher-resolution climate models geared for decadal or regional analysis [e.g., see Shukla et al., 2009] may be able to resolve processes relevant for location-specific IDF curves. However, the possible improvement in credibility of precipitation extremes at higher resolution as a function of climate model resolutions and corresponding improvements in physics or parameterizations remains a hypothesis to be tested. Our methods and results with IPCC AR4/CMIP3 data sets presented here may offer a benchmark for comparisons and hypothesis testing. Future researchers may wish to assess the improvement in regional estimates of precipitation extremes from these higher resolution models toward potentially developing regional- and local-scale IDF curves, which in turn may be valuable for regional and local water-resources management applications. While the current study has examined annual maxima, it may also be of interest for future researchers to attempt to delineate other aspects of precipitation extremes. For example, in many regions outside of the tropics, seasonal maxima may be of interest, as winter extremes may exhibit interesting and different features.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

[34] This research was funded by the Laboratory Directed Research and Development (LDRD) Program of the Oak Ridge National Laboratory (ORNL), which in turn is managed by UT-Battelle, LLC, for the U.S. Department of Energy under contract DE-AC05-00OR22725. The United States Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The authors thank Dave Bader, Evan Kodra, Cheng Liu, Richard Medina, and Karsten Steinhaeuser for helpful comments and suggestions.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methodology and Data
  5. 3. Results and Discussion
  6. 4. Conclusion
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
jgrd17081-sup-0001-t01.txtplain text document1KTab-delimited Table 1.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.