1. Introduction
The forecast skill at the Subseasonal to Seasonal Prediction project (S2S) time range (from 2 weeks to 2 months) is relatively low, due to model errors as well as outstanding scientific challenges regarding mechanisms of S2S variability and sources of predictability (Robertson and Vitart 2018; Mariotti et al. 2020). The usage of coupled models for S2S prediction is relatively new (Stan et al. 2023) and the impact of biases in surface boundary conditions on the S2S forecast skill is not well understood. The surface boundary conditions (e.g., sea surface temperature, land surface) represent an important source of predictability on the S2S time range (Guo et al. 2011; Subramanian et al. 2019). Therefore, systematic sea surface temperature (SST) bias simulated in ocean–atmosphere coupled models can greatly impact the S2S forecast skill. On one hand, SST can have a local impact on convection and precipitation. For example, ocean currents can lead to changes in air temperature and moisture content over adjacent land, which may further impact convection and precipitation. On the other hand, SST can impact weather and climate remotely through atmospheric teleconnections (Johnson et al. 2014; Stan et al. 2017; Krishnamurthy et al. 2021). Systematic errors of SST predicted by coupled models can be caused by many reasons including but not limited to, horizontal resolution of the ocean model (Minobe et al. 2008; Siqueira and Kirtman 2016; Balaguru et al. 2021), ocean–atmosphere feedbacks (Li and Xie 2014), representation of low-level clouds (Stan et al. 2010; Hu et al. 2011), and shortwave fluxes (Zuidema et al. 2016). Uncertainties in the SST initial states (Stan and Kirtman 2008) can introduce SST drifts that manifest as forecast biases at both seasonal and subseasonal time scales.
The global coupled model Unified Forecast System (UFS) is developed to replace the coupled model in the NCEP S2S forecast system, Climate Forecast System (CFS) version 2 (Saha et al. 2014), which was developed as a seasonal forecast system. Stan et al. (2023) evaluated the impact of SST biases in the tropical west Pacific (WP) on the biases over the contiguous United States (CONUS) using the UFS coupled model Prototype 5 (P5). In P5, tropical warm SST biases are found over the central and western Pacific, the Indian Ocean, west coast of Africa, and eastern Atlantic. The SST bias patterns show seasonal dependence. During boreal winter, a wet precipitation bias dominates the CONUS, except for a narrow strip along the Gulf of Mexico. During boreal summer, precipitation bias patterns change location, shape, and sign with the forecast lead time. They found that the western Pacific SST biases can be linearly related to biases in the surface temperature, precipitation over the CONUS, midtroposphere large-scale circulation and storm tracks activity during boreal summer. However, they did not examine possible physical mechanisms behind the statistical relationships, and the study only consists of analysis on the WP. Because necessary features and updating components are incrementally added to the coupled UFS, it is also essential to analyze and compare every updated version (prototype) to better understand implications of the updated model features. Using the newer version of the UFS model Prototype 6 (P6), the first objective of this study is to expand on Stan et al. (2023) in order to understand the physical mechanisms that connect the tropical SST biases to biases in the midlatitudes. The second objective is to investigate the impact of the tropical North Atlantic (TNA) SST bias on the S2S precipitation forecast skill over the CONUS along with possible physical mechanisms. In addition, seasonal differences in the impacts and mechanisms will also be examined.
There are many ways in which tropical variability and its mean state can impact the Northern Hemisphere weather from medium range (week 1–2) to S2S time scales. Intraseasonal (20–100 days) variability over the Maritime Continent and WP can impact the CONUS precipitation remotely through tropical–extratropical interactions (Ferranti et al. 1990; Yang et al. 2001; Stan et al. 2017; Lin et al. 2019; Tseng et al. 2019; Wang et al. 2020). Using the UFS coupled model Prototype 2, Krishnamurthy et al. (2021) identified ENSO, an intraseasonal oscillation (ISO) mode and a warming trend as the leading sources of S2S predictability of low-level wind, precipitation, and 2-m temperature over the CONUS. Using the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis and Climate Prediction Center (CPC) precipitation, Arcodia et al. (2020) found that MJO can impact CONUS precipitation through perturbations in the upper-tropospheric flow and low-level moisture, while ENSO can further modulate the background environment of the MJO.
The North Atlantic subtropical high (NASH) is a semipermanent high pressure system over the North Atlantic in the lower troposphere. NASH has been identified as one of the most important circulation factors affecting CONUS precipitation (Katz et al. 2003; Hu et al. 2011; Li et al. 2012). Variations in the position and intensity of NASH can modulate the hydrological cycle between the tropics and the Southeast United States, especially during summertime (Diem 2006; Gamble et al. 2008; Li et al. 2012; Zhu and Stan 2015; Nieto Ferreira and Rickenbach 2020; Zorzetto and Li 2021). Since the TNA SST has been reported to influence the NASH’s intensity (Hasanean 2004; Johnson et al. 2020), we explore the impact of the TNA SST bias on the CONUS precipitation bias via NASH.
Section 2 describes data sources and the UFS model. In section 3, biases in the SST and CONUS precipitation are shown. Multivariate linear regression of the precipitation bias against the SST bias and the mechanisms of the remote impact of the SST biases are further analyzed in section 4. Finally, section 5 gives the summary and conclusions.
