Abstract

Trends in extreme temperature and precipitation in two regional climate model simulations forced by two global climate models are compared with observed trends over the western United States. The observed temperature extremes show substantial and statistically significant trends across the western United States during the late twentieth century, with consistent results among individual stations. The two regional climate models simulate temporal trends that are consistent with the observed trends and reflect the anthropogenic warming signal. In contrast, no such clear trends or correspondence between the observations and simulations is found for extreme precipitation, likely resulting from the dominance of the natural variability over systematic climate change during the period. However, further analysis of the variability of precipitation extremes shows strong correspondence between the observed precipitation indices and increasing oceanic Niño index (ONI), with regionally coherent patterns found for the U.S. Northwest and Southwest. Both regional climate simulations reproduce the observed relationship with ONI, indicating that the models can represent the large-scale climatic links with extreme precipitation. The regional climate model simulations use the Weather Research and Forecasting (WRF) Model and Hadley Centre Regional Model (HadRM) forced by the ECHAM5 and the Hadley Centre Climate Model (HadCM) global models for the 1970–2007 time period. Comparisons are made to station observations from the Historical Climatology Network (HCN) locations over the western United States. This study focused on temperature and precipitation extreme indices recommended by the Expert Team on Climate Change Detection Monitoring and Indices (ETCCDMI).

1. Introduction

Recent research has shown that an increased frequency of many types of extreme events has already occurred in the twentieth century in many regions worldwide (Brown et al. 2008) and over North America in particular (Thomas et al. 2008; Peterson et al. 2008; DeGaetano 2009). A recent Intergovermental Panel on Climate Change (IPCC) report concludes that “it is very likely that there has been an overall decrease in the number of cold days and nights and very likely that there has been an overall increase in the number of warm days and nights in most regions” around the world (Field et al. 2012). Trends in precipitation are less certain, with “a likely increase in observed heavy precipitation in many regions in North America, despite statistically non-significant trends and some decreases in some subregions” (Field et al. 2012). For western North America, significant positive trends are observed only locally in the Pacific Northwest (PNW) with smaller positive trends and scattered negative trends elsewhere (DeGaetano 2009; Mass et al. 2011). These changes are broadly consistent with the anticipated effects of anthropogenic climate change (Gutowski et al. 2008). It is unclear, however, whether local trends are discernible from the natural variability in the climate. We will address this problem in the current paper by comparing observed trends to trends simulated by global and regional climate models.

The climate of the western United States is rather diverse. Within a few hundred kilometers, one can go from a marine-type climate to rain forests, glaciers, desert, and near desert conditions. This diversity in climate comes mostly from weather systems interactions with complex terrain that ranges from sea–land interfaces to mountainous ranges (Fig. 1). These interactions lead to small-scale weather features such as land–sea breezes, leeside wind storms, gap winds, orographic precipitation, or rain shadows that play a considerable role in determining the climate and weather at regional and local scales (Mass et al. 2002). Global climate models are unable to simulate such small-scale weather features. To simulate such small features, a realistic representation of the local complex terrain and the heterogeneous land surfaces is needed. Therefore, the use of limited-area regional climate models with horizontal resolutions on the order of at least tens of kilometers is required in order to represent the physical processes controlling local climate and extreme temperature and precipitation events (Mass et al. 2002; Leung et al. 2004; Duffy et al. 2006; Christensen et al. 2007; Salathé et al. 2010).

Fig. 1.

(left) WRF Model domains 1 and 2 and HadRM domain and (right) HCN stations in the states of California, Idaho, Nevada, Oregon, and Washington. Shadings represent terrain height (m) for the corresponding WRF domain. Grid spacing for each domain is as follows: WRF domain 1, 108 km; WRF domain 2, 36 km; and HadRM domain, 25 km.

Fig. 1.

(left) WRF Model domains 1 and 2 and HadRM domain and (right) HCN stations in the states of California, Idaho, Nevada, Oregon, and Washington. Shadings represent terrain height (m) for the corresponding WRF domain. Grid spacing for each domain is as follows: WRF domain 1, 108 km; WRF domain 2, 36 km; and HadRM domain, 25 km.

It is important to evaluate to what extent the regional climate models capture the observed trends and variability in extremes before examining future changes. Several high-resolution modeling studies have been conducted around the world (e.g., Huntingford et al. 2003; Fowler et al. 2005; Buonomo et al. 2007; Beniston et al. 2007; Alexander and Arblaster 2009) and many more are now ongoing. More particularly, for the U.S. Pacific Northwest, Dulière et al. (2011) examined the National Centers for Environmental Prediction (NCEP)/Department of Energy (DOE) Reanalysis 2 (Kalnay et al. 1996; Kanamitsu et al. 2002), the Weather Research and Forecasting (WRF) Model, and Hadley Centre Regional Model (HadRM) simulations using the reanalysis fields as the boundary conditions. These simulations were compared to observations from the Historical Climatology Network (HCN) over a 5-yr period. The regional models showed substantial improvement over the reanalysis in terms of temperature and precipitation extremes. Since the reanalysis incorporates the available observations at the time of the processing, the resulting large-scale fields represent a close approximation to the actual state of the atmosphere including both historic patterns of daily weather and interannual variability. Thus, these results show that regional climate model simulations can reproduce the observed distribution of extreme events when realistic large-scale boundary conditions are used.

Changes in extreme events as a result of human influences on the climate will depend on the sensitivity of extremes to global climate change and our ability to project these changes depends on the ability of global and regional models to represent the relevant atmosphere and land surface processes. In this paper, we examine whether simulations forced only by the historically observed radiative forcing can reproduce the observed trends in extreme temperature and precipitation. It is possible that the observed trends are driven by anthropogenic influences on physical processes, such as indirect aerosol forcing, that are not well represented in current models (Jiang et al. 2012) and thus might emerge in simulations with more advanced models. Alternatively, observed trends may be the result of natural variability and not a direct response to anthropogenic forcing on the climate.

To this end, we use the same two regional models (WRF and HadRM) as in Dulière et al. (2011) but driven by global climate model simulations for the 38-yr time period 1970–2007. The initial and lateral boundary conditions for the WRF Model and HadRM were interpolated from the Max Planck Institute for Meteorology (MPI-M) ECHAM5 and a version of the Hadley Centre's third generation coupled ocean–atmosphere general circulation model (HadCM3Q0) data, respectively (see Zhang et al. 2009 for details). The global climate simulations are taken from phase 3 of the Climate Model Intercomparison Project (CMIP3), which specifies the historic radiative forcing including observed solar, volcanic, and greenhouse gas forcing. The models are otherwise unconstrained by observations and generate independent realizations of natural or internal variations from daily weather to decadal variability. Thus, to the extent that observed trends are for the most part attributable to historic radiative forcing, and from anthropogenic effects in particular, the model simulations should be able to reproduce them. If, however, observed trends are dominated by natural variability, then the simulated trends would not correspond with observations. Note that the simulated results depend on both the global model, which provides the large-scale temperature and circulation, and regional model, which simulates local feedbacks and terrain effects.

We expect natural climate variability to play an important role in the trends of climate parameters over periods of a few decades. In particular, El Niño–Southern Oscillation (ENSO) and the Pacific decadal oscillation (PDO) have substantial influences on temperature and precipitation variability in western North America [see Gershunov (1998) and Cayan et al. (1999) regarding ENSO, and Mantua et al (1997) and Gershunov and Barnett (1998) regarding decadal variability]. Given that substantial regional warming attributable to greenhouse gas emissions has been observed only since approximately 1970 (see, e.g., Tebaldi et al. 2012), the time scale of anthropogenic climate change and of natural variability are similar. This is especially true for the PDO, which was in a persistent warm phase during the period 1977–97. Thus, the observed trends in extreme events since the 1970s are likely a result of the combined effects of climate change and decadal variability. Furthermore, since there is little anthropogenic trend prior to the 1970s, extending the period earlier would not likely improve the detection of anthropogenic trends.

