1. Background
Land–atmosphere interactions contribute significantly in the warm season to regional extreme weather and climate (e.g., floods or droughts) (Beljaars et al. 1996; Giorgi et al. 1996; Bosilovich and Sun 1999; Pal and Eltahir 2002, 2003). Among the land hydrologic variables involved, soil moisture is arguably most important for its effects on a wide range of energy and moisture processes and, hence, the evolution of both local and remote atmosphere. Midtropospheric circulation systems may be linked to drought through anomalous surface sensible heating, possibly providing a positive feedback (e.g., Namias 1991; Hong and Kalnay 2002; Fischer et al. 2007). In addition, a soil moisture–air temperature feedback (e.g., Fischer et al. 2007; Vautard et al. 2007; Seneviratne et al. 2010) can exacerbate the soil moisture deficit and, thus, the lack of subsequent moisture supply to the atmosphere, a critical mechanism reinforcing drought.
By definition, soil moisture is largely lacking in a drought. How responsible is this lack of soil moisture for the accumulated precipitation deficit? Few studies have probed how various mechanisms may be responsible for such soil moisture–precipitation feedbacks affecting a drought event: for example, 1) the interaction between soil moisture anomalies and local atmospheric thermodynamic structures; 2) the reshaping of drought by this interaction: specifically, its effect on the more local-based convective precipitation processes; and 3) the external factors that control or modify this interaction. The coupling of soil moisture with the thermodynamic structure of the low-level atmosphere can trigger and maintain moist convection (e.g., Findell and Eltahir 1997, 2003a,b; Eltahir 1998; Hohenegger et al. 2009). For example, Findell and Eltahir (2003a,b) found that the local soil moisture–precipitation interaction is controlled by two parameters that characterize the low-level thermodynamic instability and moisture availability, respectively. Myoung and Nielsen-Gammon (2010a) identified two very similar but monthly mean parameters closely related to convective inhibition (CIN) and showed that these anomalies were strongly correlated with monthly convective precipitation over Texas. CIN is the energy barrier that an air parcel must overcome to rise from near the surface to its level of neutral buoyancy. They suggested that a soil moisture anomaly can modulate CIN by affecting near-surface dewpoint temperature and thermodynamic structure, while a large-scale circulation system can impact CIN by affecting the 700-hPa temperature and humidity. Hohenegger et al. (2009) showed that either an increase or a decrease of soil moisture could reduce the impacts of capping inversion above the boundary layer, with the sign of the precipitation response to the soil moisture anomaly depending on whether parameterized or explicit convection was used. Using a soil–plant hydrology model coupled with a simplified boundary layer (BL) model, Siqueira et al. (2009) indicated that the feedback of soil moisture on rainfall triggering (especially for dry soil) could be significantly dependent on the water vapor above the boundary layer (e.g., as from moisture provided by large-scale circulation).
This study investigates the soil moisture–precipitation feedback present in a regional drought. It uses, as an example, the month of April during the 2011 drought in the southern Great Plains (SGP; i.e., central and western Texas), eastern New Mexico, and central and western Oklahoma (see Fig. 1). From the 2010–11 winter until September 2011, this area experienced severe dryness (Nielsen-Gammon 2012); in particular, its anomalously dry conditions at the end of March persisted in April and into the summer. This record-breaking and quickly developed spring drought was representative of extreme events for the related area and set the stage for a prolonged summer and early autumn drought (Nielsen-Gammon 2012). The temperature of that April was unusually warm (with the daily average surface temperature above 20°C), facilitating active exchange of heat and moisture between land and atmosphere. This event provides a suitable test bed for examining soil moisture–precipitation coupling within a regional drought context.
(top) The NARR-observed April 2011 precipitation (mm) for the model domain between 20° and 50°N of North America. (bottom) The WRF control run (with KF convective parameterization) simulated April 2011 precipitation (mm) for the same area. The rectangular box in the figure represents the southern Great Plains in our study, i.e., where the prescribed anomalously wet soil moisture data are used to replace model-simulated soil moisture in the WET run.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
We use a regional climate model to address the following issues: 1) what is the general structure of simulated soil moisture–precipitation feedback in this drought event (e.g., the geographic features of both sign and strength of the feedback); 2) what could be the processes, from a local thermodynamic perspective, that can drive or regulate this soil moisture–precipitation feedback; and 3) what are the simulated large-scale atmospheric responses to the soil moisture anomaly, and how are they related to the regional soil moisture–precipitation coupling. Our paper is organized as follows. Section 2 introduces the model, dataset, and experimental design. Section 3 presents a brief evaluation of model control simulations. Section 4 elaborates on the regional latent heat (LH) flux and precipitation responses to the prescribed soil moisture anomaly as simulated by the model. Section 5 explores the physical and dynamical mechanisms that lead to the soil moisture–precipitation feedback. A discussion and concluding remarks are given in section 6.
2. Model, data, and experimental design
a. The regional climate model
The Weather Research and Forecasting (WRF) Model, fully compressible and nonhydrostatic, is used in this study. We employed WRF version 3.3, which contains a spectrum of options regarding grid nesting schemes, BL schemes, cumulus schemes, microphysics, shortwave and longwave radiation schemes, and land surface models (LSMs), among others. Boundary conditions include atmospheric states at different levels in the vertical, sea surface temperature (SST), topography, vegetation, soil data, et cetera. WRF has been frequently used in studies investigating the land–atmosphere coupling at local or regional scales (e.g., Jiang et al. 2009; Vivoni et al. 2009; Santanello et al. 2011).
