1. Introduction
Extreme summer climate anomalies in the United States, including droughts and floods, have been the subject of many past studies and are attracting renewed interest because of the widespread drought conditions in the past several years. Numerous studies have found a significant association of precipitation over continental United States with sea surface temperature (SST) anomalies in the tropical and North Pacific and in the Atlantic during the 1988 summer drought and 1993 summer flood. For example, Trenberth et al. (1988) and Namias (1991) suggested that the deficit in precipitation in the seasons preceding the event in association with extratropical SST pattern led to the 1988 drought. Trenberth and Guillemot (1996) indicated that the anomalous tropical SST played a key role in initiating the 1988 drought and 1993 flood, but the events were persisted and possibly enhanced by the resulting soil moisture anomalies. The low-level jet over Great Plains has been suggested to link the SST anomalies from surrounding oceans to the regional hydroclimate of the central United States (Weaver and Nigam 2008; Weaver et al. 2009a,b).
In addition to remote oceanic forcing influencing precipitation, land surface conditions provide an important local forcing for precipitation through water, energy, and momentum flux exchanges with the atmosphere. Following Namias (1952), many studies have since focused on land–atmosphere interactions, especially the feedback between precipitation and soil moisture over regions of the United States (e.g., Koster et al. 2004; Findell and Eltahir 2003a,b; Kim and Wang 2007). Soil moisture anomalies alter evapotranspiration, influencing the planetary boundary layer and thus affecting local convection that is responsible for most of the summertime precipitation over the United States. A potential positive feedback between soil moisture and precipitation tends to sustain anomalous hydrological conditions such as floods and droughts. This has been demonstrated in several studies that focused particularly on the 1988 drought and 1993 flood in the United States (e.g., Bosilovich and Sun 1999; Pal and Eltahir 2001). The atmospheric response to soil moisture anomalies induced by precipitation prolongs the precipitation anomalies. This promotes a long land memory and improves the predictability of the land–atmosphere system, which underlies the contribution of initial soil moisture to subseasonal and seasonal climate predictability and prediction (Dirmeyer 2000; Koster and Suarez 2001; Dirmeyer et al. 2009; Koster et al. 2010). Soil moisture may serve as a potentially useful predictor in subsequent precipitation over regions with both a strong land–atmosphere coupling and long land memory.
Several subregions of the United States have been identified as hotspots of strong land–atmosphere coupling (Koster et al. 2004, 2010; Wang et al. 2007), where initial soil moisture anomalies are expected to play an important role in the development of summer precipitation anomalies. Indeed, numerous modeling studies have demonstrated a positive feedback between soil moisture and precipitation over the Midwest and Great Plains regions, where wet (dry) soil tends to enhance (suppress) precipitation through its impact on evapotranspiration (Seth and Giorgi 1998; Pal and Eltahir 2001, 2002; Oglesby et al. 2002; Schubert et al. 2004; Koster et al. 2010, 2011; Kim and Wang 2007; Dirmeyer et al. 2009). Model parameterizations related to surface water budget may influence the strength of soil moisture impact in a specific model (Wu and Dickinson 2005; Wang et al. 2007). It is therefore necessary that observation-based analysis be used to gauge the model results. Mei and Wang (2012) used a conditional correlation analysis to derive a land–atmosphere coupling metric from observational data and model outputs. They found that the land–atmosphere coupling strength from the two is comparable for the Midwest and Great Plains subregions, and the regions of strong coupling based on observational data are more extensive than those in models. A study by Meng and Quiring (2009) found that spring soil moisture serves as a good predictor of summer precipitation over the Great Plains when El Niño summer SST anomalies are not persistent. Wu and Kinter (2009) performed a correlation analysis with SST anomalies, soil moisture, and observational drought indices and found a more significant contribution of soil moisture relative to that of SST anomalies in long-term droughts (more than 6 months) than in short-term droughts (less than 3 months) over the United States. Mei and Wang (2011) found that the role of soil moisture feedback is especially important in years of extreme precipitation anomalies (e.g., droughts and floods), and in years when prediction based on SST alone suffers from substantial errors. This is consistent with the findings of Koster et al. (2010) in the Global Land–Atmosphere Coupling Experiment, phase 2 (GLACE2), and of Mei et al. (2013) that the contribution of land surface initialization to subseasonal precipitation prediction skill in numerical models is stronger for more extreme conditions. It is evident that simulating or predicting droughts and floods over the United States needs to account for the impact of both large-scale forcing and local land surface feedback.
This study attempts to explore the role of late spring and early summer soil moisture in the development of extreme summer conditions focusing on the 1988 drought, 1993 flood, and the 2012 drought over the United States that feature very different large-scale forcings. While previous studies collectively show the role of initial soil moisture anomalies in the 1988 drought and 1993 flood, its role in the recent U.S. droughts is not well understood. Hoerling et al. (2013, 2014) indicated that the 2012 drought is likely a result of climate internal variability with little predictability. Their studies suggested that the 2012 drought was not triggered by low initial soil moisture, although it did not exclude the possibility for soil moisture feedback to enhance the drought after its onset. In this study, using a regional climate model coupled with a land surface model, we attempt to quantify the impact of soil moisture feedback on the 2012 drought and examine how it may be similar to or different from the 1988 drought and 1993 flood. Section 2 describes the data and model used in the study and experimental design. The model performance is evaluated is section 3. Results from the swapped soil moisture experiments are described and discussed in section 4. Section 5 presents the summary and conclusions.
2. Model, data, and experimental design
The regional climate model used in this study is the International Centre for Theoretical Physics (ICTP) limited-area Regional Climate Model, version 4.1 (RegCM4.1; Elguindi et al. 2011; Giorgi et al. 2012), coupled with the Community Land Model, version 4 (CLM4; Oleson et al. 2008; Wang et al. 2016a). RegCM4.1 is a hydrostatic, compressible, sigma-p vertical coordinate model run on an Arakawa B grid. Its dynamic core originated from the Fifth-generation Pennsylvania State University (PSU)–National Center for Atmospheric Research (NCAR) Mesoscale Model (MM5; Grell et al. 1994). The physics schemes in RegCM4.1 are based on the NCAR Community Climate Model, version 3 (CCM3; Kiehl et al. 1996); the planetary boundary layer scheme is developed by Holtslag et al. (1990); the Subgrid Explicit Moisture Scheme (SUBEX) is used as the large-scale precipitation scheme (Pal et al. 2000); and the Zeng and Beljaars (2005) parameterization for ocean surface fluxes is used. The model includes several different options for convective parameterizations, of which the Grell cumulus convection scheme (Grell 1993) with Fritsch and Chappell (1980) closure scheme is used here.