We examine several temperature and precipitation extreme indices, most of them recommended by the Expert Team for Climate Change Detection Monitoring and Indices (ETCCDMI), and compare the simulated values to values computed from the HCN observations. This work is organized as follows. The regional models and experimental design are briefly described in sections 2 and 3, respectively. The methodology is presented in section 4 and a comparison of the models simulations with observations is discussed in section 5. Major conclusions and discussions are presented in section 6.

2. Models descriptions

a. WRF Model

The WRF Model is a state-of-the-art, mesoscale numerical weather model designed for both short-term weather forecasting and long-term climate simulation (http://www.wrf-model.org). The Advanced Research WRF (ARW-WRF) Modeling system includes a large selection of physical parameterizations suitable for a broad spectrum of applications and scales ranging from meters to thousands of kilometers. The physics package includes microphysics, cumulus parameterization, planetary boundary layer (PBL), land surface models (LSMs), and longwave and shortwave radiation (Skamarock et al. 2005). The full nonhydrostatic form of the continuity equation is solved in the WRF Model.

In this work, the microphysics and convective parameterizations used were the WRF single-moment 5-class (WSM5) scheme (Hong et al. 2004) and the Kain–Fritsch scheme (Kain and Fritsch 1993), respectively. The WSM5 microphysics explicitly resolves water vapor, cloud water, rain, cloud ice, and snow. The Kain–Fritsch convective parameterization utilizes a simple cloud model with moist updrafts and downdrafts that includes the effects of detrainment and entrainment. The land surface model used was the NCEP, Oregon State University, Air Force, and Hydrologic Research Laboratory (NOAH) 4-layer soil temperature and moisture model with canopy moisture and snow cover prediction (Chen and Dudhia 2001). The LSM includes root zone, evapotranspiration, soil drainage, and runoff, taking into account vegetation categories, monthly vegetation fraction, and soil texture. The PBL parameterization used was the Yonsei University (YSU) scheme (Hong et al. 2006), which is an updated version of Hong and Pan (1996). This scheme includes countergradient terms to represent heat and moisture fluxes resulting from both local and nonlocal gradients. Atmospheric shortwave and longwave radiations were computed by the National Center for Atmospheric Research (NCAR) Community Atmosphere Model (CAM) shortwave scheme and longwave scheme (Collins et al. 2004), respectively.

b. HadRM model

The HadRM used in this study is a version of the third-generation regional climate model developed at the Met Office Hadley Centre (HadRM3P; Jones et al. 2004). It is a limited-area high-resolution version of HadAM3H, an improved version of the atmospheric component of the latest Hadley Centre coupled atmosphere–ocean coupled general circulation model (AOGCM) HadCM3 (Gordon et al. 2000; Johns et al. 2003).

HadRM incorporates a simplified hydrostatic version of the full primitive equations. Model parameterizations include dynamical flow, horizontal diffusion, clouds and precipitation, radiative processes, gravity wave drag, land surface, and deep soil (Jones et al. 2004). A mass flux penetrative convective scheme (Gregory and Rowntree 1990) is used with an explicit downdraft (Gregory and Allen 1991). It includes the direct impact of vertical convection on heat, moisture, and momentum (Gregory et al. 1997). The boundary layer parameterization uses a first-order turbulent mixing scheme that vertically mixes the conserved thermodynamic variables and momentum (Smith 1990). The land surface model used is based on the Met Office Surface Exchange Scheme (MOSES; Cox et al. 1999) with the inclusion of a radiative (instead of a conductive) heat coupling of vegetated surfaces to the underlying soil. It includes evapotranspiration, soil drainage, and surface runoff and takes into account climatological vegetation type and fractional cover within a grid box. A radiation scheme computes shortwave and longwave fluxes as a function of temperature, water vapor, ozone, carbon dioxide, clouds, and other trace gases. For detailed information, please refer to Jones et al. (2004).

The horizontal resolution of the HadRM model grid is 0.22° × 0.22° (although a resolution of 0.44° × 0.44° is also available). The HadRM latitude–longitude grid is rotated in such a way that the equator lies inside the region of interest. This permits a quasi-uniform gridbox area over the region of interest with a minimum horizontal resolution of approximately 25 km at the rotated equator.

HadRM was released as part of the Providing Regional Climates for Impacts Studies (PRECIS) package (http://www.metoffice.gov.uk/precis/). This package also includes software to allow the processing and display of the model output data. The PRECIS package is flexible, user-friendly, and computationally inexpensive. It can be easily applied over any region of the globe to provide detailed climate information for regional climate studies and climate change impacts assessment.

3. Experimental design

The experimental design follows Zhang et al. (2009) and is briefly described here.

The WRF Model was set up by using one-way nesting (Fig. 1, left panel). The outer domain at 108-km resolution covers nearly the entire North American continent and much of the eastern Pacific and western Atlantic Oceans. The inner domain at 36-km resolution encompasses the continental United States and part of Canada and Mexico (Fig. 1, left panel). The WRF Model was configured to use 31 vertical levels with the highest resolution (~20–100 m) in the boundary layer.

We chose the highest available resolution (~25 km) for the domain of HadRM (Fig. 1, left panel). The HadRM domain includes a large part of the eastern Pacific Ocean, western United States, and part of Mexico and Canada to better represent the synoptic weather systems that affect the continental United States. There are 19 vertical hybrid levels in HadRM spanning from the surface to 0.5 mb.

For the current work, we used the WRF Model and HadRM simulations that were initialized at 0000 UTC 1 January and 1 December 1969, respectively, and ended at 0000 UTC 1 January 2008. The first 1-yr simulation by WRF and 1-month simulation by HadRM were regarded as model spinup. The initial and lateral conditions for the WRF Model and HadRM were interpolated from the MPI-M ECHAM5 and HadCM3Q0 data, respectively (see Zhang et al. 2009 for details). The external forcing for the twenty-first-century period (2001–08) is from the Special Report on Emissions Scenarios (SRES) A1B emissions scenario (Nakicenovic et al. 2000), which is close to the actual emissions during this period (Raupach et al. 2007). The lateral boundary conditions were updated every 6 hours for both models and both model simulations were output hourly.

4. Methodology

We mostly focus on temperature and precipitation extreme indices recommended by the joint World Meteorological Organization (WMO) Commission for Climatology (CCI)/Climate Variability and Predictability (CLIVAR)/Joint WMO–Intergovernmental Oceanographic Commission (IOC) Commission on Oceanography and Marine Meteorology (JCOMM) Expert Team on Climate Change Detection and Indices. The data quality control and computation of these indices are done using the Fclimdex FORTRAN program (available at http://etccdi.pacificclimate.org). In addition, we also look at several heat wave indices as suggested by Beniston et al. (2007). These indices were chosen to assess aspects of a changing climate that include changes in intensity, frequency, and duration of temperature and precipitation extremes. Definitions of these indices are found in Tables 1 and 2.

Table 1.

Extreme temperature indices used in this study. ID is indicator.

Extreme temperature indices used in this study. ID is indicator.
Extreme temperature indices used in this study. ID is indicator.
Table 2.

Extreme precipitation indices used in this study. PRCP stands for the daily precipitation.

Extreme precipitation indices used in this study. PRCP stands for the daily precipitation.
Extreme precipitation indices used in this study. PRCP stands for the daily precipitation.

The computation of extreme indices requires daily precipitation and daily maximum and minimum temperatures. The daily extreme temperatures are obtained from simulated hourly temperature with terrain adjustment as described in Zhang et al. (2009) and Dulière et al. (2011). The terrain adjustment accounts for differences in altitude between the station that provides the observations and the corresponding model grid box. Note that no terrain adjustment was applied to precipitation as a lapse rate of precipitation over complex terrain would depend on mountain width, buoyancy, moisture field, and winds (Smith and Barstad 2004; Esteban and Chen 2008).

Daily observations from 136 HCN stations in the states of California, Idaho, Nevada, Oregon, and Washington (Fig. 1, right panel) are used to validate the simulated changes in extreme indices. The HCN observations used in this study have been subject to a suite of quality assurance checks such as tests to detect duplicated data, climatological outliers, and various inconsistencies (internal, temporal, and spatial; http://www.ncdc.noaa.gov/oa/climate/ghcn-daily/). Stations where at least 80% of the daily precipitation and temperature measurements are available during the period 1970–2007 and where model grid boxes corresponding to their locations are land cells are used. The locations of the selected HCN stations are indicated in Fig. 1 (right panel).