b. North American Regional Reanalysis (NARR)
The NARR (Mesinger et al. 2006) is used for model evaluation as well as for initial and boundary conditions. It combines Eta model estimates with ground observations (ingested at hourly frequency where available) and atmospheric rawinsondes and dropsonde-observed temperature, wind and moisture profiles (ingested at 3-hourly frequency), and satellite radiances. Its fields are gridded at 32 km × 32 km in the horizontal and 3 h in time for 30 unequally spaced vertical pressure levels, and they span from 1979 to current. One essential feature of NARR is its assimilation of high-resolution precipitation observations, especially over the continental United States (CONUS), making it suitable for interpreting key mechanisms controlling land–atmosphere coupling (e.g., Luo et al. 2007; Dominguez and Kumar 2008a,b; Findell et al. 2011), as well as for verifying the model-represented local convection-related parameters.
c. Experimental design
To clarify the soil moisture–precipitation feedbacks in the SGP area, two groups of experiments were conducted for April 2011: a control (CTL) and a wet (WET) experiment. Both experiments have the same domain covering the entire CONUS on a 30-km horizontal grid and 29 vertical layers with uneven spacing and are driven by the initial and lateral boundary conditions derived from NARR. In each experiment, the following parameterization schemes are used: the Lin et al. (1983) microphysics scheme, the Yonsei University planetary BL scheme (Hong and Pan 1996), the simple cloud interactive radiation scheme (Dudhia 1989), the Rapid Radiative Transfer Model longwave radiation scheme (Mlawer et al. 1997), and the Noah LSM. To investigate the sensitivity of model results to cumulus parameterization, both the Kain–Fritsch (KF; Kain and Fritsch 1990) and Grell et al. (1994) schemes are tested. Additionally, each experiment contains six-member ensemble simulations starting from slightly different initial dates: 1800 UTC 30 March 2011, 0000 UTC 31 March 2011, 0600 UTC 31 March 2011, 1200 UTC 31 March 2011, 1800 UTC 31 March 2011, and 0000 UTC 1 April 2011. All other configurations are kept the same across the ensemble members. All simulations run a full month to 0000 UTC 1 May 2011.
The WET experiment, as defined above, is constructed similarly to an approach of previous studies [e.g., that of Pal and Eltahir (2002, 2003)], which used a regional climate model to identify soil moisture impacts on precipitation in CONUS. Such experiments with a fixed soil wetness anomaly help emphasize and clarify the response of the atmosphere to soil moisture perturbations. Additional experiments we did with only the initial soil moisture perturbed give similar results but show a weaker soil moisture feedback. The model-simulated feedback strength could also depend on the size of the area of perturbed soil moisture, whereas our selection (the box shown in Fig. 1) does not cover the whole area where a substantial soil moisture deficit occurred. To examine sensitivity to the soil moisture anomaly area, besides the above-mentioned CTL and WET simulations, we perform additional experiments that alter the anomaly area, as described in section 5f.
3. Model control simulation
This section compares the CTL simulation with the observationally derived NARR datasets to qualitatively evaluate the atmospheric model’s basic performance. Results are only presented for the KF cumulus scheme, because they are similar with the Grell scheme. All model results are based on an average of the ensemble simulations, unless explicitly specified otherwise.
a. Precipitation
Figure 1 compares the NARR-observed April precipitation with that of CTL. In general, the CTL performs reasonably well in capturing the geographic distribution of the observed precipitation, including the low-precipitation regime situated in the SGP area (ranging from, in both NARR and CTL runs, around 5 mm month−1 at the western edge to approximately 60 mm month−1 at the eastern edge) and the broad precipitation-maximum area located northeast of the SGP. The CTL run, however, somewhat overestimates the magnitude of precipitation for most of the domain, especially in the southeastern, north-central, and northwestern United States.
b. Atmospheric circulation
Figure 2 compares the differences between the CTL-simulated April wind, specific humidity, and temperature fields at various levels and the corresponding NARR data, showing that the 850-hPa wind speed and patterns are generally in agreement with the reanalysis. The 850-hPa specific humidity field is also reasonably well simulated but is somewhat underestimated in the Southeast, including the eastern portion of the SGP area (with a low bias around or below 0.6 g kg−1 in the SGP). A comparison at 925 hPa provides similar results (not shown).
Difference between the WRF CTL run and NARR of (a) the monthly averaged specific humidity (g kg−1) and wind (m s−1) at 850 hPa and (b),(c) monthly averaged temperature (K) and wind (m s−1) at 700 and 500 hPa, respectively.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
The wind and temperature at 700 and 500 hPa are also well simulated in general by the model (Figs. 2b,c), except for a cold bias, which is centered in the Great Plains area with a magnitude around 1.5 K at both pressure levels. Results at other levels (e.g., 600, 400, and 300 hPa) are similar, with only slightly different magnitudes of temperature bias. The following analyses suggest that these temperature biases do not significantly impact the model’s capability of representing the local thermodynamic stability, which is a function of temperature differences between two pressure levels.
Figure 3 compares the NARR-observed and CTL run–simulated monthly geopotential height (GPH) anomaly and wind anomaly fields at 500 hPa using the retrospective climatology of the period of 1979–2011. The model reproduces the major (anomalous) circulation patterns in the midtroposphere. Specifically, both results from the model and NARR indicate anomalous westerlies across the northern flank of the SGP, with a negative (positive) GPH anomaly located to the north (south). These circulation features imply storm tracks weakened for much of the SGP but enhanced farther north, and they provide the large-scale dynamic background of the SGP drought.