In this study, the model is run at a horizontal resolution of 30 km (~0.27°), with 18 sigma vertical levels over a domain covering the conterminous United States centered at (35.5°N, 97.5°W) using the Lambert conformal map projection with total grid points of 198 (west–east) × 150 (south–north; Fig. 1). The lateral boundary conditions used to drive RegCM4.1 were derived from the 6-hourly National Centers for Environmental Prediction–U.S. Department of Energy (NCEP–DOE) AMIP-II reanalysis data (R-2 hereafter; Kanamitsu et al. 2002) at 2.5° × 2.5°, including near-surface atmospheric temperature, pressure, humidity, and wind. The sea surface boundary condition used is from the NOAA Optimum Interpolation SST version 2 (OISSTv2) (Reynolds et al. 2002) at 1° × 1°.
RegCM domain with topography. The outlined box is the region over which spatial averages are taken.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
The capability of RegCM4.1–CLM4 in simulating the climate mean and extremes over the U.S. domain is evaluated based on a 27-yr simulation spanning 1986–2012, driven with initial and boundary conditions from the R-2 atmospheric data and OISSTv2 oceanic forcing data. With the year 1986 discarded as a spinup, model output from the remaining 26-yr period 1987–2012 is used to derive the model climatology. To evaluate the model performance, we compared the model climate with the Climatic Research Unit Time Series, version 3.21 (CRU), monthly gridded dataset at 0.5° resolution from the University of East Anglia (Jones and Harris 2013) for surface temperature and precipitation, with the R-2 data for atmospheric circulation and with the Global Land Data Assimilation System (GLDAS) data (Rodell et al. 2004) for soil moisture.
To identify the potential contribution of initial soil moisture to extreme anomalies of summer precipitation, a large number of simulations are conducted using RegCM4.1–CLM4, including a set of control simulations and a set of swapped experiments. All simulations start on 1 May and 1 June of several extreme years, including the 1988 and 2012 drought years and the 1993 flood year, and run through the summer season. The control simulations for each year are initialized with the soil moisture conditions produced by the 27-yr evaluation run for the corresponding specific year; in the swap experiments, in the whole domain and throughout the entire soil depth, soil moisture on 1 May and 1 June in each year is swapped with soil moisture of the same time from a different year, and the swapping is primarily between drought and flood years. Table 1 lists the details of the experimental design for soil moisture swapping.
Initial soil moisture swapping experiments.
Among the many aspects of land surface conditions that are intrinsically linked, the swapping of soil moisture alone (as is done in the experimental design here) may cause some inconsistency for example between soil temperature and soil moisture in the model. However, sensitivity experiments with simultaneous swapping of both soil moisture and temperature produce results that are essentially the same as experiments with the swapping of soil moisture alone; soil temperature does not add much information beyond what is already included in soil moisture anomalies. This is especially true for model precipitation; swapping soil moisture alone slightly underestimates the impact on air temperature (relative to swapping both), which is expected. The primary reason for the low sensitivity to soil temperature anomalies when soil moisture anomalies are already included is because, although not present at initialization, soil temperature anomalies can quickly develop as a result of soil moisture anomalies (and the resulting precipitation anomalies) because of the relatively slow variation of soil moisture.
3. Model performance
a. Climatology and extremes of surface climate
The precipitation and 2-m air temperature climatology of RegCM4.1–CLM4 (RegCM hereafter) for the March–May (MAM) and June–August (JJA) seasons (which are the focus of this study) are compared with the CRU data in Fig. 2. For precipitation, the model captures quite well the observed spatial contrast between the eastern and western United States. The percentage difference between model and CRU data (Fig. 2, right) indicates spatially extensive large wet bias (above 60% overestimation) over the western United States during MAM and dry bias (above 60% underestimation) over the U.S. Southwest during JJA. The overestimation of precipitation in areas with complex terrain (specifically the western United States) seems to be a common feature of many regional climate models, and the underestimation of precipitation over the southwest monsoon region in JJA was also reported by other studies (Leung et al. 2003; Liang et al. 2004; Duffy et al. 2006; Mearns et al. 2012; Wang et al. 2009; Tawfik and Steiner 2011). The model-simulated temperature is in good agreement with the CRU data. The difference (Fig. 2, right) shows a warm bias over the Great Plains region during all seasons, with a magnitude of 0–2 K during spring but larger (2–4 K) during subsequent seasons, consistent with previous studies (e.g., Mearns et al. 2012). RegCM’s performance across the United States is comparable with the regional climate models used in the North American Regional Climate Change Assessment Program (NARCCAP; Mearns et al. 2012), which generally perform well in simulating temperature patterns but suffer from substantial precipitation biases.
The 1987–2012 climatology of precipitation (mm day−1) and temperature (K) for MAM and JJA from (left) RegCM, (center) CRU, and (right) their difference (RegCM − CRU; % for precipitation and K for temperature).
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
Despite the large bias in modeling the mean climate, RegCM captures the magnitude and spatial pattern of precipitation anomalies in extreme years remarkably well. Figure 3 shows the precipitation anomalies in JJA simulated by the model and from the CRU data using 1988, 1993, and 2012 as examples. These are also the extreme summers this study focuses on. Comparison of precipitation anomalies in other extreme years shows a similar level of good agreement between model and data. A noticeable difference is found over the Southeast in 2012 where the model simulates little anomalies while the data shows strong wet anomalies.