2. Model, data, and methods
a. Model
The set of deterministic reforecasts analyzed in this study has been created using the UFS coupled global model, P6. In this prototype, the atmospheric model uses the FV3 dynamical core on the cubed-sphere grid (Putman and Lin 2007; Harris and Lin 2013) and the Common Community Physics Package (CCPP, https://github.com/NCAR/ccpp-physics) for physics parameterizations. The atmospheric physics package is updated from NOAA’s Global Forecast System (GFS) v15.2 used in P1–P5 to GFSv16. The GFSv15.2 physics package uses the Rapid Radiative Transfer Method for General Circulation Models (RRTMG) scheme for shortwave or longwave radiation (Mlawer et al. 1997; Iacono et al. 2000; Clough et al. 2005), the hybrid eddy-diffusivity mass-flux (EDMF) scheme for the planetary boundary layer (PBL; National Centers for Environmental Prediction 2019), the Noah land surface model (LSM) scheme for the land surface option (Chen et al. 1997), the simplified Arakawa–Schubert (SAS) deep convection for cumulus parameterization (Arakawa and Schubert 1974; Grell 1993), and a Geophysical Fluid Dynamics Laboratory (GFDL) microphysics scheme (Chen and Lin 2013). Changes in GFSv16 include a scale-aware EDMF scheme (Strobach 2021, 2022), a gravity wave drag parameterization scheme that includes subgrid scale nonstationary waves, updates in the calculation of the solar radiation absorption by water clouds and cloud overlap assumptions in the RRTMG, updates of the microphysics scheme to compute the ice-cloud effective radius (Zhou et al. 2022), and upgrades in Noah LSM in the calculation of heat flux over snow covered surface as well as incorporation of vegetation impact on surface energy budget over urban areas. In the preceding prototypes, each atmospheric grid consists of either 100% land or 100% water. P6 is the first prototype using fractional grids, which allow a grid cell along the coastlines and lake shores to be partially covered by land, open water, and sea ice. Horizontal resolution of the atmospheric model in all prototypes (P1–P6) is the same at ∼0.25° (C384). P6 is the first prototype with increased atmospheric model vertical resolution (64 levels in previous prototypes and 127 in P6) and higher top (54 km in previous prototypes and 80 km in P6). The ocean component is the GFDL Modular Ocean Model 6 (MOM6) (Adcroft et al. 2019), the sea ice component is version 6 of the Los Alamos sea ice model, referred to as the Community Ice Code (CICE6), and the sea waves component is the WAVEWATCH III model (WW3DG, 2019). The ocean and sea ice models use a tripolar grid with a horizontal resolution of 0.25°. A complete technical description of UFS prototypes is given in Stefanova et al. (2022).
The reforecasts are initialized on the first and fifteenth of each month between April 2011 and March 2018 (168 reforecasts for the entire period) and span 35 days. The atmospheric initial conditions were interpolated from the Climate Forecast System Version 2 (CFSv2; Saha et al. 2014) real time data assimilation, while the CPC Hybrid Global Ocean Data Assimilation System provided the initial conditions for the ocean model. The sea ice model was initialized by the CPC ice analysis and the initial conditions for the wave model were produced using forcing generated with the CFSv2. The initial conditions for ocean, sea ice, and waves components are the same as those used for the reforecasts produced by P5 and analyzed by Krishnamurthy and Stan (2022) and Stan et al. (2023). The initial conditions for atmosphere and land have been recreated to consider the fractional grid decomposition.
b. Data
The reforecasts will be compared to observations and reanalysis data. The daily mean observation datasets include the Optimally Interpolated SST version 2 (OISST2) dataset developed by NOAA (Reynolds et al. 2007) on a 0.25° × 0.25° grid, CPC unified (CPCU) gauge-based analysis of daily precipitation (Xie et al. 2010) on a 0.5° × 0.5° grid, and geopotential, specific humidity, and wind fields from the European Centre for Medium-Range Weather Forecasts interim reanalysis (ERAI; Dee et al. 2011) on the T255 horizontal grid (∼0.7° resolution).
c. Methods
The reforecast data were saved every 6 h. The 6-hourly data are averaged into daily means. For the calculation of biases, the model output is regridded to CPCU and ERAI grids. The daily biases are calculated by subtracting the daily mean in observation/reanalysis from the UFS daily values. Similarly, the weekly means and biases are computed by averaging the daily means and biases, respectively, over the week under consideration. The analysis is conducted for two seasons: boreal summer [June–September (JJAS)] and boreal winter [December–March (DJFM)].
The potential linkage between the CONUS precipitation biases and tropical SST biases is investigated by analyzing their lag-zero multivariate linear regression maps. The statistical significance of regression is estimated using the effective degrees of freedom developed by Bretherton et al. (1999). The effective degrees of freedom at each grid point is defined as
3. Forecast biases
Before investigating the impact of SST bias on precipitation, this section first analyzes the UFS biases in SST and CONUS precipitation. Figure 1 shows the weekly SST biases between the UFS model and OISST2 in boreal summer (JJAS) and winter (DJFM). Red colors represent warmer conditions in the UFS than observations, signifying a warm bias in the model. During both seasons, the strongest SST biases are found over the topics, including the tropical Pacific (the WP, the tropical central Pacific, and the tropical east Pacific), the tropical east Atlantic, and the tropical Indian Ocean. The North Pacific and the west coast of North America also show strong warm SST biases in JJAS. Interestingly, the UFS model captures most of the observed SST features around the Gulf Stream. This achievement may be related to the relatively high horizontal resolution (0.25°). For example, the horizontal resolution of the ocean component of the GFDL-Forecast-oriented Low Ocean Resolution (FLOR) model used in Johnson et al. (2020) is 1°, which may not be high enough to represent an accurate SST along the Gulf Stream. Tsartsali et al. (2022) has compared six global climate models (GCMs) with respect to their performance along the Gulf Stream. They found that increasing ocean resolution results in enhanced atmosphere–ocean coupling and better agreement with reanalysis and observations. The SST biases increase with the forecast lead time in both seasons. Compared to its predecessor P5 (Stan et al. 2023), the warm SST biases in P6 improve moderately over the tropical Indo-west Pacific Ocean but deteriorate over the tropical east Pacific and tropical east Atlantic. These differences found here are also consistent with Stefanova et al. (2022).
Difference in the mean SST (K) between UFS reforecasts and observation (OISST) for the first four weeks of reforecasts during (a) JJAS and (b) DJFM. The reforecasts initiated from the 1st and 15th each month of JJAS and DJFM during 2011–18 are used in computing the bias. The black boxes covering 120°E–180°, 10°S–10°N and 15°–55°W, 5°–20°N are used to define the average SST bias over the tropical west Pacific (WP) and the tropical North Atlantic (TNA). Mean biases for the WP and TNA are shown in the top left and top right above each weekly plot, respectively.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
To isolate the impact of the tropical Pacific and Atlantic SST biases, two representative regions are selected. The area of 120°E–180°, 10°S–10°N is used to calculate the daily average WP SST bias, whose value is around 0.31 (0.23) K in week 1 and increases up to 0.63 (0.43) K in week 4 in JJAS (DJFM). The impact of WP SST bias will be explored in section 4b. The area of 15°–55°W, 5°–20°N over the Atlantic is used to represent the average SST biases over the TNA, which will be used later in section 4c to investigate the relationship between TNA SST and NASH. The TNA SST bias is 0.16 (0.03) K in week 1 and increases to 0.27 (0.13) K in week 4 in JJAS (DJFM). Other tropical oceanic regions may also impact the precipitation variability over the CONUS. For example, using EOF analysis, Krishnamurthy et al. (2021) reported that one of the leading modes of S2S variability in the USF model is related to ENSO, which is further related to tropical east Pacific SST. Since there are only two El Niño years (2014 and 2015) and two La Niña years (2016 and 2017) in the reforecasts, this study will not consider the role of tropical east Pacific and focus only on the WP and TNA regions.