(top) NARR-observed and (bottom) WRF CTL run–simulated April 2011 GPH (m) and wind anomalies at 500 hPa from 1979–2011 observed and modelled climatologies.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
c. Atmospheric stability
Given the importance of atmospheric stability for regional-scale soil moisture–precipitation coupling (e.g., Hohenegger et al. 2009; Myoung and Nielsen-Gammon 2010a,b), we further assess the model’s ability to reproduce the observed lapse rate: particularly, its geographic distribution and variation with height. Figure 4 displays the monthly (and ensemble) averaged lapse rate for April 2011 at three levels [750, 650, and 550 hPa (comparisons at other levels provide similar results and are not shown)] at 2100 UTC (1500 LT for the SGP area; comparisons for other periods offer similar results), derived from NARR and the corresponding difference fields between CTL and NARR. We only examine the lapse rate within and surrounding the SGP, because we assume that its relationship to the soil moisture–precipitation feedback is mainly local. The NARR dataset (Fig. 4a) at 750 hPa shows large lapse rates on the western side of the SGP, characteristic of the weak stratification imposed by the daytime boundary layer but relatively stable stratification to the east characteristic of the capping layer above the daytime boundary layer. At 650–550 hPa, the stable capping layer is shifted westward, and the lapse rates to the east steepen to values characteristic of the midtroposphere, as do all lapse rates at higher levels. These features, in general, are reproduced well by the CTL simulation (Fig. 4b), although there are a number of discrepancies, including a modest underestimation in the south of the SGP at 550 hPa, a slight overestimation in the southwest at 650 hPa, and an overestimation in the north-central area at 750 hPa.
(a) The NARR data–derived monthly averaged lapse rates (°C km−1) at 2100 UTC (1500 LT for the SGP area) [comparisons for other periods (e.g., 1200 or 1800 UTC) give similar results] for (left)–(right) 750, 650, and 550 hPa. The lapse rate at a given pressure level is calculated using temperature data at the nearest levels (e.g., the 650-hPa lapse rate is derived using temperature at 600 and 700 hPa). (b) As in (a), but for the difference between WRF CTL and NARR. The WRF estimates are derived using ensemble averages.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
4. LH flux and precipitation responses to the wet soil moisture anomaly
We evaluate in this section the impacts of the wet soil moisture anomalies on the LH flux and precipitation. For precipitation, we focus on the local responses (i.e., over the SGP area). The remote effects of soil moisture via atmospheric teleconnection, though possibly important as demonstrated by studies investigating regional flooding (Bosilovich and Sun 1999; Pal and Eltahir 2002, 2003), are not discussed.
a. LH flux responses
Figure 5 shows the monthly (April) LH flux differences between the WET (with 70% effective saturation) and CTL simulations, both configured with the KF scheme (similar results are obtained with the Grell scheme). Significant responses are almost exclusively localized in the (soil moisture) anomaly region, where the LH flux anomaly is large and relatively uniform (exceeding 60 W m−2 almost everywhere): that is, there is a strong increase of evapotranspiration (ET), driven by the wetter soil and consistent with previous conclusions about the relatively high sensitivity of ET to a soil moisture anomaly in this area (e.g., Koster et al. 2004; Dirmeyer 2011) and with the background drought conditions and its anomalously warm air and strong surface heating.
The monthly ET difference (W m−2) between the wet (70% effective saturation) and control runs in the SGP and some nearby areas. The ET difference outside the SGP box is almost zero everywhere. Significant values of ET difference are concentrated in the southern Great Plains, where the wet soil moisture anomaly is added in the WET run (the soil moisture anomaly area in the WET run is outlined by the rectangular box). Refer to section 2c for the definition of effective saturation.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
b. Precipitation responses
How does the additional ET from the WET run [around ~(2–3) mm day−1, as shown in Fig. 5] affect precipitation? Figure 6 shows the monthly precipitation differences between the WET (with 70% effective saturation) and CTL simulations from both the KF and Grell simulations and with their convective and explicit components, respectively. The responses with both parameterizations show similar spatial patterns, highlighting a relatively strong increase of precipitation over the eastern portion of the SGP but a much smaller change for most of the western portion. These spatial patterns are largely reproduced by the convective precipitation component but absent in the large-scale precipitation (right column of Fig. 6), which has little west–east gradient within the SGP and is concentrated in a relatively small area at the northern end of the SGP.
April 2011 precipitation difference (mm) between the WET run (70% relative saturation) and control run for (left) total precipitation; (center) convective precipitation; and (right) large-scale precipitation, as well as for WRF using the (top) KF and (bottom) Grell convective parameterization; the brown box represents the southern Great Plains where the wet soil is prescribed (for the WET run).
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
To evaluate this spatial structure in greater detail, we divide the SGP area into six subareas according to their climatological surface pressure, which closely follow the west–east sloping terrain. Specifically, we use their upper and lower limits to denote these areas, which, extending from west to east, are 850–875 (also including a very small number of grids with surface pressure below 850 hPa), 875–900, 900–925, 925–950, 950–975, and 975–1000 hPa. Figure 7a shows the climatological surface pressure (in hPa) in the SGP and nearby area, thus providing geographical details about the six bands mentioned above. Also shown (Fig. 7b) is the CTL run–simulated ensemble mean soil moisture (averaged for the total 2-m soil column), which generally decreases from the east to west. In the following analyses, we compare the spatially averaged ET and precipitation changes (WET minus CTL) between these different subareas. The above approach facilitates our analyses of local thermodynamic structure and related quantities that are usually discussed in terms of pressure levels. Since the use of elevation to categorize subareas would provide almost equivalent results, we also term the subareas as elevation bands in our following text.