JJA precipitation anomalies from (a)–(c) RegCM and (d)–(f) CRU, ET anomalies from (g)–(i) RegCM, and P − ET anomalies from (j)–(l) RegCM (left) 1988, (center) 1993, and (right) 2012.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
Also shown in Fig. 3 are the model-simulated anomalies of evapotranspiration (ET) and the anomalies of precipitation P minus ET (P − ET). When averaged at monthly or longer time scales, changes of atmospheric moisture storage are negligible, and mass conservation for water vapor in the atmosphere dictates that the atmospheric moisture convergence is balanced by P − ET (Trenberth et al. 2007; Seager et al. 2010). Therefore, at the time scale of interest in this study, P − ET is approximately equal to the atmospheric moisture convergence and will be referred to as such in the rest of the paper. In all three years, ET anomalies of interior North America show a southwest–northeast dipole pattern, with strong anomalies in the central United States and weaker anomalies of opposite sign in the region to the northeast spanning over the Great Lakes and surrounding areas. The areas of strong anomalies in atmospheric moisture convergence are more extensive and extend farther toward the northeast direction than those of strong ET anomalies. Over the central United States, anomalies of moisture convergence and ET are of the same sign, indicating that both contribute to the precipitation anomalies; over the Great Lakes and the surrounding region, the two are of opposite signs and the anomalies in atmospheric moisture convergence are much larger in magnitude and dominate the precipitation anomalies in these extreme summers.
In wet (dry) summers, higher (lower)-than-normal evapotranspiration means larger (smaller) evaporative cooling, leading to lower (higher)-than-normal temperature (Fig. 4). This mechanism is a major contributor for the heat waves that accompanied the droughts in 1988 and 2012 and the substantial cooling observed in the 1993 flooding summer. Although the model overestimates the magnitude of temperature anomalies, it captures the spatial pattern very well, including, for example, the south–north dipole pattern in 1988 and the southeast–northwest dipole pattern in 1993. In both the model and CRU data, the maxima of temperature anomalies are located northwest of precipitation anomalies in 1988 and 1993 but largely coincide with precipitation anomalies in 2012, reflecting possible differences in the role of large-scale circulation between the 2012 drought and other extreme years.
JJA 2-m air temperature anomalies (K) in (left) 1988, (center) 1993, and (right) 2012 from (a)–(c) RegCM and (d)–(f) CRU.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
Figure 5 shows the RegCM-simulated soil moisture anomalies in the top 10 cm of soil for each of the three extreme years in the months leading to summer (March–June). Over most of the United States, soil conditions from spring to early summer are predominantly drier than normal in 1988 and 2012 and predominantly wetter than normal in 1993. The central U.S. region shows the largest soil moisture anomalies, especially during the months of May and June. Figure 6 presents the monthly soil moisture anomalies based on GLDAS data for comparison with RegCM. The locations of the strongest soil moisture anomalies generally agree well between RegCM and GLDAS, with exceptions in the early spring months of 1988 and 2012 over Mexico and part of the southern Great Plains, where RegCM produces a strong wet signal while GLDAS data indicate a drought signal. The large magnitude of spring–summer soil moisture anomalies over the central United States makes it possible for soil moisture–precipitation feedback to contribute to the development of summer precipitation anomalies in these extreme years (Koster et al. 2010; Mei and Wang 2012), which will be closely examined in section 4.
Top 10-cm soil water equivalent depth (mm) for March–June from RegCM output for the years (top) 1988, (middle) 1993, and (bottom) 2012.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
Top 10-cm soil water equivalent depth (mm) for March–June from GLDAS data for the years (top) 1988, (middle) 1993, and (bottom) 2012.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
b. Atmospheric circulation in extreme years
Persistent large-scale atmospheric circulation anomalies have been associated with droughts in the United States (Namias 1982; Chang and Wallace 1987; Trenberth et al. 1988; Schubert et al. 2004). At 200 hPa, the model simulates a strong anticyclonic circulation over the northwestern United States during JJA of 1988 and 2012, which compares well with the R-2 data (Fig. 7). These large-scale anticyclonic atmospheric circulation patterns produce subsidence that inhibits the normal occurrence of midsummer rainfall and leads to prolonged hot conditions at the surface (Kunkel et al. 1996; Palecki et al. 2001; Xoplaki et al. 2003; Meehl and Tebaldi 2004; Fischer et al. 2007; Hoerling et al. 2014). On the other hand, both the model and R-2 data produce an anomalous cyclonic circulation centered over the northwestern United States (Fig. 7) in the wet 1993 summer. At the 850-hPa level, the anomalous cyclonic circulation in drought years is more centered over the Midwest in 1988. Geopotential height at 850 hPa shows a strong gradient along the southeast–northwest direction in all three years, decreasing toward the south or east in 1988 and 2012 while increasing in 1993. The similarity between model and R-2 data is remarkable in the geopotential height anomalies at 850 hPa. Figure 7 also reveals several differences between model and data in circulation patterns other than the dominant ones identified above. For example, in drought years, the model produces a cyclonic pattern at 200 hPa over northern Mexico in 1988 and farther south in 2012, while the reanalysis data place it over the U.S. Gulf Coast in 1988 and northern Mexico in 2012; in the flood year 1993, the reanalysis data show a strong anticyclonic pattern at 200 hPa over the eastern United States, but the model produces a very weak one.
Anomalies of wind (vectors) and of geopotential height (shading) at 200 and 850 hPa during JJA in the three extreme summers (left) 1988, (center) 1993, and (right) 2012, from RegCM and the R-2 data. The 10 and 4 m s−1 reference wind vectors are given below each panel of the figure.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
A primary source of water vapor in summer over the central United States is the Gulf of Mexico, with the transport of moist air inland and northward by mean southerly winds (Dirmeyer and Kinter 2009; Hoerling et al. 2014). In RegCM, the low-level (850 hPa) anomalous circulation during JJA 1988 shows the moist air being drawn out from the Midwest and Great Plains regions of the United States toward the Gulf of Mexico, which compares well with the R-2 data (Fig. 7). On the contrary, during JJA 1993, the 850-hPa wind anomalies draw moist air from the Gulf of Mexico inland toward the U.S. Midwest in the model, and this reversal in the wind anomalies pattern agrees well with the R-2 data (Fig. 7). For JJA 2012 in both the model and reanalysis data, the wind anomalies are much weaker than in 1988 and draw moist air mainly from the southern Great Plains and not as far inland as in 1988 (Fig. 7). By and large, the model and reanalysis data agree reasonably well in the main features of atmospheric circulation during these extreme summers.