Figure 2 shows the weekly precipitation biases defined as the difference between the UFS model and CPCU observations in the two seasons. Red colors show wetter conditions in the UFS than observations, signifying a wet bias in the model. In JJAS, an overall wet bias is found in the west of CONUS, except for the Pacific Northwest region, which shows a weak dry bias. A dry bias is seen in the middle of CONUS, the Florida peninsula, Nevada, Utah, and Arizona. Starting from week 2, a wet bias is also seen along the Appalachian Mountains and West Virginia/eastern Tennessee. Although varying slightly from week to week, the precipitation biases in JJAS are generally persistent and do not show a clear trend with the lead time. In DJFM, a wet bias dominates most of the CONUS, with the strongest magnitude in the east half of CONUS. This wet bias increases from week 1 to 3. The dry bias in the south along the Gulf of Mexico coast showing in weeks 1 and 2 shrinks in area in week 3 and re-emerges with enhanced amplitude in week 4. A weak dry bias in California is also found in weeks 1 and 2. The precipitation biases in P6 are generally consistent with P5, including their amplitude, distribution and weekly trends.
Difference in the mean precipitation (mm day−1) between UFS reforecasts and observation (CPCU) for the first four weeks of reforecasts during (a) JJAS and (b) DJFM. The reforecasts initiated from the 1st and 15th each month of JJAS and DJFM during 2011–18 are used in computing the bias.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
4. Impact of SST biases
a. Multivariate linear regression
In this subsection, we will first investigate heuristically the impact of SST biases from the two tropical regions using a multivariate linear regression method. In this method, daily CONUS precipitation biases for a week from the 56 reforecasts in a season (a total of 392 days) were regressed onto daily average SST biases indicated in the two black boxes in Fig. 1. Before applying the multivariate linear regression, we measured the multicollinearity with the variance inflation factor (VIF; Rawlings et al. 1998) between the two explanatory variables of SST biases. VIF is around 2.5 for boreal summer and is close to 1 in boreal winter, indicating that multicollinearity is not a problem in this case.
The perfect situation, lag-0 regression, was calculated, because the SST bias impact may differ in mechanism and lag days from the two different regions. In addition, because the persistence of SST biases is longer than that of precipitation, it is reasonable to obtain insights using the simplest situation.
Figures 3a and 3b show the slopes of multilinear regression of the precipitation biases on the WP and the TNA SST biases for each week in boreal summer. In any given week, a positive/negative slope represents precipitation bias increasing/decreasing with warm SST bias. The number of degrees of freedom used for the calculation of t value is an area average over the CONUS and not grid point values. In Fig. 3a, significant positive correlation is found around Oklahoma in week 1, indicating that the warm SST bias over the WP leads to increased precipitation (wet bias) in this area. Parts of Alabama and Georgia show significant negative correlation, which is equivalent to a dry bias. In week 2, the significant positive correlation moves to the east, while Florida is dominated by significant negative correlation. Week 3 is in contrast with the previous weeks in that much of the eastern United States shows negative correlation with the SST biases. In addition, the northwest corner shows significant positive correlation. Finally in week 4, significant positive correlation is seen in the mid–United States, while the negative correlation remains in part of the east coast. In Fig. 3b, week 1 precipitation does not show a clear pattern of significant correlation with the SST bias over the TNA, except for some scattered regions. From week 2 to 4, significant correlation is found mainly over the eastern half of the United States, but the signs of the correlation are not persistent from region to region and week to week. To estimate the fraction of precipitation bias explained by the warm SST biases, Figs. 3c and 4c further shows the statistically significant adjusted R2 of regression, corresponding to both WP and TNA SST biases. During boreal summer (Fig. 3c), as expected, regions of the highest adjusted R2 overlap with regions of significance. In general, the adjusted R2 increases with lead time, but the maximum magnitude is about 5%, indicating that the fraction of the summertime precipitation bias explained by the SST biases is small and other model errors such as physical parameterizations may play bigger roles.
Slopes of the weekly multivariate linear regression of JJAS bias in the daily mean precipitation (UFS reforecasts − CPCU) against bias in the daily mean SST (UFS reforecasts − OISST) of the (a) WP and (b) TNA for the first four weeks of reforecasts. The regions where the regression coefficient is statistically significant at the 0.05 level are stippled. Units are in mm day−1 per unit standard deviation of the SST time series. The reforecasts initiated from the 1st and 15th each month of JJAS and DJFM during 2011–18 are used in computing the regression. (c) The adjusted R2 of the multivariable linear regression. Stippling is from both (a) and (b) inclusively.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
Similarly, Figs. 4a and 4b show the slopes of multilinear regression of the precipitation biases on the WP and the TNA SST biases for each week in boreal winter. Both WP and TNA regions exert influence in the Southeast United States (however, weeks 1 and 3 for WP do not show significance and week 1 for the TNA only shows scattered regions of significance). The influence manifests as a northeast–southwest-oriented pattern, which persists from week 1 to 4. In addition, positive correlations with the WP SST bias persist from weeks 1 to 4 throughout the western and central United States, while week 3 shows significant positive correlation across the western and central United States. Only scattered regions of significant correlation are found in weeks 1, 2, and 4. While the west coastal regions also show significant correlation with the TNA, the sign of correlation is not consistent across the weeks. The inconsistency between the first two weeks and weeks 3 and 4 may be a manifestation of the transition in sources of predictability from weather to S2S time scales. For the first 2 weeks, initial condition may play a bigger role in impacting the CONUS precipitation, while starting from week 3, long time scale processes such as SST variability, the MJO, and other remote impacts take over. During boreal winter (Fig. 4c), similarly to summer, regions of the highest adjusted R2 overlap with regions of significance and the adjusted R2 increases with lead time. Compared to summertime, the fraction of wintertime precipitation bias explained by the SST biases is larger and up to 10% for some of the regions with significant correlation. The larger adjusted R2 may be associated with stronger tropical variability during boreal winter, which will be discussed in the next subsections. The regression of precipitation bias against the SST biases for both seasons in P6 is generally consistent with P5 (Stan et al. 2023), which is no surprise because biases for SST and precipitation are similar in the two prototypes.