We examine two different aspects of changes in extreme indices. First we look at any observed and simulated averaged changes over the full domain (namely here, the states of California, Idaho, Nevada, Oregon, and Washington). To that end, trends are computed from areally averaged indices at HCN station locations across the whole domain. The domain average was done in two steps to account for the spatial heterogeneity in the station network. First, the average was done on a gridded domain (1° × 1°), and then the resulting values were averaged over the whole domain. The second aspect of changes in extreme indices studied here consists in the geographical pattern of changes. Here, trends in extreme indices are directly computed at each station location. In all cases, the trend calculation follows Alexander et al. (2006). A nonparametric Kendall's tau-based slope estimator (Sen 1968) was used to compute trends, and the serial correlation in residuals was considered when testing the statistical significance of trends. An iterative procedure was adopted, originally proposed by Zhang et al. (2000) and later refined by Wang and Swail (2001), to compute the magnitude of trends and to test their statistical significance. Note that we also tested the simple linear regression and the Kendall's tau method without considering the serial correlation in residuals. All three methods give similar results.

For precipitation, we also evaluate the simulated climate variability in extreme indices. We look at variability associated with ENSO, which is one important mode of interannual variability in the western United States and is well-represented in the global and regional models used here (Zhang et al. 2012). This evaluation considers the trend in precipitation indices with increasing values of the ONI.

5. Results

a. Extreme temperature indices

We focus here on changes in several extreme temperature indices (see Table 1 for full expansions) at HCN station locations across the states of California, Idaho, Nevada, Oregon, and Washington. We compare extreme indices computed from the observed HCN data with values computed from the HadRM and WRF simulations over the 1970–2007 period.

The time series of extreme temperature indices averaged over all HCN locations across the domain are presented in Fig. 2. The corresponding trends are listed in Table 3. Except for DTR, trends for each temperature extreme index are consistently the same sign for observed and simulated results (Fig. 2; Table 3); although, for most indices, the simulated trends are larger than observed. Even when not statistically significant, these trends correspond with the increase of mean surface temperature due to global warming, which leads to a shift in more frequent and intense warm temperature events and less frequent and intense cold temperature events. Also shown are the results for the two global forcing models, HadCM for HadRM and ECHAM5 for the WRF Model.

Fig. 2.

Observed (black line) and modeled (red: HadRM; brown: HadCM; blue: WRF; and green: ECHAM5) time series of extreme temperature indices from 1970 to 2007 averaged over HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. The straight lines represent the corresponding linear trends in time. The full lines stand for the statistically significant trends at 5% level.

Fig. 2.

Observed (black line) and modeled (red: HadRM; brown: HadCM; blue: WRF; and green: ECHAM5) time series of extreme temperature indices from 1970 to 2007 averaged over HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. The straight lines represent the corresponding linear trends in time. The full lines stand for the statistically significant trends at 5% level.

Table 3.

Observed and modeled temporal trends computed from areally averaged extreme temperature indices over the period 1970–2007 using HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 1.

Observed and modeled temporal trends computed from areally averaged extreme temperature indices over the period 1970–2007 using HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 1.
Observed and modeled temporal trends computed from areally averaged extreme temperature indices over the period 1970–2007 using HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 1.

Several of the observed trends are statistically significant at a 5% level but with various successes by the regional and global models in reproducing these trends. For frost days (FD), both the observations and regional models yield statistically significant decreases of about 0.2–0.5 day yr−1 between 1970 and 2007 (Fig. 2a; Table 3). While the global models also simulate decreasing trends, the simulated trends are not statistically significant and much smaller in the case of ECHAM5. The average percentage of cool nights (TN10p; Fig. 2e) and cool days (TX10p; Fig. 2f) decreases by a statistically significant 0.10%–0.15% yr−1 according to observations. Both the HadRM and HadCM simulations reproduce these results. The WRF Model also reproduces these results but the daytime trend is not statistically significant.

A significant increase of 0.106 yr−1 in the percentage of warm nights (TN90p; Fig. 2g) is observed at HCN stations and is robustly simulated by all models, including the global models. Consistent with the larger observed trend in nighttime (TN90p) than daytime (TX90p) extremes, there is a statistically significant decreasing trend observed in the diurnal temperature range (DTR) of −0.034°C yr−1. A similar trend in DTR is simulated by the WRF Model, but little trend is produced in the HadRM simulation. Note that the HadRM simulation has a large significant positive trend in TX90p inherited from HadCM but contrary to observations, which likely also accounts for the DTR result.

The observations show an increase in heat waves over the region (Figs. 2i–l), although a statistically significant trend (0.062 heat waves yr−1) is found only for the number of heat waves (Fig. 2j; Table 3); the duration, frequency, and intensity increase as well, but these trends are not statistically significant. Both regional models simulate the trend in the number of heat waves with WRF very close to the observed trend (0.053 heat waves yr−1) and HadRM overestimating (0.110 heat waves yr−1). The models tend to overestimate the remaining heat wave indicators including statistically significant trends in heat wave frequency 2 to 3 times the observed. These deficiencies are present in the forcing global models as well and passed to the regional simulations. Trends in the frequency of summer days (SU) and tropical nights (TR) are also too high in the model simulations, by as much as a factor of 5 in the case of HadRM.

At individual HCN stations the observed and modeled trends in several selected temperature extreme indices (FD, SU, TX10p, TX90p, and nHW) as well as in other extreme temperature indices (not shown here) largely agree with each other in sign, although some trends are substantially larger in the simulations, especially for HadRM (Fig. 3). Many of the observed and modeled trends are statistically significant, and these statistically significant trends are spread across the region. Note that trends in observations are only computed for stations with at least 20 years of available extreme indices. Both regional models show trends of consistent sign across the whole domain for each extreme index (second and fourth columns in Fig. 3), while a few opposite trends are found in observed extreme temperature indices (cf. the northern part of Idaho in Fig. 3a). Note also that in the southern part of the domain, frost days are few if not unusual, which results in mostly small and not statistically significant trends. At the same time, the annual number of summer days, warm days, and heat waves show nearly the opposite trend at HCN stations.

Fig. 3.

(left) Observed and modeled temporal trends for (center left) HadRM, (center) HadCM, (center right) WRF, and (right) ECHAM5 in extreme temperature indices from 1970 to 2007 at HCN station locations. The crosses represent trends of magnitude between −0.001 and 0.001. Statistically significant trends at a level of 5% are contoured in black.

Fig. 3.

(left) Observed and modeled temporal trends for (center left) HadRM, (center) HadCM, (center right) WRF, and (right) ECHAM5 in extreme temperature indices from 1970 to 2007 at HCN station locations. The crosses represent trends of magnitude between −0.001 and 0.001. Statistically significant trends at a level of 5% are contoured in black.

Although both regional models simulate the overall pattern of changes in extreme indices well, they fail to simulate these opposite trends. Note that natural variability in the climate will not coincide between the models and observations since it is not related to the external climate forcing. In fact, since the PDO was in a persistent warm phase during the period 1977–97, it is likely that a residual trend from the PDO is present in the observed temperature record, which could explain both the regional variations in trends and the disparities between observations and simulations. Favre and Gershunov (2006) show an intensification of cyclonic activity in the North Pacific coincident with the 1977 regime change in the PDO. This intensification induced an increased frequency of warm nights and decreased frequency of cold nights as compared to warm and cold days. It is likely that this natural climate shift contributes to the differences between observed and simulated temperature extremes.

Thus, results for stations with opposite trends are likely the result of natural climate variability that locally reverses the regional trend. The much higher regional uniformity in the simulated results may reflect weaker natural variability in the forcing models, particularly at the local scale. In fact, in an analysis of temporal standard deviation over all stations, the observed standard deviation exceeds the simulated values for precipitation, maximum temperature, and minimum temperature by a factor of 2–3. Likewise, the regional models simulated higher standard deviations than the global models. Another reason could be that the regional models fail to capture some local weather features as a result of insufficient spatial resolution or deficiencies in land surface representation. Finally, these opposite trends could also result from nonclimatic inhomogeneities in the observed data (i.e., station moves, instrument changes, and changes in land-use or the environment surrounding the station).