(a) The climatological surface pressure (hPa) in the SGP (and adjacent areas), which is used for dividing the SGP into six bands; (b) CTL run–simulated (ensemble average) April 2011 volumetric soil moisture content (m3 m−3, depth weighted for the total 2-m soil column) in the same area. The box in each figure represents the SGP where soil moisture is perturbed in the WET run.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
Figure 8 compares the ET anomalies (WET minus CTL) and efficiency ratios 
(a) The monthly ET difference (ΔET) between the WET run and control run; (b) the efficiency ratio for total precipitation [ΔP(total)/ΔET]; (c) the efficiency ratio for convective precipitation [ΔP(convective)/ΔET]; and (d) the efficiency ratio for large-scale precipitation [ΔP(large-scale)/ΔET] as a function of magnitude (%) of soil moisture relative saturation prescribed for the WET run (the x axis for each plot). All the estimates are derived from the monthly ET and precipitation (components) averaged over six different elevation bands in the southern Great Plains. The (ground) pressure level ranges used to define the elevation bands are provided. Effective saturation for each layer is defined by S = (θ − w)/(φ − w), where θ is the volumetric soil moisture content, φ is the porosity, and w is the wilting point.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
The efficiency ratio (
The box plot of the distribution of the ET efficiency ratio for total precipitation, ΔP(total)/ΔET, at each elevation band (subarea). The central bar represents the range from the lower to upper quarter (25th–75th percentiles). The upper (lower) dashed line represents the range from the upper (lower) quarter to the maximum (minimum) value. The horizontal line in the central bar represents the median value. The results are from the experiment using the KF convective parameterization and the wet soil moisture anomaly = 0.7. Samples are taken from the grids in the corresponding elevation band (subarea) and their monthly accumulated ET and precipitation are used.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
5. Processes related to the spatially different soil moisture–precipitation feedback strengths
The previous section has shown that the local soil moisture–precipitation feedback strength in the SGP drought event appears to be spatially heterogeneous, although the signs of the feedback were mostly positive. This section examines the underlying processes related to this pattern: that is, first, the local moisture convection-related mechanisms, in consideration of the highly consistent patterns between the total precipitation and its convective component shown in Figs. 8b and 8c, then the nonlocal (regional or large-scale) mechanisms (especially those present in the lower troposphere). The following analyses examine results derived from the KF parameterization and a 70% soil effective saturation for the WET run. Results derived from simulations using the Grell parameterization or different magnitudes of anomaly yield similar conclusions.
a. CAPE and CIN
Figure 10 displays the model-derived joint probability density function (pdf) of convective available potential energy (CAPE) and CIN for three subareas and for both the CTL run and the difference between WET and CTL simulations. The samples are taken from the hourly averaged outputs of ensemble mean (using six ensemble members produces similar results). The subareas examined are the locations with weak (850–975 hPa), intermediate (925–950 hPa), and strong (950–975 hPa) soil moisture–precipitation feedback strengths, respectively. With other conditions the same, a larger CAPE in this joint pdf favors stronger moist convection once it gets initiated (which is likely to produce more precipitation) and vice versa; while a lower CIN implies more frequent (larger likelihood) moist convection and vice versa. The increased ET affects these parameters in three ways. It increases moisture of the boundary layer, lowering the lifting condensation level (LCL) and raising its temperature, which lowers CIN and elevates CAPE. It also reduces sensible fluxes and the height of boundary layer, which thus leads to a decrease of the entrainment of dry air from above the boundary, further increasing the boundary layer humidity and its effect on CIN and CAPE. However, this cooling of the surface also lowers the temperature of convective plumes at the top of the boundary layer and, hence, weakens the net thermodynamic effects (via the alteration of CIN and CAPE) on the feedback.
(left) Probability density function, represented by the frequency of occurrence [in hours (i.e., hourly averages); the total size is 720 h], of the joint distribution of CAPE and CIN in the control run for (top to bottom) the elevation bands of 850–875, 925–950, and 950–975 hPa in the SGP. (right) The difference of the probability (of joint distribution of CAPE and CIN) between the wet and control run for the corresponding elevation bands. For each plot, the samples comprise the ensemble averaged model outputs at hourly frequency at those grids belonging to the corresponding elevation band. For the state space of CAPE (y axis), the bin (−inf, 0) denotes the situation where the local atmosphere is so stable that CAPE is nonexistent and deep convection is restricted. Note that the similarity of the results for the 975–1000 hPa-elevation band to those of 950–975 hPa and for the 875–900 and 900 hPa elevation bands to those of 850–875 hPa.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
The figure reveals some well-organized differences among the subareas for the CTL simulation (the left column of Fig. 10). The convective parameters of the 850–875-hPa band are characterized by low values of CAPE and high values of CIN that limit deep moist convection. In contrast, the band of 950–975 hPa more frequently favors a higher CAPE and lower CIN, which should sustain more and stronger moist convection. The band of 925–950 hPa, as expected, sits in between. More importantly, the wetter soil magnifies this spatial difference in the joint CAPE and CIN distribution (the right column of Fig. 10). At the subarea of 950–975 hPa, the wetter soil gives rises to the largest increase of probability of zones favoring deep convection (i.e., high CAPE and low CIN). For example, there is more than a 10-h increase in the frequency of occurrence of local soundings with CAPE larger than 2000 J kg−1 and CIN lower than 50 J kg−1. The higher-elevation subareas receive a progressively smaller increase in the occurrence in convection-favoring zones. Results for the three bands not shown are consistent with these shown (e.g., the band of 975–1000 hPa is similar to 950–975 hPa, and so on).
b. Cloud water content
The cloud water content (CWC) can either indicate the extent to which the LCL is lowered into the boundary layer or reflect the strength of moist convection and its associated precipitation. Here, we use the most western and eastern bands in the SGP to investigate the spatial variability of the CWC difference between the CTL and WET simulations and its relation to the soil moisture–precipitation feedback strength, as observed in Figs. 6, 8, and 9. Figure 11 presents the monthly mean difference of the CWC (WET − CTL) at each hour for the 850–875 hPa (west) and 975–1000 hPa (east) bands. For the most western band, significant increases of CWC associated with wetter soil only occur around local midnight (e.g., 0000–0400 LT), with minimal CWC change in the daytime (0900–1800 LT). The nighttime increments of CWC indicate low-level fog, a result of a relatively intense daytime moistening of the boundary layer by ET, and surface cooling at night. These results indicate that the wet soil has a negligible effect on local moist convection at the most western band, especially in the daytime, in agreement with the marginal CAPE and CIN changes for the western bands (Fig. 10).