4. Impact of initial soil moisture on summer extremes
The impact of soil moisture anomalies on subsequent precipitation can be sensitive to the timing of the anomalies and becomes increasingly important from spring toward summer when convective activities are most pronounced (Kim and Wang 2007). However, from the predictability and prediction point of view, information too late in the season is less useful because of the short lead time. In this study, 1 May and 1 June are used as the timing of soil moisture anomalies to examine the contribution of land–atmosphere interactions to the development of extreme summer drought or flood.
Figure 8 plots the precipitation differences in the three summers caused by the swapping of soil moisture between drought and flood years based on simulations starting on 1 May and 1 June (Figs. 8a–f). These differences (swapped − control) reflect the sensitivity of precipitation to soil moisture changes. Also plotted are the precipitation difference between the swapped experiments and the 1987–2012 climatology (Figs. 8g–l). These differences (swapped − mean) can be considered as what the precipitation anomalies would have been if the initial soil in 1993 were as dry as 1988 or if the initial soil in 1988 and 2012 were as wet as 1993.
(a)–(f) JJA precipitation difference (mm day−1) caused by the swapping of soil moisture (swapped − control) on 1 May and 1 Jun (left) 1988, (center) 1993, and (right) 2012; (g)–(l) JJA precipitation anomalies simulated by the 1 May and 1 Jun swapped experiments (swapped − mean) in the three years.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
The precipitation response to soil moisture anomalies is highly heterogeneous. Even the direction of the response is not spatially uniform. This is especially the case for the experiments with a 1 May starting date. Relative to the control, replacing the 1 June soil moisture in 1988 and 2012 with the wet condition from 1993 produces a predominantly wet signal, especially over areas of strongest precipitation deficit in the upper Mississippi River basin. This agrees qualitatively with findings from previous studies that support a positive soil moisture–precipitation feedback in that region (e.g., Kim and Wang 2007). For 1993, however, swapping the 1 June soil moisture with the dry condition from 1988 produces an additional increase of precipitation over the region of severe flood and some decrease of precipitation to the south. The magnitude of the precipitation response to soil moisture anomalies on 1 May is generally smaller than the response to soil moisture anomalies on 1 June (Figs. 8a–f). In addition, the summer precipitation changes following soil moisture swapping on 1 June are more dominated by local ET changes than 1 May (not shown), while the moisture convergence plays a more important role in the precipitation response to 1 May soil moisture swapping. In terms of spatial coverage, although soil moisture anomalies are applied over the entire domain, precipitation response is most evident over the Midwest and the central United States. As can be deduced from Figs. 5 and 6, large soil moisture differences between dry and wet years are not limited to these regions. The stronger response of precipitation in these regions is more a reflection of strong soil moisture–precipitation coupling (Koster et al. 2004; Wang et al. 2007).
Differences between Figs. 8g–l and Figs. 3a–c reflect how much soil moisture feedback may have influenced the severity and extent of the summer drought and flood. In 1988 and 2012, swapping the 1 June soil moisture with 1993 reduces both the spatial extent of the model-simulated drought and the severity of the drought in areas most influenced. The impact is especially strong in 2012. In 1993, however, replacing the 1 June soil moisture with the 1988 values has little impact on areas that are most influenced by the flood. For all three years, the impact of 1 May soil moisture swapping on precipitation is rather minimal. This is consistent with the findings of several previous studies (Bosilovich and Schubert 2001; Pal and Eltahir 2001; Oglesby et al. 2002) that land conditions had stronger influence during drought than during flood.
The soil moisture swapping impact on temperature is spatially more coherent than precipitation. As shown in Fig. 9, relative to the control simulations, replacing 1 June soil moisture in 1988 and 2012 with that of 1993 leads to cooling in the central United States in both years, and this cooling extends to the southeastern United States in the case of 2012; replacing 1 June soil moisture in 1993 with that of 1988 leads to a warming over the central United States and southern Great Plains. For all three years, swapping of the 1 May soil moisture produces a similar impact but of smaller magnitude. For 2012, both the 1 May and 1 June experiments produce a warming (relative to the control) over the region northwest of the cooling. Compared with temperature anomalies relative to the long-term mean (Figs. 9g–l vs Figs. 4a–c), the areas influenced by heat waves in 1988 and 2012 are significantly reduced as a result of the soil moisture swapping. However, in 1993, the warming induced by soil moisture swapping has little overlapping with the cooling anomalies in the control. This leads to a reduction of cooling anomalies over a fraction of the northern Great Plains and significant warming in the southern Great Plains.
As in Fig. 8, but for 2-m air temperature (K).
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
The differences in wind and geopotential height caused by soil moisture swapping in 1988 and 1993 (Fig. 10) show spatial patterns that are opposite of the anomalies in the corresponding control simulations, but have a much smaller magnitude. So the impact of soil moisture swapping on regional circulation within the model domain is consistent with the relationship between precipitation anomalies and circulation anomalies in the model control climate. For the year 2012, a general high developed at 200 hPa and a strong low at 800 hPa. While the circulation anomalies in the control simulations are much weaker in 2012 than in 1988 and 1993 (Fig. 7), the swapping-induced circulation changes are much stronger in 2012 than in 1988 and 1993 (Fig. 10).
Model differences in JJA wind field (vector) and geopotential height (shading) (at 200 and 850 hPa) caused by swapping soil moisture on 1 Jun 1988, 1993, and 2012.