b. Impact of tropical west Pacific SST bias
WAF is used to evaluate whether the impact of SST biases on CONUS manifests through the known tropical–extratropical teleconnection mechanisms (Stan et al. 2017). WAF can be expressed as a “snapshot” of a propagating package of quasigeostrophic wave disturbances. WAF is parallel to the group velocity of Rossby waves and is a good indicator of Rossby wave energy propagation (Takaya and Nakamura 2001). Figure 5 shows weekly 500-hPa horizontal WAF and perturbation streamfunction biases between the model reforecasts and ERAI reanalysis. In boreal summer (Fig. 5a) during week 1, a wave train originating near 120°E connects the WP and North America. The WAF bias over the North Pacific is eastward and therefore stronger than ERAI. In week 2, a discernable wave train of the streamfunction height bias appears and links the WP and CONUS through the extratropics, but the WAF bias does not have a uniform propagation direction. In weeks 3 and 4, WAF bias over the Pacific and the CONUS is generally westward (except for some of the WAF vectors near 170°E that are eastward) and the Rossby waves in the model are weaker than in ERAI, suggesting that impact of the WP on the CONUS in UFS model is weaker than in ERAI. In boreal winter (Fig. 5b), the WAF bias is generally stronger than in summer. In week 1, the WAF bias does not have a uniform propagation direction. A clear wave train of the streamfunction bias appears from week 2, connecting the North Pacific and the CONUS through the extratropics. However, the origin of the WAF bias does not seem to be in the WP. In weeks 3 and 4, the wave train stems from the WP and propagates to the CONUS through the extratropics, and the WAF in the model is generally weaker than in ERAI. In both seasons, a clear westward WAF bias connecting the WP and the CONUS through the extratropics emerges after week 3, indicating that the WP bias can influence the CONUS weather on the S2S range via Rossby wave train propagation. The weaker than observed WAF predicted by UFS suggests that in the model the mean state of WP has a weaker impact on CONUS than seen in reanalysis.
Weekly 500-hPa horizontal wave-activity flux biases (vectors) and perturbation streamfunction biases (shaded) between the UFS reforecasts and ERAI reanalysis for (a) JJAS and (b) DJFM.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
Convection can be directly influenced by the SST underneath, therefore, SST over the WP–Maritime Continent region can impact the CONUS weather through tropical–extratropical interactions generated by tropical convective activity (Zhang 1993; Sabin et al. 2013; Tseng et al. 2019). Numerous previous studies have also reported that the WP SST plays an important role in the propagation of the MJO (Arnold et al. 2013; Stan 2018; Roxy et al. 2019), which is one of the most important sources of predictability on the S2S time scale. The regions of significant correlation shown by Figs. 3 and 4 and WAF bias shown by Fig. 5 imply the WP SST bias may be able to affect the CONUS via the abovementioned teleconnections. Although the wave train pattern connecting the WP and North America has some bias, it is not entirely due to the WP SST bias alone. Other sources of errors in the model may also contribute to the WAF bias. For example, physical parameterizations may lead to biases in convection, which can also impact the CONUS weather and alter the response of the region to tropical forcing. In addition, the 500-hPa bias can also contribute to the WAF bias through the wave–mean flow interactions. Further investigation to isolate the effects of WP SST bias will be conducted in future studies.
c. Impact of tropical North Atlantic SST bias
Before evaluating the impact of TNA SST bias on the NASH, Fig. 6 shows the weekly 850-hPa geopotential height forecasted by P6 and ERAI. The 850-hPa geopotential height is used as a proxy for NASH instead of surface pressure to avoid the topographic effect. In both seasons, the model generally forecasts a realistic NASH in terms of its amplitude and position. Both observed and forecasted NASH are stronger and shifted to the west in boreal summer relative to winter, consistent with previous studies (Davis et al. 1997; Nigam and Chan 2009). In boreal summer, the reforecast shows a gradual northeast displacement of NASH relative to its observed position and the displacement increases with lead time. Conversely, during boreal winter, the observed and forecasted centers generally are collocated. To better illustrate changes in the forecasted NASH with lead time, Fig. 7 shows the NASH biases overlaid by NASH in ERAI. In boreal summer (Fig. 6a), a positive geopotential height bias is seen to the northeast of the observed center and its amplitude increases with time, consistent with the northeast displacement in Fig. 6a. The NASH’s center in UFS is slightly weaker than in ERAI. In boreal winter (Fig. 6b), a negative bias is seen near the NASH center in weeks 1–2. In week 3, a negative bias is almost collocated with NASH’s center, but a positive bias appears to the northwest of the observed NASH. This seemingly inconsistency in time may be related to the transition in model predictability from initial conditions to the boundary conditions after two weeks, which requires further investigation. In week 4, the negative bias is still near NASH’s center, but the positive bias in week 3 moves to the north of observed NASH.
Weekly 850-hPa geopotential height (gpm) of the UFS model reforecasts (shaded) overlaid by ERAI reanalysis (contours) for (a) JJAS and (b) DJFM. The 1560-gpm contour used to compute NASH’s ridge is emphasized in black.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
Weekly 850-hPa geopotential height (gpm) biases (UFS reforecasts − ERAI reanalysis; shaded) overlaid by ERAI reanalysis (contours) for (a) JJAS and (b) DJFM. The 1560-gpm contour used to compute NASH’s ridge is emphasized in black.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
Can the TNA SST influence NASH on a daily basis? To address this question, the daily bias of NASH’s center expressed as the 850-hPa geopotential height was regressed against the daily TNA SST bias. Here we used the average daily SST bias over the TNA (5°–15°N, 15°–45°W), which has the strongest correlation with the NASH. In boreal winter, the correlation coefficient between the daily TNA SST bias and daily bias of NASH’s center (27°–37°N, 20°–40°W) is −0.29, which is significant at 99% confidence interval. Hasanean (2004) also found a similar significant negative correlation (−0.43) between seasonal TNA SST and the NASH pressure during winter. In boreal summer, the correlation coefficient between the daily TNA SST bias and daily bias of NASH’s center (30°–38°N, 30°–50°W) bias is −0.13, significant at 95% confidence interval. The statistically significant negative correlations indicate that a warmer SST bias will likely contribute to a weaker NASH center.