The vertical distribution of cloud water content (g kg−1) difference between the WET and CTL simulations for the subareas of (left) 850–875 hPa and (right) 975–1000 hPa. For each pressure layer, the monthly, subarea, and ensemble averaged model outputs are provided for each hour average, beginning from 0600 UTC (0000 LT).
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
At the most eastern band, on the contrary, the CWC increases substantially from near the surface to around 300 hPa above the surface for both local night and daytime. These CWC increments extend to heights well above the boundary layer, suggesting enhanced (deep) moist convection, consistent with the significant CAPE and CIN changes in the WET run for the eastern bands (Fig. 10).
The aforementioned results (i.e., in sections 5a and 5b) are apparently linked to the spatially differential feedback strengths shown in Fig. 6 (especially for the convective precipitation). We further explore the local thermodynamic structures that are responsible for, or closely connected to, the above spatially varying patterns.
c. Atmospheric water availability
The occurrence of moist convection depends not only on humidity near the surface but also on that immediately above the boundary layer (e.g., Siqueira et al. 2009). The monthly averaged atmospheric precipitable water (PW) is used to assess the connections of moist convection to total water vapor and its change. Figure 12 shows the distribution of PW for the CTL and WET for the total atmospheric columns and for the two lowest layers [1–0.9 sigma and 0.9–0.7 sigma (i.e., from ground to 900 hPa and from 900 to 700 hPa) at a ground pressure level of 1000 hPa], respectively. The results are shown as boxplots for each subarea. They show that the monthly PW increases monotonically from the western to eastern bands. Figure 12 also shows that the additional water added by ET accumulates almost entirely in these lowest layers and is about 2 mm in most of the bands, being somewhat less for the western two.
The boxplots of monthly and ensemble averaged precipitable water, as in Fig. 9, in each elevation band (subarea) with samples taken from grids for the corresponding elevation band (subarea). Precipitable water (top) of the total column; (middle) between the model sigma layers 1 and 0.9; and (bottom) between the model sigma layers 0.9 and 0.7. Red (blue) represents the CTL (WET) simulation result.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
d. Atmospheric stability and humidity conditions
Soil moisture–precipitation feedback for a given area can be largely controlled by the low-level thermal and moisture conditions (Findell and Eltahir 2003a): in particular, the convective triggering potential (CTP) and dewpoint temperature depression (HIlow). CTP measures, in a skew T–logp map, the area between the ambient atmospheric temperature and a moist adiabatic from 100 to 300 hPa above ground. HIlow sums the dewpoint temperature depressions at 50 and 150 hPa above the ground. Both use atmospheric profiles at local early morning to emphasize free-atmosphere effects. Figure 13 summarizes the distributions of model-derived CTP and HIlow using box plots for each elevation band. The CTP decreases from approximately 150 J kg−1 at the western side to negative values on the eastern side. The wet surface decreases it by up to 50 J kg−1. Findell and Eltahir (2003a) found that wet soil would increase probability of convection for 5° ≤ HIlow < 10°C and CTP < 200 J kg−1. For HIlow 10° ≤ HIlow ≤ 15°C, convection could still occur, but preferentially with dry soil if CTP > 200 J kg−1. This CTP limit was derived for summer conditions and should be shifted downward with a colder surface (Su et al. 2013), as can occur in springtime. According to these criteria, the three areas on the western side are almost always too dry [with monthly HIlow above 20°C, and the daily HIlow always above 15°C (not shown)] for moist convection to occur, even after their soil is excessively wet. However, over the three areas on the eastern side, the additional soil moisture at some grid points reduces the monthly HIlow index to below the 15°C, with daily fluctuations dropping below 10°C (not shown), hence permitting moist convection under an appropriate CTP (which is generally near or below 100 J kg−1).
As in Fig. 12, but for (left) the CTP and (right) HIlow summed at 50 and 150 hPa above the ground. Both are derived from model results at 0600 LT (1200 UTC).
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
The extent to which HIlow is reduced and column PW is increased depends on the balance between moistening from ET and the air advected from its western side, which is of lower humidity than the air exiting on its eastern side. HIlow can be further affected by dry air entrained from the top of the boundary layer. The advection effect is strongest on the western edge, where entering air has not experienced any previous moistening so that HIlow is only reduced by about 5°C, consistent with the small increase of PW (Fig. 12). However, all the remaining areas experience a lowering of HIlow by about 10°C (~25% increase of relative humidity), indicating insensitivity to domain size. The west–east increase of relative humidity is somewhat less than the increase of PW, consistent with the lowering of elevation. We also note that our modeling results, in terms of the CTP and HIlow ranges in the SGP area, agree with, to some extent, the corresponding estimates from Findell and Eltahir (2003b) using long-term observational datasets in summer.