The third and last columns in Fig. 3 show the trends in extreme temperatures as represented in the global models used to force the two regional climate model simulations. In general, the trends in the global models are consistent with the regional models but with localized differences in magnitude. This result indicates that large-scale warming of the western United States over the period 1970–2007 has resulted in a concomitant increase in extreme events. There are important localized differences between the global model simulations and the regional simulations, especially for some parameters such as FD and SU. These differences reflect the importance of local land–atmosphere interactions that depend on the local terrain and land cover.

No systematic differences are noted in extreme temperature trends between the U.S. Northwest and Southwest (Fig. 3), as is typically seen in warm and cold ENSO or PDO events (Ropelewski and Halpert 1986; Cayan 1996; Kumar and Hoerling 1998; Gershunov 1998; Favre and Gershunov 2006; Favre and Gershunov 2009; Zhang et al. 2012). This suggests that a common and strong external forcing is at play for the two regions. Between 1970 and 2007, the annual number of frost days and cold days have significantly decreased at most of the HCN stations in observations and across the whole domain in the model simulations (Fig. 3). As suggested previously, these changes in extreme indices clearly reflect the global warming signal due to the anthropogenic forcing.

b. Extreme precipitation indices

In this section, we first examine the observed and simulated magnitude of annual maximum 1- and 5-day precipitation (RX1day_10 and RX5day_10) at a 10-yr return period. This is one way of evaluating the models performance in simulating the observed climatological statistics of extreme precipitation. Figure 4 shows the scatterplots of simulated RX1day_10 and RX5day_10 against observations at each HCN station. Both regional models simulate the observed return periods reasonably well. The correlation coefficients between the modeled and observed return periods are always larger than 0.7 with the linear regression slopes around 1, especially for the WRF Model. There is indication of overestimation by HadRM, which has a linear regression exceeding 1 for both RX1day_10 and RX5day_10. Also shown are the results for the two global models used as boundary conditions for the regional simulations. In both cases, and especially for HadCM, the global models perform relatively poorly in representing the spatial variations in the observed intensity of heavy precipitation. Thus, the regional models are able to generate a geographically improved climatology of heavy precipitation from the forcing models.

Fig. 4.

Scatterplots of (left) HadRM, (left center) HadCM, (right center) WRF, and (right) ECHAM5 annual maximum (top) 1-day (RX1day_10) and (bottom) 5-day (RX5day_10) precipitation with a return period of 10 years against observed ones. Colors from light gray to black represent station density. The two numbers in bottom-right of each scatterplot correspond to the correlation coefficient and linear regression slope, respectively. Units are mm.

Fig. 4.

Scatterplots of (left) HadRM, (left center) HadCM, (right center) WRF, and (right) ECHAM5 annual maximum (top) 1-day (RX1day_10) and (bottom) 5-day (RX5day_10) precipitation with a return period of 10 years against observed ones. Colors from light gray to black represent station density. The two numbers in bottom-right of each scatterplot correspond to the correlation coefficient and linear regression slope, respectively. Units are mm.

We now turn to the trends in several precipitation indices over the 38-yr period: the simple daily intensity index (SDII), the annual total wet day precipitation (PRCPTOT), the number of days with heavy and very heavy precipitation (R10mm and R20mm, respectively), the consecutive number of dry days (CDD), the consecutive number of wet days (CWD), the total precipitation on very wet and extreme wet days (R95p and R99p), and finally the annual maximum 1- and 5-day precipitation (RX1day and RX5day). We compare changes in observed extreme indices with changes in simulated extreme indices over the 1970–2007 time period averaged over all 136 HCN station locations.

Figure 5 shows the time series of extreme precipitation indices averaged over all HCN locations across the domain. Most of the regionally averaged indices from observations suggest a modest but statistically insignificant increase over the 1970–2007 period (Fig. 5; Table 4). The regional simulations also give insignificant trends with the WRF simulations more consistently positive than observed and the HadRM simulations similar to the observations with variable trends among the indices. Madsen and Figdor (2007) showed that an unusual increase in precipitation intensity has occurred over the contiguous United States since 1970 when compared to the 1948–69 period. Heavy, very heavy, and extreme precipitation events all increased but at different rates while in the meantime, the annual number of days with rain or snowfall has decreased. Similar to the regional simulations, insignificant trends in extreme indices are also noted in the driving global models (Fig. 5; Table 4). The trends in extreme indices averaged over two subregions [namely the Pacific Northwest and Pacific Southwest (PSW)] give similar results both in the observations and simulations (not shown).

Fig. 5.

As in Fig. 2, but for extreme precipitation indices.

Fig. 5.

As in Fig. 2, but for extreme precipitation indices.

Table 4.

Observed and modeled temporal trends computed from areally averaged extreme precipitation indices over the period 1970–2007 using HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 2.

Observed and modeled temporal trends computed from areally averaged extreme precipitation indices over the period 1970–2007 using HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 2.
Observed and modeled temporal trends computed from areally averaged extreme precipitation indices over the period 1970–2007 using HCN station locations in the states of California, Idaho, Oregon, Nevada, and Washington. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 2.

The observed and simulated trends in precipitation indices (SDII, PRCPTOT, R10mm, R95p, and RX1day) as well as in other extreme precipitation indices (not shown here) show large spatial heterogeneity across individual HCN stations (Fig. 6). For observations, increasing trends are found in northern Washington and southern California with decreasing trends in Oregon. Madsen and Figdor (2007) and Mass et al. (2011) also showed increasing trends in the frequency of extreme precipitation over northern Washington and southern California and decreasing trends in Oregon during the period of 1948–2006. In the interior of the domain, observed trends are rather small (Fig. 6). The observed trends are not statistically significant at most HCN stations. Note that Kunkel et al. (1999) analyzed the 1931–96 period and found a different pattern of extreme precipitation trends in that the U.S. Southwest, which exhibited a highly statistically significant upward trend in short duration (1–7 days) extreme precipitation events.

Fig. 6.

As in Fig. 3, but for extreme precipitation indices.

Fig. 6.

As in Fig. 3, but for extreme precipitation indices.

Large differences are noted in the spatial patterns of the observed and simulated trends at HCN stations (Fig. 6). For HadRM, increasing trends are found in the Pacific Northwest and decreasing trends are noted in California, except for PRCPTOT and R10mm for which decreasing trends are also found in the Pacific Northwest. For WRF, increasing trends are always noted over the Pacific Northwest, while decreasing trends are identified over California in terms of PRCPTOT and R10mm. The correspondence of R10mm with PRCPTOT suggests that the 10-mm threshold does not capture extreme events that respond differently to large-scale climate variations from total precipitation, as seen for R95p and RX1day. Note that most of the simulated trends in extreme precipitation indices are not statistically significant.

As for the temperature results (Fig. 3), the trends in the global forcing models are also shown for comparison with the regional models in Fig. 6. Looking first at total precipitation (second row in Fig. 6), it is clear that the large-scale changes simulated by the global models are passed on to the regional models. The large-scale changes reflect shifts in the moisture flux and storm patterns that strongly condition the mesoscale simulations. HadCM simulates a modest domainwide reduction in precipitation, which is reproduced in HadRM at nearly all stations. For the extreme precipitation indices SDII and R95p, however, HadCM simulates statistically significant increases across the state of Washington. In the regional simulation, HadRM restricts these increases to the western part of the state along the windward slopes of the Cascades. This connection between large-scale zonal winds, moisture flux, and precipitation is seen in other modeling studies for this region, for example in Leung et al. (2003). ECHAM5, by contrast, simulates increasing precipitation in the Northwest with decreases elsewhere. This large-scale pattern is only partly transferred to the WRF result with considerable local differences, not statistically significant (i.e., the eastern part of the study area and southern California). Similar to HadCM, ECHAM5 simulates statistically significant increases in heavy precipitation (SDII and RX1day; SDII and R95p for HadCM) in the Northwest.