Next, how NASH affects precipitation over the CONUS is investigated. Figure 8a shows the linear regression slope between the daily CONUS precipitation bias and daily bias of NASH’s center in winter. From week 1 to 4, precipitation in the Southeast United States shows persistent significant negative correlation with NASH’s center, suggesting that the simulated weaker NASH leads to enhanced precipitation in the Southeast United States. Using NCAR Community Atmosphere Model 3 (CAM3), Wang et al. (2007) found that NASH can modulate moisture transport from the tropical North Atlantic to the Caribbean Sea and the United States via the tropical easterly trade winds. To evaluate how moisture transport in the UFS model is impacted by the NASH bias, Fig. 8b shows the weekly differences in 850-hPa MFC and 850-hPa wind between the UFS reforecasts and ERAI reanalysis. Unlike Wang et al. (2007), where the Atlantic warm pool weakens the southwestern edge of NASH, the NASH in UFS P6 is weaker than in reanalysis at its center. Therefore, the associated circulation is also different. Figure 8b shows that UFS low level winds are stronger than ERAI from the Gulf of Mexico and the Atlantic toward the Southeast United States, leading to increased moisture transport from the Gulf of Mexico and the Atlantic in the model. Although the UFS 850-hPa specific humidity is slightly higher over the Gulf of Mexico (not shown), wind difference is found to dominate the differences seen in the MFC. The locations receiving increased moisture from the ocean are consistent with regions of significant negative correlation in Fig. 8a (i.e., the Southeast United States), supporting that the weaker NASH would lead to enhanced precipitation in the Southeast United States in winter. This result partly explains the wet bias over the eastern United States noted in Fig. 2b.
(a) Weekly linear regression between the daily precipitation bias (UFS reforecast − CPCU) and daily bias of the NASH’s center represented by 850-hPa geopotential height (UFS reforecast − ERAI) in DJFM. Regions where the regression coefficient is statistically significant at the 0.05 level are stippled. (b) Weekly differences in 850-hPa MFC (shaded) and 850-hPa horizontal wind (m s−1; vectors) between UFS reforecast and ERAI.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
In boreal summer, the NASH center in UFS is slightly weaker than in observations and displaced to the northeast relative to observations (Fig. 7a). Regression of the CONUS precipitation against the bias of the NASH’s center only shows scattered significance (not shown). However, previous studies have found that the western ridge of NASH plays an important role in regulating the U.S. summer rainfall. For example, using the NCEP/NCAR reanalysis, Li et al. (2012) categorized the seasonal western ridge of NASH into four locations and reported different precipitation patterns corresponding to each category. Johnson et al. (2020) also reported that the western portion of the NASH can affect summertime precipitation in the central United States in the FLOR coupled model. To examine the possible relationship between the NASH western ridge bias and the U.S. precipitation bias, we first represent the western ridge as the westmost extension of NASH at 1560 gpm, following Li et al. (2012). Because the position of the ridge is also affected by synoptic scale variability, a 5-day running mean is applied to daily geopotential height to remove the high-frequency variability. A comparison of the western ridge locations reveals displacements in both latitudinal and longitudinal directions. To evaluate the impact of simultaneous displacements in both directions, multilinear regression is used.
Figure 9 shows the multilinear regression slope between the daily CONUS precipitation bias and daily NASH western ridge displacement. In Fig. 9a, week 1 does not show any robust displacement in latitude. From week 2 to 4, a significant positive relationship is found over the states along the western part of the Gulf of Mexico, indicating that a northern displacement of the NASH ridge will lead to increased precipitation (wet bias) in this region. In addition, significant negative correlation is seen right to the north of the regions of positive correlation. In Fig. 9b, precipitation in week 1 shows significant positive correlation over Georgia and South Carolina, and significant negative correlation around Louisiana with displacement in longitude. Starting from week 2, significant positive correlation is found over the state of Texas, which spreads to the Great Plains and Mississippi River Valley in weeks 3 and 4. Positive correlation is also found over the southeast during week 3. The result suggests that an eastern displacement of the NASH ridge will lead to increased precipitation (wet bias) in this region. Excluding week 1 when the relationship between precipitation and the NASH ridge position is inconclusive, Fig. 9 illustrates that a northeast displacement of the NASH ridge tends to enhance precipitation (wet bias) over the Southeast United States and reduce precipitation (dry bias) to the north during later weeks. Although based on daily data, this relationship is consistent with Li et al. (2012), where a northeast NASH western ridge corresponds to increased rainfall over the Southeast United States and decreased rainfall in the northeast during the summer, with a southeast ridge corresponding to the opposite precipitation distribution. Diem (2006) also suggested that the NASH western ridge is inclined to locate westward during dry summers compared to wet summers.
Weekly multivariate linear regression of 5-day running mean precipitation bias (UFS reforecast − CPCU) against (a) latitude and (b) longitude of 5-day running mean NASH western ridge position bias (UFS reforecast − ERAI) in JJAS. Regions where the regression coefficient is statistically significant at the 0.05 level are stippled.
Citation: Weather and Forecasting 38, 6; 10.1175/WAF-D-22-0162.1
Overall, during boreal winter and summer, a warm TNA SST bias results in a weaker NASH’s center. In winter, the weaker forecasted NASH favors the moisture transport from the Gulf of Mexico, leading to enhanced precipitation (wet bias) over the Northeast United States in the forecast relative to observations. During summer, the NASH center in UFS is slightly weaker and does not have a significant impact over the CONUS precipitation. However, the position of the NASH western ridge can modulate the precipitation distribution in summer, i.e., a south/southeast wet–northeast dry bias distribution tends to emerge during a northeast displacement of NASH western ridge. Besides the TNA, SST along the Gulf Stream may also impact NASH. For example, Zhang et al. (2022) found a positive SST–NASH correlation along the Gulf Stream. They reported that the during the wet (dry) conditions of the Southeast United States, the warm (cold) SST and high (low) pressure center of the NASH lead to increased northward (southward) moisture transport and low-level convergence (divergence), which further produce upward motion and more decadal precipitation over the Southeast United States.