In addition, we found from evaluation of bulk lapse rate (as quantified by temperature difference between two arbitrary levels) between levels of 300 and 500 hPa above ground that the western bands (in both CTL and WET runs) have similar thermodynamic stability to the eastern bands. These results indicate that, at levels well above the boundary layer top, the westerns bands are unlikely to gain more convective strength than their eastern counterparts. These lower- and upper-level moisture and thermal conditions, as shown above, help shape a west–east contrast regarding potential (capability) of moisture convection in the SGP, which is consistent with those presented in Fig. 10.
e. Large-scale circulation consequences
1) Low-level circulation and temperature
Soil moisture anomalies can drive, to some degree, anomalous large-scale circulation patterns (e.g., Pal and Eltahir 2003). Figure 14 shows an anomalous anticyclonic flow (at low level) centered at the western and central SGP in response to the wetter soil, which tends to drive the water vapor out of those areas [demonstrated by the corresponding moisture flux anomaly map (not shown)] and leads to a relatively drier low-level atmosphere there. In particular, this moisture surplus advected from the western to eastern bands can increase their PW difference, which is consistent with the weaker PW increments at the western bands (Fig. 12). The nonlocal impacts of this moisture advection are further examined in section 5f. The larger GPH increase (for wet soil) at the western SGP is attributed to a stronger cooling effect of wet soil there: that is, a larger near-surface–low-level temperature decrease than that of the eastern area in the WET run (Fig. 15). This difference in temperature response is, in part, caused by the sloping terrain in the SGP area and, hence, the gradient in surface pressure, which leads to a higher sensitivity of low-level atmospheric temperatures to anomalous surface turbulent heating at the western side. Less cloudiness toward the west also contributes to this pattern (not shown). Note that, in the SGP area, the circulation response is largely confined to the lower troposphere, whereas in remote areas (e.g., U.S. Northeast), a significant response could extend to the midtroposphere (e.g., 500 hPa). Also, the significant low pressure anomaly in the U.S. Northeast (as shown in Fig. 14) could be related to the remote precipitation response (in that area) in a less straightforward way, the analysis of which is beyond the scope of this study.
The monthly GPH difference (m) and horizontal wind difference (m s−1) at (a) 925 and (b) 850 hPa between the WET and CTL simulations.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
As in Fig. 14, but for the monthly air temperature difference (K).
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
2) Moisture divergence and convergence
We further explore how an SGP soil moisture anomaly could affect large-scale moisture convergence–divergence patterns. First, we examine the monthly moisture divergence (convergence), integrated between the surface and 300 hPa (where water vapor is negligible), for the CTL run. The results (Fig. 16a) show distinct patterns between the SGP and broad areas east of it (e.g., the U.S. Midwest and Northeast), with the former, broadly speaking, dominated by moisture divergence and the latter by convergence. Within the SGP area, the divergent pattern is more pronounced in the western bands. These results suggest that, in the CTL run, the SGP (the areas to the east of it) serves as, in general, a moisture source (sink). This background flow structure helps explain the precipitation increase (in the WET run) in the area outside of the SGP (as shown in Fig. 6, which can reach up to around 100 mm). The water vapor surplus fueled by wetter soil that is not precipitated in the SGP tends to be advected downwind and rained out in places where moisture convergence is favored. That said, the structure shown in Fig. 16a might be strengthened given a wetter soil in the SGP. These have been demonstrated by Fig. 16b, where the SGP (areas east of it) has a stronger moisture divergence (convergence) in the WET run. The above result may be also induced by the effects shown in Fig. 14: that is, the more divergent pattern in the SGP is, in part, driven by a low-level wind anomaly in response to a wetter soil.
(a) Monthly (April 2011) moisture divergence (mm) for the local atmospheric column between bottom and 300 hPa in the CTL run (negative value represents moisture convergence). (b) As in (a), but for the difference between WET and CTL.
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
f. The nonlocal effects on precipitation responses
The model results regarding low-level circulation and moisture divergence responses to the wet soil moisture anomaly, as shown in Figs. 14 and 16, respectively, further imply that some nonlocal impacts arising from the wetter soil may help shape the spatial gradient of feedback strength. Specifically, the relatively strong increase of precipitation in the eastern SGP may result at least partially from the eastward moisture advection from the wetter western SGP. Similarly, the negligible feedback strength over the western SGP could be related to the location of the soil moisture anomaly area (in other words, its lack of extra moisture advection from its west). To examine these issues, we run three additional experiments: 1) apply the wet soil perturbation only over the eastern three bands of the original SGP domain (denoted as the East_WET run); 2) apply the wet soil perturbation only over the western three bands of the original SGP domain (denoted as the West_WET run); and 3) expand the original SGP to its west (see Fig. 17c for its geographic location) and apply the wet soil perturbation to this new domain (denoted as the Expand_WET run). For each experiment, all other configurations are the same as the original WET run: in particular, with the effective saturation S = 0.7 and KF as the convective parameterization.
The April precipitation difference (mm) between the CTL run and (a) the East_WET run, (b) the West_WET run, and (c) the Expand_WET run. Refer to the text for the definitions of these simulations. The solid brown box in (c) denotes the new wet soil moisture anomaly area to which Expand_WET run is configured. The dashed line in (c) denotes the western boundary for the original SGP area. (d) The difference between (c) and the top-left panel in Fig. 6 (i.e., how much precipitation is changed by expansion of the wet area).
Citation: Journal of Climate 30, 3; 10.1175/JCLI-D-13-00185.1
The monthly rainfall responses (each wet soil run minus CTL run) for the above experiments are presented in Fig. 17. Figure 17a clearly demonstrates that, without its western counterpart, a wetter soil situated over the eastern SGP alone leads to much less local precipitation increase: that is, only 31% of the increase in the original WET run (Fig. 6a). This result indicates that nonlocal soil moisture (anomaly) contributes to the feedback pattern shown in Fig. 6. This conclusion is further corroborated by the result from the West_WET run (Fig. 17b), where the wetter western SGP alone can bring an increase of more than 14 mm on average of rainfall to its east (i.e., around 50% of the increase over the eastern SGP shown in the original WET run) via moisture advection.