The small and insignificant trends in extreme precipitation indices with a large spatial heterogeneity across the domain suggest that the anthropogenic forcing associated with global warming cannot be the dominant forcing for changes in precipitation across the region. This is in contrast to the extreme temperature indices where the global warming footprint is dominant in their changes (Figs. 2 and 5). Temperature as well as precipitation is affected by natural variability (e.g., ENSO and PDO) while climate change affects temperature in a stronger, more direct, and uniform way than it affects precipitation. In particular, trends and variability in heavy precipitation are closely linked to large-scale circulation patterns that are controlled by ENSO and PDO on interannual to decadal time scales (Gershunov and Cayan 2003). This results in a more detectable climate change signal in temperature (on top of natural variability) than in precipitation. Statistically significant precipitation trends are coincident across observations and the two regional models only for western Washington in SDII, RX1day, and R95p (not significant in the WRF simulation). A climate change–induced increase in heavy precipitation in this particular region is supported by studies that project a poleward shift and intensification of the Pacific storm track (Favre and Gershunov 2009; Salathé 2006; Ulbrich et al. 2008, 2009) due to climate change. Nevertheless, while the present results suggest an emerging effect of climate change, such a conclusion is not statistically supported by the data presented here.

c. ENSO teleconnection patterns

Since the observed trends in extreme precipitation during the period 1970–2007 do not appear to be well-simulated by the regional models, we hypothesize that this is a result of the partial sampling of natural variability, particularly the PDO, which undergoes regime shifts on the same time scale as the 40-yr period analyzed here. Even in the absence of systematic climate change, a 38-yr record would not be sufficient to fully average out natural variability. In this case, however, the residual trends would not reveal a “fingerprint” of anthropogenic climate change.

ENSO is one of the most important modes of natural variability over the western United States affecting precipitation (Cayan and Roads 1984; Ropelewski and Halpert 1986; Redmond and Koch 1991; Wallace et al. 1992; Cayan 1996; Mitchell and Blier 1997; Dettinger et al. 1998; Kumar and Hoerling 1998; Pandey et al. 1999). Since ENSO cycles are not related to external forcing, variability in precipitation related to ENSO will not correspond by year among the models. Since ENSO is high frequency, we do not expect systematic or common trends in observations. Nevertheless, it would be useful to evaluate whether the observed relationship between large-scale climate related to ENSO and local precipitation extremes is well simulated by the regional models. To this end, we shall examine the relationship between precipitation extremes and the ONI. Note that since ONI is quasi periodic, the correlation of an index with increasing ONI does not imply an increase over time.

During the wet season (October to March), the warm phase of ENSO (i.e., El Niño) is correlated with warm–dry weather in the PNW and cool–wet weather in the southwest United States. The cold phase of ENSO (i.e., La Niña) is characterized by cool–wet weather over the PNW and warm–dry weather of the southwest United States. An ENSO influence on precipitation and temperature extremes in the western United States have been reported by Gershunov (1998) and Cayan et al. (1999). The ability of the global and regional models used in this study to represent ENSO and its teleconnections across the western United States has been examined by Zhang et al. (2012), and both the HadCM–HadRM and ECHAM5–WRF simulations perform quite well.

The PDO is more likely than ENSO to produce observed trends over the 1970–2007 period since the 40-yr period averages over several ENSO cycles. We performed a similar analysis for both ENSO and PDO. Results for the PDO were largely consistent with the results for ENSO, but because of the short period of analysis compared to PDO cycles, their statistical significance is weak. Therefore, we focus here on an analysis of ENSO behavior because 1) it is well simulated by the models in contrast to PDO and 2) the effects of PDO and ENSO on temperature and precipitation in western North America are very similar (Mantua et al. 1997).

Following Zhang et al. (2012), ONI was used to identify El Niño and La Niña events in the tropical Pacific from the driving model (ECHAM5 and HadCM3) SST fields. ONI is defined as the running 3-month mean SST anomalies for the Niño-3.4 region (i.e., 5°N–5°S, 120°–170°W). Events are defined as 5 consecutive months at or above the +0.5°C anomaly for warm phase and at or below the −0.5°C anomaly for cold phase. Figure 7 shows the running 3-month mean SST anomalies for the Niño-3.4 region (ONI) based on observations and HadCM3 and ECHAM5 simulations for 1970–2007. Table 5 presents the frequency (defined as the number of occurrence) and intensity (defined as the average of the maximum ONI) of El Niño and La Niña events. The ONI peaks around January in both the observations and simulations. During 1970–2007, 8 El Niño events with an intensity of 1.8°C and 7 La Niña events with an intensity of −1.6°C are indicated in the observations (Table 5; Fig. 7, top); HadCM3 shows 4 El Niño events with an intensity of 1.9°C and 5 La Niña events with an intensity of −1.4°C (Table 5; Fig. 7, middle); and ECHAM5 shows 9 El Niño events with an intensity of 2.2°C and 11 La Niña events with an intensity of −2.0°C (Table 5; Fig. 7, bottom). The ENSO events in ECHAM5 are very regular. Since ECHAM5 and HadCM3 are free-running global climate models, the simulated ENSO events do not match the observed sequence of ENSO events but the distributions are similar (Fig. 7). The ENSO events in ECHAM5 and HadCM3 last for about 1–2 years as in the observations.

Fig. 7.

Running 3-month mean ONI (dark gray) and PDO (light gray) indices during 1970–2007 constructed from (a) observations, (b) HadCM3, and (c) ECHAM5 global climate model simulations. HadCM PDO was not computed because of the lack of data.

Fig. 7.

Running 3-month mean ONI (dark gray) and PDO (light gray) indices during 1970–2007 constructed from (a) observations, (b) HadCM3, and (c) ECHAM5 global climate model simulations. HadCM PDO was not computed because of the lack of data.

Table 5.

Frequency and average intensity (°C) of El Niño and La Niña events for the 1970–2007 period based on ONI. Frequency stands for the number of events during the period and intensity for the average of the maximum ONI for all events.

Frequency and average intensity (°C) of El Niño and La Niña events for the 1970–2007 period based on ONI. Frequency stands for the number of events during the period and intensity for the average of the maximum ONI for all events.
Frequency and average intensity (°C) of El Niño and La Niña events for the 1970–2007 period based on ONI. Frequency stands for the number of events during the period and intensity for the average of the maximum ONI for all events.

As mentioned above, the ENSO influence is most pronounced in the wet season (October–March). Therefore, extreme precipitation indices are computed here only over the wet season. This contrasts with the first part of our study where we focused on annual indices.

Figure 8 shows the observed and modeled extreme precipitation indices averaged over HCN stations plotted against the corresponding ONI value. The extreme indices are computed separately for the Pacific Northwest (left panel) and the Pacific Southwest (defined here as the states of California and Nevada; right panel). For each observed index, opposite slopes with increasing ONI are found between the two regions. Both models capture these opposite trends in extreme indices except for WRF-simulated R10mm, CDD, and CWD. Compared to the time series in extreme precipitation indices (Fig. 5), the regression coefficients are statistically significant for many more parameters.

Fig. 8.

Scatterplots of observed (black) and modeled (red: HadRM and blue: WRF) extreme precipitation indices averaged over HCN station locations against ONI indices from 1970 to 2007 in the states of (left) Idaho, Oregon, and Washington and (right) California and Nevada. The straight lines represent the corresponding linear regression with ONI. The full lines stand for the statistically significant linear regression at 5% level. Extreme indices are computed over the wet seasons only.

Fig. 8.

Scatterplots of observed (black) and modeled (red: HadRM and blue: WRF) extreme precipitation indices averaged over HCN station locations against ONI indices from 1970 to 2007 in the states of (left) Idaho, Oregon, and Washington and (right) California and Nevada. The straight lines represent the corresponding linear regression with ONI. The full lines stand for the statistically significant linear regression at 5% level. Extreme indices are computed over the wet seasons only.