5. Conclusions
This study investigated the impact of warm SST biases of the tropical Pacific and the tropical Atlantic on the S2S precipitation biases over the CONUS in the UFS coupled model P6. The study also explored the physical mechanisms through which tropical SST biases over the two oceans remotely influence the CONUS precipitation. The impact of SST from two tropical regions—the tropical west Pacific and the tropical North Atlantic—was evaluated using multivariate linear regression. In both boreal summer and winter, SST biases over the two tropical oceans show significant regional influence on the CONUS precipitation bias. The fraction of CONUS precipitation biases explained by the tropical SST biases increases from week 1 to 4, up to about 5% in summer and 10% in winter.
The SST bias over the WP–Maritime Continent region can impact the CONUS weather remotely via tropical–extratropical interactions generated by tropical convection. In both seasons, a clear westward WAF bias connecting the WP and the CONUS through the extratropics emerges after week 3. The weaker than observed WAF forecasted by the model suggests that the WP in the model has a weaker impact over the CONUS than in ERAI on the S2S time scale via Rossby wave train propagation.
The SST bias over the TNA can remotely impact the CONUS precipitation through the direct impact on NASH. During boreal winter, a linear regression analysis shows that the warm SST bias weakens the center of NASH in the forecast. The weaker than observed NASH favors an enhanced moisture transport from the Gulf of Mexico and the Atlantic, leading to increased precipitation over the Southeast United States. During boreal summer, the NASH center is slightly weaker than in reanalysis but this bias does not have a significant impact on the CONUS precipitation. However, a northeast bias displacement of the NASH western ridge in summer modulates the precipitation distribution to have a south/southeast wet–northeast dry bias pattern.
This study used (multivariate) linear regression to investigate the correlation between the SST bias and the CONUS precipitation bias. However, correlation does not imply causation. In addition, it is difficult to untangle biases rooted in SST and in other error sources without conducting sensitivity experiments. Therefore, future investigation using mechanistic experiments is needed to quantify the impact of SST biases.
Acknowledgments.
This study was supported by the Unified Forecast System Research to Operation (UFS R2O) Project, which is jointly funded by NOAA’s Office of Science and Technology Integration (OSTI) of National Weather Service (NWS) and Weather Program Office (WPO), [Joint Technology Transfer Initiative (JTTI)] of the Office of Oceanic and Atmospheric Research (OAR) through NOAA Grant NA19NES4320002 [Cooperative Institute for Satellite Earth System Studies (CISESS)]. The study was partially supported by resources provided by the Office of Research Computing at George Mason University and funded in part by grants from the National Science Foundation (Awards 1625039 and 2018631).
Data availability statement.
UFS data used in this study are publicly available at https://registry.opendata.aws/noaa-ufs-s2s/. ERA-Interim data can be downloaded from https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era-interim. CPC Global Unified Precipitation and Temperature, and OISST2 data provided by the NOAA/OAR/ESRL PSL, Boulder, Colorado, can be downloaded from their website at https://psl.noaa.gov/data/gridded/.
REFERENCES
Adcroft, A., and Coauthors, 2019: The GFDL global ocean and sea ice model OM4. 0: Model description and simulation features. J. Adv. Model. Earth Syst., 11, 3167–3211, https://doi.org/10.1029/2019MS001726.
Arakawa, A., and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large-scale environment, part I. J. Atmos. Sci., 31, 674–701, https://doi.org/10.1175/1520-0469(1974)031<0674:IOACCE>2.0.CO;2.
Arcodia, M. C., B. P. Kirtman, and L. S. Siqueira, 2020: How MJO teleconnections and ENSO interference impacts U.S. precipitation. J. Climate, 33, 4621–4640, https://doi.org/10.1175/JCLI-D-19-0448.1.
Arnold, N. P., Z. Kuang, and E. Tziperman, 2013: Enhanced MJO-like variability at high SST. J. Climate, 26, 988–1001, https://doi.org/10.1175/JCLI-D-12-00272.1.
Balaguru, K., L. P. Van Roekel, L. R. Leung, and M. Veneziani, 2021: Subtropical eastern North Pacific SST bias in earth system models. J. Geophys. Res. Oceans, 126, e2021JC017359, https://doi.org/10.1029/2021JC017359.
Bretherton, C. S., M. Widmann, V. P. Dymnikov, J. M. Wallace, and I. Bladé, 1999: The effective number of spatial degrees of freedom of a time-varying field. J. Climate, 12, 1990–2009, https://doi.org/10.1175/1520-0442(1999)012<1990:TENOSD>2.0.CO;2.
Chen, F., Z. Janjić, and K. Mitchell, 1997: Impact of atmospheric surface-layer parameterizations in the new land-surface scheme of the NCEP mesoscale Eta model. Bound.-Layer Meteor., 85, 391–421, https://doi.org/10.1023/A:1000531001463.
Chen, J.-H., and S.-J. Lin, 2013: Seasonal predictions of tropical cyclones using a 25-km-resolution general circulation model. J. Climate, 26, 380–398, https://doi.org/10.1175/JCLI-D-12-00061.1.
Clough, S. A., M. W. Shephard, E. J. Mlawer, J. S. Delamere, M. J. Iacono, K. Cady-Pereira, S. Boukabara, and P. D. Brown, 2005: Atmospheric radiative transfer modeling: A summary of the AER codes. J. Quant. Spectrosc. Radiat. Transfer, 91, 233–244, https://doi.org/10.1016/j.jqsrt.2004.05.058.
Davis, R. E., B. P. Hayden, D. A. Gay, W. L. Phillips, and G. V. Jones, 1997: The North Atlantic subtropical anticyclone. J. Climate, 10, 728–744, https://doi.org/10.1175/1520-0442(1997)010<0728:TNASA>2.0.CO;2.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Diem, J. E., 2006: Synoptic-scale controls of summer precipitation in the southeastern United States. J. Climate, 19, 613–621, https://doi.org/10.1175/JCLI3645.1.
Ferranti, L., T. N. Palmer, F. Molteni, and E. Klinker, 1990: Tropical-extratropical interaction associated with the 30–60 day oscillation and its impact on medium and extended range prediction. J. Atmos. Sci., 47, 2177–2199, https://doi.org/10.1175/1520-0469(1990)047<2177:TEIAWT>2.0.CO;2.
Gamble, D. W., D. B. Parnell, and S. Curtis, 2008: Spatial variability of the Caribbean mid-summer drought and relation to North Atlantic high circulation. Int. J. Climatol., 28, 343–350, https://doi.org/10.1002/joc.1600.
Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev., 121, 764–787, https://doi.org/10.1175/1520-0493(1993)121<0764:PEOAUB>2.0.CO;2.
Guo, Z., P. A. Dirmeyer, and T. DelSole, 2011: Land surface impacts on subseasonal and seasonal predictability. Geophys. Res. Lett., 38, L24812, https://doi.org/10.1029/2011GL049945.
Harris, L. M., and S.-J. Lin, 2013: A two-way nested global-regional dynamical core on the cubed-sphere grid. Mon. Wea. Rev., 141, 283–306, https://doi.org/10.1175/MWR-D-11-00201.1.
Hasanean, H. M., 2004: Variability of the North Atlantic subtropical high and associations with tropical sea-surface temperature. Int. J. Climatol., 24, 945–957, https://doi.org/10.1002/joc.1042.
Hu, Z.-Z., B. Huang, Y.-T. Hou, W. Wang, F. Yang, C. Stan, and E. K. Schneider, 2011: Sensitivity of tropical climate to low-level clouds in the NCEP climate forecast system. Climate Dyn., 36, 1795–1811, https://doi.org/10.1007/s00382-010-0797-z.
Iacono, M. J., E. J. Mlawer, S. A. Clough, and J.-J. Morcrette, 2000: Impact of an improved longwave radiation model, RRTM, on the energy budget and thermodynamic properties of the NCAR community climate model, CCM3. J. Geophys. Res., 105, 14 873–14 890, https://doi.org/10.1029/2000JD900091.
Johnson, N. C., D. C. Collins, S. B. Feldstein, M. L. L’Heureux, and E. E. Riddle, 2014: Skillful wintertime North American temperature forecasts out to 4 weeks based on the state of ENSO and the MJO. Wea. Forecasting, 29, 23–38, https://doi.org/10.1175/WAF-D-13-00102.1.
Johnson, N. C., L. Krishnamurthy, A. T. Wittenberg, B. Xiang, G. A. Vecchi, S. B. Kapnick, and S. Pascale, 2020: The impact of sea surface temperature biases on North American precipitation in a high-resolution climate model. J. Climate, 33, 2427–2447, https://doi.org/10.1175/JCLI-D-19-0417.1.
Katz, R. W., M. B. Parlange, and C. Tebaldi, 2003: Stochastic modeling of the effects of large-scale circulation on daily weather in the southeastern U.S. Climatic Change, 60, 189–216, https://doi.org/10.1023/A:1026054330406.
Krishnamurthy, V., and C. Stan, 2022: Prediction of extreme events in precipitation and temperature over CONUS during boreal summer in the UFS coupled model. Climate Dyn., 59, 109–125, https://doi.org/10.1007/s00382-021-06120-0.
Krishnamurthy, V., and Coauthors, 2021: Sources of subseasonal predictability over CONUS during boreal summer. J. Climate, 34, 3273–3294, https://doi.org/10.1175/JCLI-D-20-0586.1.
Li, G., and S.-P. Xie, 2014: Tropical biases in CMIP5 multimodel ensemble: The excessive equatorial Pacific cold tongue and double ITCZ problems. J. Climate, 27, 1765–1780, https://doi.org/10.1175/JCLI-D-13-00337.1.
Li, L., W. Li, and Y. Kushnir, 2012: Variation of the North Atlantic subtropical high western ridge and its implication to Southeastern U.S. summer precipitation. Climate Dyn., 39, 1401–1412, https://doi.org/10.1007/s00382-011-1214-y.
Lin, H., R. Mo, F. Vitart, and C. Stan, 2019: Eastern Canada flooding 2017 and its subseasonal predictions. Atmos.–Ocean, 57, 195–207, https://doi.org/10.1080/07055900.2018.1547679.
Mariotti, A., and Coauthors, 2020: Windows of opportunity for skillful forecasts subseasonal to seasonal and beyond. Bull. Amer. Meteor. Soc., 101, E608–E625, https://doi.org/10.1175/BAMS-D-18-0326.1.
Minobe, S., A. Kuwano-Yoshida, N. Komori, S. P. Xie, and R. J. Small, 2008: Influence of the Gulf Stream on the troposphere. Nature, 452, 206–209, https://doi.org/10.1038/nature06690.
Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16 663–16 682, https://doi.org/10.1029/97JD00237.
National Centers for Environmental Prediction, 2019: The Global Forecast System (GFS)—Global Spectral Model (GSM). NOAA, accessed 1 March 2022, https://www.emc.ncep.noaa.gov/emc/pages/numerical_forecast_systems/gfs/documentation.php.
Nieto Ferreira, R., and T. M. Rickenbach, 2020: Effects of the North Atlantic subtropical high on summertime precipitation organization in the southeast United States. Int. J. Climatol., 40, 5987–6001, https://doi.org/10.1002/joc.6561.
Nigam, S., and S. C. Chan, 2009: On the summertime strengthening of the Northern Hemisphere Pacific sea level pressure anticyclone. J. Climate, 22, 1174–1192, https://doi.org/10.1175/2008JCLI2322.1.
Putman, W. M., and S.-J. Lin, 2007: Finite-volume transport on various cubed-sphere grids. J. Comput. Phys., 227, 55–78, https://doi.org/10.1016/j.jcp.2007.07.022.
Rawlings, J. O., S. G. Pantula, and D. A. Dickey, 1998: Applied Regression Analysis: A Research Tool. 2nd ed. Springer, 660 pp.
Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 5473–5496, https://doi.org/10.1175/2007JCLI1824.1.
Robertson, A., and F. Vitart, Eds., 2018: Sub-seasonal to Seasonal Prediction: The Gap between Weather and Climate Forecasting. 1st ed. Elsevier, 585 pp.
Roxy, M. K., P. Dasgupta, M. J. McPhaden, T. Suematsu, C. Zhang, and D. Kim, 2019: Twofold expansion of the Indo-Pacific warm pool warps the MJO life cycle. Nature, 575, 647–651, https://doi.org/10.1038/s41586-019-1764-4.
Sabin, T. P., C. A. Babu, and P. V. Joseph, 2013: SST–convection relation over tropical oceans. Int. J. Climatol., 33, 1424–1435, https://doi.org/10.1002/joc.3522.
Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 2185–2208, https://doi.org/10.1175/JCLI-D-12-00823.1.