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
The boxed region (35°–45°N, 100°–87.5°W) shown in Fig. 1 encompasses some of the strongest anomalies in precipitation, ET, and soil moisture in the three extreme summers considered in this study and is used here to derive the spatial average of surface climate variables for more detailed analysis. Figure 11 shows the spatially averaged anomalies of precipitation, ET, moisture convergence, 2-m air temperature, sensible heat flux, and moisture in the top 30 cm of soil in 1988, 1993, and 2012 from the control simulations, 1 May swapped experiments, and 1 June swapped experiments. In the control simulations, the precipitation anomalies are overwhelmingly dominated by anomalies in the atmospheric moisture convergence in 1993. The contributions from anomalies in local moisture supply (through ET) in 1988 and 2012 are larger than in 1993, especially during June–August, but are still generally smaller than the contribution from moisture convergence. Interestingly, in May, ET anomalies are positive in 1988 and 2012 and negative in 1993, although the negative soil moisture anomalies in 1988 and 2012 and positive soil moisture anomalies in 1993 all started in May and last through all summer. The positive ET anomalies in May of 1988 and 2012 likely result from the strong warming anomalies in these two years that enhance ET when temperature is still low. During the peak summer in July and August, the moisture limitation for ET becomes dominant, and ET influences temperature through evaporative cooling, contributing to the warm anomalies in 1988 and 2012 and cold anomalies in 1993. During most of the summer months, anomalies of sensible heat fluxes are in opposite direction of the ET anomalies.
JJA anomalies averaged over the boxed region shown in Fig. 1, for different variables in the three extreme summers (1988 in orange with triangle, 1993 in green with open circle, and 2012 in red with asterisk), based on the RegCM control simulation (solid lines), swapped experiments starting on 1 May (dotted lines), and swapped experiments starting on 1 Jun (dashed line).
Citation: Journal of Hydrometeorology 17, 8; 10.1175/JHM-D-15-0168.1
In 1993, the response of precipitation to soil moisture swapping is dominated by the response of atmospheric moisture convergence (Figs. 11a–c). Drier soil at the beginning of May and June in 1993 would have slightly reduced the positive precipitation anomalies, but precipitation would still have been significantly above normal for most of the summer. Despite the very weak response in precipitation to soil moisture swapping, the changed soil moisture anomalies persist throughout the summer season, and the resulting warming reduces the temperature anomalies by approximately 50% in the 1 June swapped experiment.
In 1988, the soil moisture swapping on 1 May and 1 June both lead to significant increases of ET throughout the summer, but precipitation in JJA increases only slightly in the 1 June experiment and even decreases in the 1 May experiment because of the compensating effects of the moisture convergence response. Note that despite the positive initial soil moisture anomalies in the swapped experiments, soil moisture drops to a level significantly below normal within a month because of the large ET made possible by the large warm anomalies. However, the soil moisture level is still higher than in the control. Compared to the control simulation, the magnitude of soil drought throughout the summer is reduced by more than 50% in the 1 June swapped experiment, and correspondingly summer heat is significantly reduced.
Precipitation in the summer of 2012 shows the strongest response to soil moisture swapping. Similar to 1988, wetter soil in 2012 would have caused a significant increase of ET, and ET in the 1 June swapped experiment is close to the 1993 level. Different from 1988, the moisture convergence response in 2012 is relatively weak. As such, the precipitation response to soil moisture swapping is dominated by the ET response. From the control to the 1 June swapped experiment, the magnitude of precipitation anomalies is reduced by approximately 50%, and a major fraction of the dry soil moisture anomalies are eliminated. The severity of summer heat as measured by temperature anomalies is also reduced by approximately 50%.
Based on both the spatial distribution and areal average of the model results, of the three extreme summers analyzed, 2012 shows the strongest response to a swapping of soil moisture between dry and wet conditions. The summer drought and heat of 2012 would have been less severe if soil moisture in the beginning of summer were wetter.
5. Summary and conclusions
In this study, a series of simulations using RegCM4.1–CLM4 were performed to investigate the impact of soil moisture anomalies on subsequent summer climate over the United States focusing on extreme drought and flood years. First, we evaluated the performance of the model in simulating the climatology of precipitation and surface temperature over the United States and in reproducing the surface climate and large-scale circulation anomalies in extreme years using the 1988 and 2012 droughts and 1993 flood as examples. We then investigated how initial soil moisture anomalies influence extreme summer conditions through swapping soil moisture between drought and flood years, using 1 May and 1 June as starting dates. The main findings are as follows:
Despite model biases in the mean precipitation and temperature in the western United States, the model performs remarkably well in capturing the magnitude and spatial extent of the climate anomalies during extreme summers and the associated anomalies in atmospheric circulation.
Results from the swapped soil moisture experiments support the general notion of positive soil moisture–climate feedback, with increase of soil moisture leading to higher ET and precipitation and lower temperature primarily over the Midwest and the Great Plains. The response in the 1 June experiments is much stronger than that in the 1 May experiments.
Swapping of the 1 June soil moisture reduces the magnitude and spatial extent of soil moisture and temperature anomalies during the subsequent summer, and this impact is consistent among all three years examined.
The degree of the reduction in precipitation anomalies caused by soil moisture swapping varies significantly from year to year. Of the three years examined, 1993 is the least sensitive and 2012 is the most sensitive.
Accurate soil moisture initialization contributes to improving precipitation prediction during the years of large precipitation anomalies (Koster et al. 2010, 2011; Mei and Wang 2011). Results from the present study indicate that early summer (1 June) soil moisture anomalies can be used to derive added forecast skill for seasonal precipitation prediction. There is, however, one caveat—the extreme summer predictions based on 1 June soil moisture anomalies leaves little lead time.
It is worth noting here that the model sensitivity to changes in soil moisture anomalies may depend on the mean climate of the model (Koster et al. 2006; Wang et al. 2007; Kim and Wang 2012). Although the model used in this study contains large biases in precipitation and temperature over the western and southwestern United States, it performs fairly well over the Midwest and central United States, regions that are the most influenced by the 1988 and 2012 droughts and 1993 flood. These regions are also regions of strong land–atmosphere coupling based on previous studies (Koster et al. 2004; Mei and Wang 2012; Mei et al. 2013). The decent performance of the model in simulating mean climate in these regions and the remarkable performance in reproducing the severity and extent of extreme drought and flood indicate that the most important processes underlying the soil moisture–precipitation feedback in these regions are correctly captured by the model.