For the Pacific Northwest, the observed extreme precipitation indices decrease with ONI except for CDD (Fig. 8; Table 6). This suggests a decrease in intensity and frequency of precipitation extremes during El Niño years compared to La Niña years. Both models simulate this observed relationship between extremes and ONI, except for WRF-simulated R10mm, CDD, and CWD. The HadRM results are closer to the observed compared to the WRF Model. The WRF Model is able to capture changes in very intense precipitation indices such as R95p, R99p, RX1day, and RX5day. However, WRF displays wet biases as reflected by overestimation of CWD, PRCPTOT, and R10mm for El Niño years. This probably also explains the WRF simulated decreasing trend in CDD as opposed to the observations and HadRM simulation. This result is consistent with the findings in Zhang et al. (2012), who showed a weak ENSO response in Pacific Northwest precipitation in ECHAM5 as compared to reanalysis and HadCM. This deficiency in the forcing model is inherited by the associated WRF simulation.

Table 6.

Regression slopes of observed and modeled extreme precipitation indices with ONI computed from areally averaged extreme precipitation indices over the period 1970–2007 using HCN station locations. The PNW region includes the states of Washington, Oregon, and Idaho and the PSW region includes the states of California and Nevada. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 2.

Regression slopes of observed and modeled extreme precipitation indices with ONI computed from areally averaged extreme precipitation indices over the period 1970–2007 using HCN station locations. The PNW region includes the states of Washington, Oregon, and Idaho and the PSW region includes the states of California and Nevada. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 2.
Regression slopes of observed and modeled extreme precipitation indices with ONI computed from areally averaged extreme precipitation indices over the period 1970–2007 using HCN station locations. The PNW region includes the states of Washington, Oregon, and Idaho and the PSW region includes the states of California and Nevada. Boldface indicates statistically significance at 5% level. Index definitions are given in Table 2.

For the Pacific Southwest, each index increases with ONI in both observations and model simulations except for CDD for which the observed and simulated trends are decreasing (Fig. 8; Table 6). Among the two regional models, WRF simulates the magnitude of the regression coefficients with ONI over the Pacific Southwest better than HadRM. This is in contrast to the Pacific Northwest where HadRM performs better than the WRF Model in resolving the observed precipitation changes with ONI. Again, this result can be traced to the performance of the forcing global models, HadCM and ECHAM5.

Figure 9 shows the spatial distributions of the regression coefficients of several extreme precipitation indices (namely SDII, PRCPTOT, R10mm, R95p, and RX1day) with ONI at individual HCN stations. Consistent with the spatially averaged ONI regression coefficients discussed above, decreasing and increasing coefficients are noted at most of the HCN stations over the Pacific Northwest and Southwest, respectively. This is true for both observations and model simulations.

Fig. 9.

(left) Observed and modeled regression of extreme precipitation indices for (center left) HadRM, (center) HadCM, (center right) WRF, and (right) ECHAM5 with ONI over the 1970–2007 period at each HCN station. The crosses represent ONI regression coefficients between −0.001 and 0.001. Statistically significant coefficients at a level of 5% are contoured in black. Extreme indices are computed over the wet seasons only.

Fig. 9.

(left) Observed and modeled regression of extreme precipitation indices for (center left) HadRM, (center) HadCM, (center right) WRF, and (right) ECHAM5 with ONI over the 1970–2007 period at each HCN station. The crosses represent ONI regression coefficients between −0.001 and 0.001. Statistically significant coefficients at a level of 5% are contoured in black. Extreme indices are computed over the wet seasons only.

In the interior of the domain and on the lee side of mountain ranges (namely the Cascades, Rockies, and Sierra Nevada), the regression coefficients with ONI are rather small both in observations and model simulations. In contrast to the temporal trends in Fig. 6, the regression coefficients with ONI are statistically significant at many HCN stations (Fig. 9).

Thus, while the simulated temporal trends in extreme precipitation do not correspond well with observations, this does not appear to reflect on model performance. Indeed, based on the results for ENSO, the models do a good job of simulating natural variability and the response of extreme precipitation to natural climate variability. This result suggests that the weak results for simulating temporal trends are due to the lack of robust signal of anthropogenic climate forcing on extreme precipitation during the period 1970–2007. Any trend that may occur because of anthropogenic greenhouse gas forcing, which could be captured by the regional and global climate models, is not clearly discernible from natural climate variability and/or could be masked by decadal variability such as the PDO (Favre and Gershunov 2009).

6. Discussion and conclusions

In this paper, we have examined the performance of two regional climate simulations in simulating the observed trends and variability in extreme temperature and precipitation for the recent past (1970–2007) over the western United States. The regional climate simulations are forced by global climate model simulations from the HadCM and ECHAM5 models; thus, the comparison evaluates both the global and regional model results. These simulations include only external climate forcings resulting from volcanic aerosol, solar forcing, and anthropogenic greenhouse gases. Thus, anthropogenic greenhouse gases provide the only systematic trends in the forcing for the simulations.

The observed temperature extremes show substantial and statistically significant trends across the region, with consistent results among individual stations. Observed indices of cold extremes, such as the annual number of frost days or percentage of cool nights, decreased with time over the 1970–2007 period. Indices of hot extremes, such as the annual number of summer days, percentage of warm days, or number of heat waves, increased over the same period. Both regional models simulate these temporal trends consistently with the observations. Several measures of extreme temperature, however, show much stronger trends in the simulations than have been observed. Since the observed trends in temperature extremes are so strongly simulated in the regional climate models, they are very likely to reflect the global warming signal as a result of the anthropogenic forcing. The overestimate of these trends is likely the result of deficiencies in both the regional and global climate models since the trends in each are consistent. One possible source of this discrepancy is aerosols, including their indirect effects, which are not well represented in the models used here (Jiang et al. 2012).

For precipitation, the observed trends are statistically significant at only a few stations, with both positive and negative trends depending on the station. No regional average trend is found, which is consistent with other previous studies, which typically show a northward shift in precipitation with climate change. Specifically, the observations indicate increases in extreme precipitation at stations in western Washington and coastal northern California. Decreases are found in some indices for Oregon stations. Insignificant trends are found for stations through much of the interior. Some of these spatial features, such as increases in western Washington, are weakly reproduced in the regional models but not consistently for all precipitation indices. There is no clear correspondence between the observed and simulated results for extreme precipitation.

To test whether the weak results for precipitation are related to modeling deficiencies, we examine the variability of precipitation extremes with natural variability associated with ENSO. Here, a strong correspondence is found between observed precipitation indices and ONI, with regionally coherent patterns found for the Northwest and Southwest. Both regional climate simulations reproduce the observed ONI relationship, indicating that the models can represent the climatic links with extreme precipitation.

Thus, the lack of correspondence between observed and simulated trends for extreme precipitation likely results from the dominance of natural variability over anthropogenic trends in heavy precipitation during the period 1970–2007. The simulated natural variability in the free-running climate simulations is not correlated in time with the observed variability, so residual trends due to variability will not produce consistent geographical results. Over time, as the anthropogenic influence amplifies, temporal trends in extreme precipitation may emerge. In fact, the initial analysis of future climate change simulations with the models used in this paper show statistically significant increases in some indices, and this issue is the subject of ongoing research.

Acknowledgments

This work is funded by the State of Washington as part of the Washington Climate Change Impacts Assessment, by an Environmental Protection Agency STAR Grant, by the National Science Foundation (ATM0709856), and by a Microsoft Corporation gift to the Climate Impacts Group. We thank the PRECIS team from the U.K. Met Office and especially Richard Jones, David Hein, David Hassell, and Simon Wilson for supplying the PRECIS package and helping us to use it. Part of the WRF simulations was performed at the National Center for Atmospheric Research (NCAR) Computational and Information System Laboratory (CISL). This publication is also partially funded by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement NA17RJ1232.