Siqueira, L., and B. P. Kirtman, 2016: Atlantic near‐term climate variability and the role of a resolved Gulf Stream. Geophys. Res. Lett., 43, 3964–3972, https://doi.org/10.1002/2016GL068694.
Stan, C., 2018: The role of SST variability on the simulation of MJO. Climate Dyn., 51, 2943–2964, https://doi.org/10.1007/s00382-017-4058-2.
Stan, C., and B. P. Kirtman, 2008: The influence of atmospheric noise and uncertainty in ocean initial conditions on the limit of predictability in a coupled GCM. J. Climate, 21, 3487–3503, https://doi.org/10.1175/2007JCLI2071.1.
Stan, C., M. Khairoutdinov, C. A. DeMott, V. Krishnamurthy, D. M. Straus, D. A. Randall, J. L. Kinter III, and J. Shukla, 2010: An ocean-atmosphere climate simulation with an embedded cloud resolving model. Geophys. Res. Lett., 37, L01702, https://doi.org/10.1029/2009GL040822.
Stan, C., D. M. Straus, J. S. Frederiksen, H. Lin, E. D. Maloney, and C. Schumacher, 2017: Review of tropical-extratropical teleconnections on intraseasonal time scales. Rev. Geophys., 55, 902–937, https://doi.org/10.1002/2016RG000538.
Stan, C., and Coauthors, 2023: The impact of tropical Pacific SST biases on the S2S forecast skill over North America in the UFS global coupled model. J. Climate, 36, 2439–2456, https://doi.org/10.1175/JCLI-D-22-0196.1.
Stefanova, L., and Coauthors, 2022: Description and results from UFS coupled prototypes for future global, ensemble and seasonal forecasts at NCEP. NOAA/NCEP Office Note 510, 201 pp., https://doi.org/10.25923/knxm-kz26.
Strobach, E. J., 2021: Single column model evaluations of mixing length formulations and constraints for the sa-TKE-EDMF scheme. NOAA/NCEP Office Note 504, 28 pp., https://doi.org/10.25923/0sv0-4916.
Strobach, E. J., 2022: A single-column model evaluation of mixing length formulations and constraints for the sa-TKE-EDMF planetary boundary layer parameterization. Wea. Forecasting, 37, 1805–1828, https://doi.org/10.1175/WAF-D-21-0059.1.
Subramanian, A. C., and Coauthors, 2019: Ocean observations to improve our understanding, modeling, and forecasting of subseasonal-to-seasonal variability. Front. Mar. Sci., 6, 427, https://doi.org/10.3389/fmars.2019.00427.
Takaya, K., and H. Nakamura, 2001: A formulation of a phase-independent wave-activity flux for stationary and migratory quasigeostrophic eddies on a zonally varying basic flow. J. Atmos. Sci., 58, 608–627, https://doi.org/10.1175/1520-0469(2001)058<0608:AFOAPI>2.0.CO;2.
Tsartsali, E. E., and Coauthors, 2022: Impact of resolution on the atmosphere–ocean coupling along the Gulf Stream in global high resolution models. Climate Dyn., 58, 3317–3333, https://doi.org/10.1007/s00382-021-06098-9.
Tseng, K.-C., E. Maloney, and E. Barnes, 2019: The consistency of MJO teleconnection patterns: An explanation using linear Rossby wave theory. J. Climate, 32, 531–548, https://doi.org/10.1175/JCLI-D-18-0211.1.
Wang, C., S. Lee, and D. B. Enfield, 2007: Impact of the Atlantic warm pool on the summer climate of the Western Hemisphere. J. Climate, 20, 5021–5040, https://doi.org/10.1175/JCLI4304.1.
Wang, J., H. Kim, D. Kim, S. A. Henderson, C. Stan, and E. D. Maloney, 2020: MJO teleconnections over the PNA region in climate models. Part II: Impacts of the MJO and basic state. J. Climate, 33, 5081–5101, https://doi.org/10.1175/JCLI-D-19-0865.1.
Xie, P., M. Chen, and W. Shi, 2010: CPC unified gauge-based analysis of global daily precipitation. Preprints, 24th Conf. on Hydrology, Atlanta, GA, Amer. Meteor. Soc., 2.3A, https://ams.confex.com/ams/90annual/techprogram/paper_163676.htm.
Yang, F., A. Kumar, W. Wang, H.-M. H. Juang, and M. Kanamitsu, 2001: Snow albedo feedback and seasonal climate variability over North America. J. Climate, 14, 4245–4248, https://doi.org/10.1175/1520-0442(2001)014<4245:SAFASC>2.0.CO;2.
Zhang, C., 1993: Large-scale variability of atmospheric deep convection in relation to sea surface temperature in the tropics. J. Climate, 6, 1898–1913, https://doi.org/10.1175/1520-0442(1993)006<1898:LSVOAD>2.0.CO;2.
Zhang, W., B. Kirtman, L. Siqueira, B. Xiang, J. Infanti, and N. Perlin, 2022: Decadal variability of southeast U.S. rainfall in an eddying global coupled model. Geophys. Res. Lett., 49, e2021GL096709, https://doi.org/10.1029/2021GL096709.
Zhou, L., L. Harris, J.-H. Chen, K. Gao, B. Xiang, M. Tong, J. J. Huff, and M. Morin, 2022: Improving global weather prediction in GFDL SHiELD through an upgraded GFDL microphysics scheme. J. Adv. Model. Earth Syst., 14, e2021MS002971, https://doi.org/10.1029/2021MS002971.
Zhu, X., and C. Stan, 2015: Projection of summer precipitation over the southeastern United States in the superparameterized CCSM4. J. Climate, 28, 8052–8066, https://doi.org/10.1175/JCLI-D-14-00765.1.
Zorzetto, E., and L. Li, 2021: Impacts of the North Atlantic subtropical high on daily summer precipitation over the conterminous United States. J. Hydrometeor., 22, 1697–1712, https://doi.org/10.1175/JHM-D-20-0242.1.
Zuidema, P., J. Redemann, J. Haywood, R. Wood, S. Piketh, M. Hipondoka, and P. Formenti, 2016: Smoke and clouds above the southeast Atlantic: Upcoming field campaigns probe absorbing aerosol’s impact on climate. Bull. Amer. Meteor. Soc., 97, 1131–1135, https://doi.org/10.1175/BAMS-D-15-00082.1.