The use of a regional climate model in the experiments conducted in this study poses some limitations and uncertainties. Specifically, driving the model with lateral boundary conditions from the R-2 data limits the impact of soil moisture swapping on large-scale circulation. For example, for the initial soil in 1993 to have been as dry as 1988, the large-scale atmospheric circulation for this to happen would have been markedly different, and the swapped soil moisture might have also influenced atmospheric circulation beyond the model domain. These processes were not captured in our experiments, which may underestimate the impact of soil moisture anomalies to the extent that remote interactions matter. One solution to this limitation is to derive the lateral boundary conditions for the regional climate model from a pair of swapping and control simulations using a GCM (e.g., Mei et al. 2013; Wang et al. 2016b), which would better capture the impact on large-scale circulation but suffer from biases and potential erroneous responses of the GCM (Wang et al. 2016b). On the other hand, the RCM’s limitation of not fully capturing the impact on large-scale circulation will be less of a concern when a large domain is used, as is in this study. For example, soil moisture anomalies were found to influence nonlocal climate through planetary waves (Koster et al. 2014), and the spatial scale of those waves is much smaller than the model domain used here.
Another consequence of using a regional climate model is that the variability of the model climate depends heavily on the lateral boundary conditions. The model internal variability when forced by the same boundary conditions and same initial soil moisture is small in this region, and it is hard to interpret what that variability means. This lack of strong internal variability is due to the lack of disturbances from and interactions with noises outside the model domain. It makes it hard to gauge the statistical significance of the difference between swapped experiment and control in this study because of the lack of a proper, physically meaningful reference for internal variability. However, what matters more is the magnitude of the differences caused by soil moisture swapping relative to the magnitude of the anomalies in the control simulations, and they are rather substantial as shown in the side-by-side comparison in Fig. 11.
Acknowledgments
Research documented in this paper was supported by funding from the NSF (AGS-1063986, AGS-1064008). The authors thank the two anonymous reviewers for their constructive comments.
REFERENCES
Bosilovich, M. G., and Sun W. Y. , 1999: Numerical simulation of the 1993 Midwestern flood: Land–atmosphere interactions. J. Climate, 12, 1490–1505, doi:10.1175/1520-0442(1999)012<1490:NSOTMF>2.0.CO;2.
Bosilovich, M. G., and Schubert S. D. , 2001: Precipitation recycling over the central United States diagnosed from the GEOS-1 Data Assimilation System. J. Hydrometeor., 2, 26–35, doi:10.1175/1525-7541(2001)002<0026:PROTCU>2.0.CO;2.
Chang, F.-C., and Wallace J. M. , 1987: Meteorological conditions during heat waves and droughts in the United States Great Plains. Mon. Wea. Rev., 115, 253–269, doi:10.1175/1520-0493(1987)115<1253:MCDHWA>2.0.CO;2.
Dirmeyer, P. A., 2000: Using a global soil wetness dataset to improve seasonal climate simulation. J. Climate, 13, 2900–2922, doi:10.1175/1520-0442(2000)013<2900:UAGSWD>2.0.CO;2.
Dirmeyer, P. A., and Kinter J. L. III, 2009: The Maya Express—Late spring floods in the US Midwest. Eos, Trans. Amer. Geophys. Union, 90, 101–102, doi:10.1029/2009EO120001.
Dirmeyer, P. A., Schlosser C. A. , and Brubaker K. L. , 2009: Precipitation, recycling, and land memory: An integrated analysis. J. Hydrometeor., 10, 278–288, doi:10.1175/2008JHM1016.1.
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, doi:10.1175/JCLI3669.1.
Elguindi, N., and Coauthors, 2011: Regional climatic model RegCM user manual version 4.1. Abdus Salam International Centre for Theoretical Physics, Trieste, Italy, 32 pp. [Available online at http://gforge.ictp.it/gf/download/docmanfileversion/21/544/ReferenceMan4.1.pdf.]
Findell, K. L., and Eltahir E. A. B. , 2003a: Atmospheric controls on soil moisture–boundary layer interactions. Part I: Framework development. J. Hydrometeor., 4, 552–569, doi:10.1175/1525-7541(2003)004<0552:ACOSML>2.0.CO;2.
Findell, K. L., and Eltahir E. A. B. , 2003b: Atmospheric controls on soil moisture–boundary layer interactions. Part II: Feedbacks within the continental United States. J. Hydrometeor., 4, 570–572, doi:10.1175/1525-7541(2003)004<0570:ACOSML>2.0.CO;2.
Fischer, E. M., Seneviratne S. I. , Vidale P. L. , Lüthi D. , and Schär C. , 2007: Soil moisture–atmosphere interactions during the 2003 European summer heat wave. J. Climate, 20, 5081–5099, doi:10.1175/JCLI4288.1.
Fritsch, J. M., and Chappell C. F. , 1980: Numerical prediction of convectively driven mesoscale pressure systems. Part I: Convective parameterization. J. Atmos. Sci., 37, 1722–1733, doi:10.1175/1520-0469(1980)037<1722:NPOCDM>2.0.CO;2.
Giorgi, F., and Coauthors, 2012: RegCM4: Model description and preliminary tests over multiple CORDEX domains. Climate Res., 52, 7–29, doi:10.3354/cr01018.
Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev., 121, 764–787, doi:10.1175/1520-0493(1993)121<0764:PEOAUB>2.0.CO;2.
Grell, G. A., Dudhia J. , and Stauffer D. R. , 1994: A description of the Fifth generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 121 pp., doi:10.5065/D60Z716B.
Hoerling, M., and Coauthors, 2013: An interpretation of the origins of the 2012 central Great Plains drought. NOAA Assessment Rep., 50 pp. [Available online at http://www.esrl.noaa.gov/psd/csi/factsheets/pdf/noaa-gp-drought-assessment-report.pdf.]
Hoerling, M., Eischeid J. , Kumar A. , Leung R. , Mariotti A. , Mo K. , Schubert S. , and Seager R. , 2014: Causes and predictability of the 2012 Great Plains drought. Bull. Amer. Meteor. Soc., 95, 269–282, doi:10.1175/BAMS-D-13-00055.1.
Holtslag, A. A. M., de Bruijn E. I. F. , and Pan H. L. , 1990: A high resolution air mass transformation model for short-range weather forecasting. Mon. Wea. Rev., 118, 1561–1575, doi:10.1175/1520-0493(1990)118<1561:AHRAMT>2.0.CO;2.