REFERENCES

REFERENCES
Alexander
,
L. V.
, and
J. M.
Arblaster
,
2009
:
Assessing trends in observed and modelled climate extremes over Australia in relation to future projections
.
Int. J. Climatol.
,
29
,
417
435
.
Alexander
,
L. V.
, and
Coauthors
,
2006
:
Global observed changes in daily climate extremes of temperature and precipitation
.
J. Geophys. Res.
,
111
, D05109,
doi:10.1029/2005JD006290
.
Beniston
,
M.
, and
Coauthors
,
2007
:
Future extreme events in European climate: An exploration of regional climate model projections
.
Climatic Change
,
81
,
71
95
.
Brown
,
S. J.
,
J.
Caesar
, and
C. A. T.
Ferro
,
2008
:
Global changes in extreme daily temperature since 1950
.
J. Geophys. Res.
,
113
, D05115, doi:10.1029/2006JD008091.
Buonomo
,
E.
,
R.
Jones
,
C.
Huntingford
, and
J.
Hannaford
,
2007
:
On the robustness of changes in extreme precipitation over Europe from two high resolution climate change simulations
.
Quart. J. Roy. Meteor. Soc.
,
133
,
65
81
.
Cayan
,
D. R.
,
1996
:
Interannual climate variability and snowpack in the western United States
.
J. Climate
,
9
,
928
948
.
Cayan
,
D. R.
, and
J. O.
Roads
,
1984
:
Local relationships between United States West Coast precipitation and monthly mean circulation parameters
.
Mon. Wea. Rev.
,
112
,
1276
1282
.
Cayan
,
D. R.
,
K. T.
Redmond
, and
L. G.
Riddle
,
1999
:
ENSO and hydrologic extremes in the western United States
.
J. Climate
,
12
,
2881
2893
.
Chen
,
F.
, and
J.
Dudhia
,
2001
:
Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity
.
Mon. Wea. Rev.
,
129
,
569
585
.
Christensen
,
J. H.
,
T. R.
Carter
,
M.
Rummukainen
, and
G.
Amanatidis
,
2007
:
Evaluating the performance and utility of regional climate models: The PRUDENCE project
.
Climatic Change
,
81
,
1
6
.
Collins
,
W. D.
, and
Coauthors
,
2004
:
Description of the NCAR Community Atmospheric Model (CAM 3.0). NCAR Tech. Rep. NCAR/TN-464+STR, 226 pp
.
Cox
,
P. M.
,
R. A.
Betts
,
C. B.
Bunton
,
R. L. H.
Essery
,
P. R.
Rowntree
, and
J.
Smith
,
1999
:
The impact of new land surface physics on the GCM simulation of climate and climate sensitivity
.
Climate Dyn.
,
15
,
183
203
.
DeGaetano
,
A. T.
,
2009
:
Time-dependent changes in extreme-precipitation return-period amounts in the continental United States
.
J. Appl. Meteor. Climatol.
,
48
,
2086
2099
.
Dettinger
,
M. D.
,
D. R.
Cayan
,
H. F.
Diaz
, and
D. M.
Meko
,
1998
:
North–south precipitation patterns in western North America on interannual-to-decadal timescales
.
J. Climate
,
11
,
3095
3111
.
Duffy
,
P. B.
, and
Coauthors
,
2006
:
Simulations of present and future climates in the western United States with four nested regional climate models
.
J. Climate
,
19
,
873
895
.
Dulière
,
V.
,
Y.
Zhang
, and
E. P.
Salathé
,
2011
:
Extreme precipitations and temperatures over the U.S. Pacific Northwest: A comparison between observations, reanalysis, and regional models
.
J. Climate
,
24
,
1950
1964
.
Esteban
,
M. A.
, and
Y.-L.
Chen
,
2008
:
The impact of trade wind strength on precipitation over the windward side of the island of Hawaii
.
Mon. Wea. Rev.
,
136
,
913
928
.
Favre
,
A.
, and
A.
Gershunov
,
2006
:
Extra-tropical cyclonic/anticyclonic activity in north-eastern Pacific and air temperature extremes in western North America
.
Climate Dyn.
,
26
,
617
629
.
Favre
,
A.
, and
A.
Gershunov
,
2009
:
North Pacific cyclonic and anticyclonic transients in a global warming context: Possible consequences for western North American daily precipitation and temperature extremes
.
Climate Dyn.
,
32
,
969
987
.
Field
,
C. B.
, and
Coauthors
, Eds.,
2012
:
Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp
.
Fowler
,
H. J.
,
M.
Ekström
,
C. G.
Kilsby
, and
P. D.
Jones
,
2005
:
New estimates of future changes in extreme rainfall across the UK using regional climate model integrations. 1. Assessment of control climate
.
J. Hydrol.
,
300
,
212
233
.
Gershunov
,
A.
,
1998
:
ENSO influence on intraseasonal extreme rainfall and temperature frequencies in the contiguous United States: Implications for long-range predictability
.
J. Climate
,
11
,
3192
3203
.
Gershunov
,
A.
, and
T.
Barnett
,
1998
:
Inter-decadal modulation of ENSO teleconnections
.
Bull. Amer. Meteor. Soc.
,
79
,
2715
2725
.
Gershunov
,
A.
, and
D.
Cayan
,
2003
:
Heavy daily precipitation frequency over the contiguous United States: Sources of climatic variability and seasonal predictability
.
J. Climate
,
16
,
2752
2765
.
Gordon
,
C.
,
C.
Cooper
,
C. A.
Senior
,
H. T.
Banks
,
J. M.
Gregory
,
T. C.
Johns
,
J. F. B.
Mitchell
, and
R. A.
Wood
,
2000
:
The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments
.
Climate Dyn.
,
16
,
147
168
.
Gregory
,
D.
, and
P. R.
Rowntree
,
1990
:
A mass-flux convection scheme with representation of cloud ensemble characteristics and stability dependent closure
.
Mon. Wea. Rev.
,
118
,
1483
1506
.
Gregory
,
D.
, and
S.
Allen
,
1991
:
The effect of convective downdraughts upon NWP and climate simulations. Preprints, Ninth Conf. on Numerical Weather Prediction, Denver, CO, Amer. Meteor. Soc., 122–123
.
Gregory
,
D.
,
R.
Kershaw
, and
P. M.
Inness
,
1997
:
Parametrization of momentum transport by convection. II: Tests in single column and general circulation models
.
Quart. J. Roy. Meteor. Soc.
,
123
,
1153
1183
.
Gutowski
,
W. J.
, and
Coauthors
,
2008
:
Causes of observed changes in extremes and projections of future changes. Weather and Climate Extremes in a Changing Climate. Regions of Focus: North America, Hawaii, Caribbean, and U.S. Pacific Islands, T. R. Karl et al., Eds., U.S. Climate Change Science Program, 81–116
.
Hong
,
S. Y.
, and
H. L.
Pan
,
1996
:
Nonlocal boundary layer vertical diffusion in a medium-range forecast model
.
Mon. Wea. Rev.
,
124
,
2322
2339
.
Hong
,
S. Y.
,
J.
Dudhia
, and
S. H.
Chen
,
2004
:
A revised approach to ice microphysical processes for the bulk parameterization of clouds and precipitation
.
Mon. Wea. Rev.
,
132
,
103
120
.
Hong
,
S. Y.
,
Y.
Noh
, and
J.
Dudhia
,
2006
:
A new vertical diffusion package with an explicit treatment of entrainment processes
.
Mon. Wea. Rev.
,
134
,
2318
2341
.
Huntingford
,
C.
,
R. G.
Jones
,
C.
Prudhomme
,
R.
Lamb
,
J. H. C.
Gash
, and
D. A.
Jones
,
2003
:
Regional climate-model predictions of extreme rainfall for a changing climate
.
Quart. J. Roy. Meteor. Soc.
,
129
,
1607
1621
.
Jiang
,
J. H.
, and
Coauthors
,
2012
:
Evaluation of cloud and water vapor simulations in CMIP5 climate models using NASA “A-Train” satellite observations
.
J. Geophys. Res.
,
117
, D14105,
doi:10.1029/2011JD017237
.
Johns
,
T. C.
, and