On the other hand, Figs. 17c and 17d indicate that the negligible (although positive) response of local precipitation in the original western SGP to wetter soil is not a result of absence of extra moisture advection from its west. In particular, with a westward expansion of the anomaly area, the local precipitation (in the western SGP) increases little beyond that of the WET run (Fig. 17d), with only some scattered enhancement of the rainfall response (in addition to the WET run). Apparently, the significant moisture advection (not shown) from its immediate west to the original western SGP would marginally affect its feedback strength, in contrast to what happens over the eastern SGP, as discussed above. This lack of sensitivity to nonlocal soil moisture impact may be largely related to the relatively strong subsidence over that region, which is supported by our additional evaluation of the vertical velocity (ω) fields at the mid-to-lower troposphere for both the CTL and WET simulations (the results are not shown here) and also implied by Figs. 16a and 16b. It is the prevailing subsidence flow that contributes to maintaining the extremely dry background atmosphere (in the CTL run) in the western SGP, as seen in Fig. 12.
6. Discussion and concluding remarks
Through examination of various atmospheric parameters, we have established why the soil moisture added to the western side of the SGP box produced very little feedback on precipitation, whereas the eastern side rendered much stronger (positive) feedback strength. In particular, several aspects regarding the local atmospheric thermodynamic properties are found responsible. First, the PW sets a stage for the feedback strength and its spatial distribution, with a very dry atmosphere over the western SGP maintained by subsidence and a much more moist atmosphere (partially from more convergent flow) on the eastern side. The regional low-level moisture and stability are thus conditioned so that at the western (eastern) side they very weakly (much more strongly) foster a positive soil moisture–precipitation feedback. All these properties jointly shape a negligible (appreciable) precipitation response at the western (eastern) SGP. Our results also indicate that, although the above features are sensitive to the magnitude of the soil moisture anomalies, their spatial patterns are largely invariant with it and are thus indicative of external control on the soil moisture–precipitation feedback revealed here. Furthermore, the impacts of soil moisture anomaly on large-scale atmospheric temperature, GPH, and wind and moisture divergence–convergence are found to be related to the spatial structure of feedback strength. Additional experiments also reveal that the nonlocal soil moisture effects providing moisture advection from the west affect significantly the local rainfall response at the eastern SGP but not the western SGP, where the background atmosphere is too dry and subsiding.
Additionally, our findings here are consistent with previous studies (e.g., Findell et al. 2011), which demonstrated a stronger potential of a morning ET anomaly to trigger afternoon precipitation in the summer over the eastern SGP (e.g., east of 100°W in their Fig. 1) than over the western SGP (west of 100°W). In particular, their approach hinges on a spatial categorization of low-level thermodynamic conditions according to the effects on soil moisture–convection coupling (Findell and Eltahir 2003a,b), a physical pathway also evaluated in our research. Our results also differ from some earlier work conducted for a similar area (e.g., Wei and Dirmeyer 2012), which examined the climatologic conditions of water vapor sources of precipitation and their linkage to ET for the period of 1979–2005 and demonstrated that summer precipitation responds actively to local ET anomalies for most of the SGP area (both west and east). This discrepancy might be attributed in part to the different conditions (i.e., climatology versus a particular drought year, respectively) and season (i.e., summer versus April, respectively) that are involved. The background circulation patterns characteristic of April 2011 (e.g., the GPH, wind and moisture transport fields at different pressure levels) may be somewhat unique in their dynamic and thermodynamic constraints on the local soil moisture–precipitation feedback (as suggested by our modeling results). A similar dependence of soil moisture–precipitation feedback on the tropospheric thermodynamic or dynamic features of a particular month or season has been implied previously (e.g., by Hohenegger et al. 2009), but a full understanding of this dependency requires further investigation.
Acknowledgments
This research was supported by the DOE (Grant DE-FG02-09ER64746) and NASA (Grant NNX11AE42G). Dr. Zong-Liang Yang is thanked for his support, and Dr. John Nielsen-Gammon and an anonymous reviewer are thanked for their constructive comments on the manuscript.
REFERENCES
Beljaars, A. C. M., P. Viterbo, M. Miller, and A. Betts, 1996: The anomalous rainfall over the United States during July 1993: Sensitivity to land surface parameterization and soil moisture anomalies. Mon. Wea. Rev., 124, 362–383, doi:10.1175/1520-0493(1996)124<0362:TAROTU>2.0.CO;2.
Bosilovich, M. G., and W. Y. Sun, 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.
Dirmeyer, P. A., 2011: The terrestrial segment of soil moisture–climate coupling. Geophys. Res. Lett., 38, L16702, doi:10.1029/2011GL048268.
Dominguez, F., and P. Kumar, 2008a: Precipitation recycling variability and ecoclimatological stability—A study using NARR data. Part I: Central U.S. plains ecoregion. J. Climate, 21, 5165–5186, doi:10.1175/2008JCLI1756.1.
Dominguez, F., and P. Kumar, 2008b: Precipitation recycling variability and ecoclimatological stability—A study using NARR data. Part II: North American monsoon region. J. Climate, 21, 5187–5203, doi:10.1175/2008JCLI1760.1.
Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 3077–3107, doi:10.1175/1520-0469(1989)046<3077:NSOCOD>2.0.CO;2.
Eltahir, E. A. B., 1998: A soil moisture–rainfall feedback mechanism: 1. Theory and observations. Water Resour. Res., 34, 765–776, doi:10.1029/97WR03499.
Findell, K. L., and E. A. B. Eltahir, 1997: An analysis of the soil moisture–rainfall feedback, based on direct observations from Illinois. Water Resour. Res., 33, 725–735, doi:10.1029/96WR03756.