Jones, P., and Harris I. , 2013: CRU TS3.21: University of East Anglia Climatic Research Unit (CRU) Time-Series (TS) version 3.21 of high resolution gridded data of month-by-month variation in climate (Jan. 1901–Dec. 2012). NCAS British Atmospheric Data Centre, 1 April 2016, doi:10.5285/D0E1585D-3417-485F-87AE-4FCECF10A992.
Kanamitsu, M., Ebisuzaki W. , Woollen J. , Yang S.-K. , Hnilo J. J. , Fiorino M. , and Potter G. L. , 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 1631–1643, doi:10.1175/BAMS-83-11-1631.
Kiehl, J. T., Hack J. J. , Bonan G. B. , Boville B. A. , Briegleb B. P. , Williamson D. L. , and Rasch P. J. , 1996: Description of the NCAR Community Climate Model (CCM3). NCAR Tech. Note NCAR/TN-420+STR, 152 pp., doi:10.5065/D6FF3Q99.
Kim, Y., and Wang G. , 2007: Impact of initial soil moisture anomalies on subsequent precipitation over North America in the coupled land–atmosphere model CAM3–CLM3. J. Hydrometeor., 8, 513–533, doi:10.1175/JHM611.1.
Kim, Y., and Wang G. , 2012: Soil moisture–vegetation–precipitation feedback over North America: Its sensitivity to soil moisture climatology. J. Geophys. Res., 117, D18115, doi:10.1029/2012JD017584.
Koster, R. D., and Suarez M. J. , 2001: Soil moisture memory in climate models. J. Hydrometeor., 2, 558–570, doi:10.1175/1525-7541(2001)002<0558:SMMICM>2.0.CO;2.
Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 1138–1140, doi:10.1126/science.1100217.
Koster, R. D., and Coauthors, 2006: GLACE: The Global Land–Atmosphere Coupling Experiment. Part I: Overview. J. Hydrometeor., 7, 590–610, doi:10.1175/JHM510.1.
Koster, R. D., and Coauthors, 2010: Contribution of land surface initialization to subseasonal forecast skill: First results from a multimodel experiment. Geophys. Res. Lett., 37, L02402, doi:10.1029/2009GL041677.
Koster, R. D., and Coauthors, 2011: The second phase of the Global Land–Atmosphere Coupling Experiment: Soil moisture contributions to subseasonal forecast skill. J. Hydrometeor., 12, 805–822, doi:10.1175/2011JHM1365.1.
Koster, R. D., Chang Y. , and Schubert S. D. , 2014: A mechanism for land–atmosphere feedback involving planetary wave structures. J. Climate, 27, 9290–9301, doi:10.1175/JCLI-D-14-00315.1.
Kunkel, K. E., Changnon S. S. , Reike B. , and Arritt R. W. , 1996: The July 1995 heat wave in the Midwest: A climatic perspective and critical weather factors. Bull. Amer. Meteor. Soc., 77, 1507–1518, doi:10.1175/1520-0477(1996)077<1507:TJHWIT>2.0.CO;2.
Leung, L. R., Qian Y. , and Bian X. D. , 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, doi:10.1175/1520-0442(2003)016<1892:HOTWUS>2.0.CO;2.
Liang, X. Z., Li L. , Kunkel K. E. , Ting M. , and Wang J. X. L. , 2004: Regional climate model simulation of U.S. precipitation during 1982–2002. Part I: Annual cycle. J. Climate, 17, 3510–3528, doi:10.1175/1520-0442(2004)017<3510:RCMSOU>2.0.CO;2.
Mearns, L., and Coauthors, 2012: The North American Regional Climate Change Assessment Program: Overview of phase I results. Bull. Amer. Meteor. Soc., 93, 1337–1362, doi:10.1175/BAMS-D-11-00223.1.
Meehl, G. A., and Tebaldi C. , 2004: More intense, more frequent, and longer lasting heat waves in the 21st century. Science, 305, 994–997, doi:10.1126/science.1098704.
Mei, R., and Wang G. L. , 2011: Impact of sea surface temperature and soil moisture on summer precipitation in the United States based on observational data. J. Hydrometeor., 12, 1086–1099, doi:10.1175/2011JHM1312.1.
Mei, R., and Wang G. L. , 2012: Summer land–atmosphere coupling strength in the United States: Comparison among observations, reanalysis data and numerical models. J. Hydrometeor., 13, 1010–1022, doi:10.1175/JHM-D-11-075.1.
Mei, R., Wang G. L. , and Gu H. , 2013: Summer land–atmosphere coupling strength over the United States: Results from the regional climate model RegCM4–CLM3.5. J. Hydrometeor., 14, 946–962, doi:10.1175/JHM-D-12-043.1.
Meng, L., and Quiring S. M. , 2009: Observational relationship of sea surface temperature and precedent soil moisture with summer precipitation in the U.S. Great Plains. Int. J. Climatol., 30, 884–893, doi:10.1002/joc.1941.
Namias, J., 1952: The annual course of month-to-month persistence in climatic anomalies. Bull. Amer. Meteor. Soc., 33, 279–285.
Namias, J., 1982: Anatomy of Great Plains protracted heat waves (especially the 1980 U.S. summer drought). Mon. Wea. Rev., 110, 824–838, doi:10.1175/1520-0493(1982)110<0824:AOGPPH>2.0.CO;2.
Namias, J., 1991: Spring and summer 1988 drought over the contiguous United States—Causes and prediction. J. Climate, 4, 54–65, doi:10.1175/1520-0442(1991)004<0054:SASDOT>2.0.CO;2.
Oglesby, R. J., Marshall S. , Erickson D. J. III, Roads J. O. , and Robertson F. R. , 2002: Thresholds in atmosphere–soil moisture interactions: Results from climate model studies. J. Geophys. Res., 107, 4244, doi:10.1029/2001JD001045.
Oleson, K. W., and Coauthors, 2008: Improvements to the Community Land Model and their impact on the hydrological cycle. J. Geophys. Res., 113, G01021, doi:10.1029/2007JG000563.
Pal, J. S., and Eltahir E. A. B. , 2001: Pathways relating soil moisture conditions to future summer precipitation within a model of the land–atmosphere system. J. Climate, 14, 1227–1242, doi:10.1175/1520-0442(2001)014<1227:PRSMCT>2.0.CO;2.