Coauthors
,
2003
:
Anthropogenic climate change for 1860 to 2100 simulated with the HadCM3 model under updated emissions scenarios
.
Climate Dyn.
,
20
,
583
612
.
Jones
,
R. G.
,
M.
Noguer
,
D. C.
Hassel
,
D.
Hudson
,
S. S.
Wilson
,
G. L.
Jenkins
, and
J. F. B.
Mitchell
,
2004
:
Generating high resolution climate change scenarios using PRECIS. Met Office Hadley Centre/National Communications Support Unit Handbook, 40 pp
.
Kain
,
J. S.
, and
J. M.
Fritsch
,
1993
:
Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 24, Amer. Meteor. Soc., 165–170
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Kanamitsu
,
M.
,
W.
Ebisuzaki
,
J.
Woollen
,
S.-K.
Yang
,
J. J.
Hnilo
,
M.
Fiorino
, and
G. L.
Potter
,
2002
:
NCEP–DOE AMIP-II Reanalysis (R-2)
.
Bull. Amer. Meteor. Soc.
,
83
,
1631
1643
.
Kumar
,
A.
, and
M.
Hoerling
,
1998
:
Annual cycle of Pacific–North American seasonal predictability associated with different phases of ENSO
.
J. Climate
,
11
,
3295
3308
.
Kunkel
,
K. E.
,
K.
Andsager
, and
D. R.
Easterling
,
1999
:
Long-term trends in extreme precipitation events over the conterminous United States and Canada
.
J. Climate
,
12
,
2515
2527
.
Leung
,
L. R.
,
Y.
Qian
, and
X.
Bian
,
2003
:
Hydroclimate of the western United States based on observations and regional climate simulation of 1981–2000. Part I: Seasonal statistics
.
J. Climate
,
16
,
1892
1911
.
Leung
,
L. R.
,
Y.
Qian
,
X.
Bian
,
W. M.
Washington
,
J.
Han
, and
J. P.
Roads
,
2004
:
Mid-century ensemble regional climate change scenarios for the western United States
.
Climatic Change
,
62
,
75
113
.
Madsen
,
T.
, and
E.
Figdor
,
2007
:
When it rains, it pours: Global warming and the rising frequency of extreme precipitation in the United States. Environment California Research and Policy Center Rep., 48 pp
.
Mantua
,
N.
,
S.
Hare
,
Y.
Zhang
,
J.
Wallace
, and
R.
Francis
,
1997
:
A Pacific decadal climate oscillation with impacts on salmon
.
Bull. Amer. Meteor. Soc.
,
78
,
1069
1079
.
Mass
,
C. F.
,
D.
Ovens
,
K.
Westrick
, and
B. A.
Colle
,
2002
:
Does increasing horizontal resolution produce more skillful forecasts?
Bull. Amer. Meteor. Soc.
,
83
,
407
430
.
Mass
,
C. F.
,
A.
Skalenakis
, and
M.
Warner
,
2011
:
Extreme precipitation over the west coast of North America: Is there a trend?
J. Hydrometeor.
,
12
,
310
318
.
Mitchell
,
T. P.
, and
W.
Blier
,
1997
:
The variability of wintertime precipitation in the region of California
.
J. Climate
,
10
,
2261
2276
.
Nakicenovic
,
N.
, and
Coauthors
,
2000
:
Special Report on Emissions Scenarios. Cambridge University Press, 570 pp
.
Pandey
,
G. R.
,
D. R.
Cayan
, and
K. P.
Georgakakos
,
1999
:
Precipitation structure in the Sierra Nevada of California during winter
.
J. Geophys. Res.
,
104
,
12 019
12 030
.
Peterson
,
T. C.
,
X.
Zhang
,
M.
Brunet-India
, and
J. L.
Vazquez-Aguirre
,
2008
:
Changes in North American extremes derived from daily weather data
.
J. Geophys. Res.
,
113
, DO7113, doi:10.1029/2007JD009453.
Raupach
,
M. R.
,
G.
Marland
,
P.
Ciais
,
C.
Le Quéré
,
J. G.
Canadell
,
G.
Klepper
, and
C. B.
Field
,
2007
:
Global and regional drivers of accelerating CO2 emissions
.
Proc. Natl. Acad. Sci. USA
,
104
,
10 288
10 293
.
Redmond
,
K. T.
, and
R. W.
Koch
,
1991
:
Surface climate and streamflow variability in the western United States and their relationship to large scale circulation indices
.
Water Resour. Res.
,
27
,
2381
2399
.
Robinson
,
P. J.
,
2001
:
On the definition of a heat wave
.
J. Appl. Meteor.
,
40
,
762
775
.
Ropelewski
,
C. F.
, and
M. S.
Halpert
,
1986
:
North American precipitation and temperature patterns associated with the ENSO
.
Mon. Wea. Rev.
,
114
,
2352
2362
.
Salathé
,
E. P.
,
2006
:
Influences of a shift in North Pacific storm tracks on western North American precipitation under global warming
.
Geophys. Res. Lett.
,
33
, L19820, doi:10.1029/2006GL026882.
Salathé
,
E. P.
,
R. L.
Leung
,
Y.
Qian
, and
Y.
Zhang
,
2010
:
Regional climate model projections for the state of Washington
.
Climatic Change
,
102
,
51
75
.
Sen
,
P. K.
,
1968
:
Estimates of the regression coefficient based on Kendall's tau
.
J. Amer. Stat. Assoc.
,
63
,
1379
1389
.
Skamarock
,
W. C.
,
J. B.
Klemp
,
J.
Dudhia
,
D. O.
Gill
,
D. M.
Barker
,
W.
Wang
, and
J. G.
Powers
,
2005
:
A description of the Advanced Research WRF version 2. NCAR Tech. Note NCAR/TN-468+STR, 88 pp
.
Smith
,
R. B.
,
1990
:
A scheme for predicting layer clouds and their water content in a general circulation model
.
Quart. J. Roy. Meteor. Soc.
,
116
,
435
460
.
Smith
,
R. B.
, and
I.
Barstad
,
2004
:
A linear theory of orographic precipitation
.
J. Atmos. Sci.
,
61
,
1377
1391
.
Tebaldi
,
C.
,
D.
Adams-Smith
, and
N.
Heller
,
2012
:
The heat is on: U.S. temperature trends. Climate Central Rep., 22 pp
.
Thomas
,
R. K.
,
G. A.
Meehl
,
C. D.
Miller
,
S. J.
Hassol
,
A. W.
Waple
, and
W. L.
Murray
, Eds.,
2008
:
Weather and Climate Extremes in a Changing Climate. Regions of Focus: North America, Hawaii, Caribbean, and U.S. Pacific Islands. U.S. Climate Change Science Program, 164 pp
.
Ulbrich
,
U.
,
J. G.
Pinto
,
H.
Kupfer
,
G. C.
Leckebusch
,
T.
Spangehl
, and
M.
Reyers
,
2008
:
Changing Northern Hemisphere storm tracks in an ensemble of IPCC climate change simulations
.
J. Climate
,
21
,
1669
1679
.
Ulbrich
,
U.
,
G. C.
Leckebusch
, and
J. G.
Pinto
,
2009
:
Extra-tropical cyclones in the present and future climate: A review
.
Theor. Appl. Climatol.
,
96
,
117
131
.
Wallace
,
J. M.
,
C.
Smith
, and
C. S.
Bretherton
,
1992
:
Singular value decompositions of winter time sea surface temperature and 500-mb height anomalies
.
J. Climate
,
5
,
561
576
.
Wang
,
X. L.
, and
V. R.
Swail
,
2001
:
Changes of extreme wave heights in Northern Hemisphere oceans and related atmospheric circulation regimes
.
J. Climate
,
14
,
2204
2220
.
Zhang
,
X.
,
L. A.
Vincent
,
W. D.
Hogg
, and
A.
Niitsoo
,
2000
:
Temperature and precipitation trends in Canada during the 20th century
.
Atmos.–Ocean
,
38
,
395
429
.
Zhang
,
Y.
,
V.
Duliere
,
P. W.
Mote
, and
E. P.
Salathé
,
2009
:
Evaluation of WRF and HadRM mesoscale climate simulations over the U.S. Pacific Northwest
.
J. Climate
,
22
,
5511
5526
.
Zhang
,
Y.
,
Y.
Qian
,
V.
Dulière
,
E.
Salathé
, and
L.
Leung
,
2012
:
ENSO anomalies over the western United States: Present and future patterns in regional climate simulations
.
Climatic Change
,
110
,
315
346
.

Footnotes

*

Joint Institute for the Study of the Atmosphere and Ocean Contribution Number 1769.

+

The National Center for Atmospheric Research is sponsored by the National Science Foundation.