Findell, K. L., and E. A. B. Eltahir, 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 E. A. B. Eltahir, 2003b: Atmospheric controls on soil moisture–boundary layer interactions. Part II: Feedbacks within the continental United States. J. Hydrometeor., 4, 570–583, doi:10.1175/1525-7541(2003)004<0570:ACOSML>2.0.CO;2.
Findell, K. L., P. Gentine, B. R. Lintner, and C. Kerr, 2011: Probability of afternoon precipitation in eastern United States and Mexico enhanced by high evaporation. Nat. Geosci., 4, 434–439, doi:10.1038/ngeo1174.
Fischer, E. M., S. I. Seneviratne, P. L. Vidale, D. Lüthi, and C. Schär, 2007: Soil moisture–atmosphere interactions during the 2003 European summer heat wave. J. Climate, 20, 5081–5099, doi:10.1175/JCLI4288.1.
Giorgi, F., L. Mearns, C. Shields, and L. Mayer, 1996: A regional model study of the importance of local versus remote controls of the 1988 drought and the 1993 flood over the central United States. J. Climate, 9, 1150–1162, doi:10.1175/1520-0442(1996)009<1150:ARMSOT>2.0.CO;2.
Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 138 pp., doi:10.5065/D60Z716B.
Hohenegger, C., P. Brockhaus, C. S. Bretherton, and C. Schär, 2009: The soil moisture–precipitation feedback in simulations with explicit and parameterized convection. J. Climate, 22, 5003–5020, doi:10.1175/2009JCLI2604.1.
Hong, S. Y., and H. L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124, 2322–2339, doi:10.1175/1520-0493(1996)124<2322:NBLVDI>2.0.CO;2.
Hong, S. Y., and E. Kalnay, 2002: The 1998 Oklahoma–Texas drought: Mechanistic experiments with NCEP global and regional models. J. Climate, 15, doi:10.1175/1520-0442(2002)015<0945:TOTDME>2.0.CO;2.
Jiang, X., G.-Y. Niu, and Z.-L. Yang, 2009: Impacts of vegetation and groundwater dynamics on warm season precipitation over the central United States. J. Geophys. Res., 114, D06109, doi:10.1029/2008JD010756.
Kain, J. S., and J. M. Fritsch, 1990: A one-dimensional entraining/detraining plume model and its application in convective parameterization. J. Atmos. Sci., 47, 2784–2802, doi:10.1175/1520-0469(1990)047<2784:AODEPM>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.
Lin, Y.-L., R. D. Farley, and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor., 22, 1065–1092, doi:10.1175/1520-0450(1983)022<1065:BPOTSF>2.0.CO;2.
Luo, Y., E. H. Berbery, K. E. Mitchell, and A. K. Betts, 2007: Relationships between land surface and near-surface atmospheric variables in the NCEP North American Regional Reanalysis. J. Hydrometeor., 8, 1184–1203, doi:10.1175/2007JHM844.1.
Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343–360, doi:10.1175/BAMS-87-3-343.
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, doi:10.1029/97JD00237.
Myoung, B., and J. W. Nielsen-Gammon, 2010a: The convective instability pathway to warm season drought in Texas. Part I: The role of convective inhibition and its modulation by soil moisture. J. Climate, 23, 4461–4473, doi:10.1175/2010JCLI2946.1.
Myoung, B., and J. W. Nielsen-Gammon, 2010b: The convective instability pathway to warm season drought in Texas. Part II: Free-tropospheric modulation of convective inhibition. J. Climate, 23, 4474–4488, doi:10.1175/2010JCLI2947.1.
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.
Nielsen-Gammon, J. W., 2012: The 2011 Texas drought. Texas Water J., 3, 59–95.
Pal, J. S., and E. A. B. Eltahir, 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., and E. A. B. Eltahir, 2003: A feedback mechanism between soil-moisture distribution and storm tracks. Quart. J. Roy. Meteor. Soc., 129, 2279–2297, doi:10.1256/qj.01.201.
Santanello, J. A., C. D. Peters-Lidard, and S. V. Kumar, 2011: Diagnosing the sensitivity of local land–atmosphere coupling via the soil moisture–boundary layer interaction. J. Hydrometeor., 12, 766–786, doi:10.1175/JHM-D-10-05014.1.
Seneviratne, S. I., T. Corti, E. L. Davin, M. Hirschi, E. B. Jaeger, I. Lehner, B. Orlowsky, and A. J. Teuling, 2010: Investigating soil moisture–climate interactions in a changing climate: A review. Earth-Sci. Rev., 99, 125–161, doi:10.1016/j.earscirev.2010.02.004.
Siqueira, M., G. Katul, and A. Porporato, 2009: Soil moisture feedbacks on convection triggers: The role of soil–plant hydrodynamics. J. Hydrometeor., 10, 96–112, doi:10.1175/2008JHM1027.1.
Su, H., R. E. Dickinson, K. L. Findell, and B. R. Lintner, 2013: How are spring snow conditions in central Canada related to early warm-season precipitation? J. Hydrometeor., 14, 787–807, doi:10.1175/JHM-D-12-029.1.
Vautard, R., and Coauthors, 2007: Summertime European heat and drought waves induced by wintertime Mediterranean rainfall deficit. Geophys. Res. Lett., 34, L07711, doi:10.1029/2006GL028001.
Vivoni, E. R., K. W. Tai, and D. J. Gochis, 2009: Effects of initial soil moisture on rainfall generation and subsequent hydrologic response during the North American monsoon. J. Hydrometeor., 10, 644–664, doi:10.1175/2008JHM1069.1.
Wei, J., and P. A. Dirmeyer, 2012: Dissecting soil moisture–precipitation coupling. Geophys. Res. Lett., 39, L19711, doi:10.1029/2012GL053038.