Pal, J. S., and Eltahir E. A. B. , 2002: Teleconnections of soil moisture and rainfall during the 1993 Midwest summer flood. Geophys. Res. Lett., 29, 1865, doi:10.1029/2002GL014815.
Pal, J. S., Small E. F. , and Eltahir E. A. B. , 2000: Simulation of regional-scale water and energy budgets: Representation of subgrid cloud and precipitation processes with RegCM. J. Geophys. Res., 105, 29 579–29 594, doi:10.1029/2000JD900415.
Palecki, M. A., Changnon S. A. , and Kunkel K. E. , 2001: The nature and impacts of the July 1999 heat wave in the midwestern United States: Learning from the lessons of 1995. Bull. Amer. Meteor. Soc., 82, 1353–1367, doi:10.1175/1520-0477(2001)082<1353:TNAIOT>2.3.CO;2.
Reynolds, R. W., Rayner N. A. , Smith T. M. , Stokes D. C. , and Wang W. , 2002: An improved in situ and satellite SST analysis for climate. J. Climate, 15, 1609–1625, doi:10.1175/1520-0442(2002)015<1609:AIISAS>2.0.CO;2.
Rodell, M., and Coauthors, 2004: The Global Land Data Assimilation System. Bull. Amer. Meteor. Soc., 85, 381–394, doi:10.1175/BAMS-85-3-381.
Schubert, S. D., Suarez M. J. , Pegion P. J. , Koster R. D. , and Bacmeister J. T. , 2004: On the cause of the 1930s Dust Bowl. Science, 303, 1855–1859, doi:10.1126/science.1095048.
Seager, R., Naik N. , and Vecchi G. A. , 2010: Thermodynamic and dynamic mechanisms for large-scale changes in the hydrological cycle in response to global warming. J. Climate, 23, 4651–4668, doi:10.1175/2010JCLI3655.1.
Seth, A., and Giorgi F. , 1998: The effects of domain choice on summer precipitation simulation and sensitivity in a regional climate model. J. Climate, 11, 2698–2712, doi:10.1175/1520-0442(1998)011<2698:TEODCO>2.0.CO;2.
Tawfik, A. B., and Steiner A. L. , 2011: The role of soil ice in land–atmosphere coupling over the United States: A soil moisture–precipitation winter feedback mechanism. J. Geophys. Res., 116, D02113, doi:10.1029/2010JD014333.
Trenberth, K. E., and Guillemot C. J. , 1996: Physical processes involved in the 1988 drought and 1993 floods in North America. J. Climate, 9, 1288–1298, doi:10.1175/1520-0442(1996)009<1288:PPIITD>2.0.CO;2.
Trenberth, K. E., Branstator G. W. , and Arkin P. A. , 1988: Origins of the 1988 North American drought. Science, 242, 1640–1645, doi:10.1126/science.242.4886.1640.
Trenberth, K. E., Smith L. , Qian T. , Dai A. , and Fasullo J. , 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758–769, doi:10.1175/JHM600.1.
Wang, G., Kim Y. , and Wang D. , 2007: Quantifying the strength of soil moisture–precipitation coupling and its sensitivity to changes in surface water budget. J. Hydrometeor., 8, 551–570, doi:10.1175/JHM573.1.
Wang, G., Yu M. , Pal J. S. , Mei R. , Bonan G. , Levis S. , and Thornton P. , 2016a: On the development of a coupled regional climate–vegetation model RCM–CLM–CN–DV and its validation in Tropical Africa. Climate Dyn., 46, 515–539, doi:10.1007/s00382-015-2596-z.
Wang, G., Yu M. , and Xue Y. K. , 2016b: Modeling the potential contribution of land cover changes to the Sahel drought using a regional climate model: Sensitivity to lateral boundary conditions and experimental approach. Climate Dyn., doi:10.1007/s00382-015-2812-x, in press.
Wang, H., Schubert S. , Koster R. , Ham Y.-G. , and Suarez M. , 2014: On the role of SST forcing in the 2011 and 2012 extreme U.S. heat and drought: A study in contrasts. J. Climate, 15, 1255–1273, doi:10.1175/JHM-D-13-069.1.
Wang, S. Y., Gillies R. R. , Takle E. S. , and Gutowski W. J. , 2009: Evaluation of precipitation in the Intermountain Region as simulated by the NARCCAP regional climate models. Geophys. Res. Lett., 36, L11704, doi:10.1029/2009GL037930.
Weaver, S. J., and Nigam S. , 2008: Variability of the Great Plains low-level jet: Large-scale circulation context and hydroclimate impacts. J. Climate, 21, 1532–1551, doi:10.1175/2007JCLI1586.1.
Weaver, S. J., Ruiz-Barradas A. , and Nigam S. , 2009a: Pentad evolution of the 1988 drought and 1993 flood over the Great Plains: An NARR perspective on the atmospheric and terrestrial water balance. J. Climate, 22, 5366–5384, doi:10.1175/2009JCLI2684.1.
Weaver, S. J., Schubert S. , and Wang H. L. , 2009b: Warm season variations in the low-level circulation and precipitation over the central United States in observations, AMIP simulations, and idealized SST experiments. J. Climate, 22, 5401–5420, doi:10.1175/2009JCLI2984.1.
Wu, R. G., and Kinter J. L. III, 2009: An analysis of the relationship of U.S. droughts with SST and soil moisture: Distinguishing the time scale of droughts. J. Climate, 22, 4520–4538, doi:10.1175/2009JCLI2841.1.
Wu, W., and Dickinson R. E. , 2005: Warm-season rainfall variability over the U.S. Great Plains and its correlation with evapotranspiration in a climate simulation. Geophys. Res. Lett., 32, L17402, doi:10.1029/2005GL023422.
Xoplaki, E., González-Rouco J. F. , Luterbacher J. , and Wanner H. , 2003: Mediterranean summer air temperature variability and its connection to the large-scale atmospheric circulation and SSTs. Climate Dyn., 20, 723–739, doi:10.1007/s00382-003-0304-x.
Zeng, X., and Beljaars A. , 2005: A prognostic scheme of sea surface skin temperature for modeling and data assimilation. Geophys. Res. Lett., 32, L14605, doi:10.1029/2005GL023030.