Analyzing Atmosphere's Impact on Soil Moisture-Boundary Layer Interactions, Lab Reports of Environmental Science

Download Analyzing Atmosphere's Impact on Soil Moisture-Boundary Layer Interactions and more Lab Reports Environmental Science in PDF only on Docsity! 552 VOLUME 4J O U R N A L O F H Y D R O M E T E O R O L O G Y q 2003 American Meteorological Society Atmospheric Controls on Soil Moisture–Boundary Layer Interactions. Part I: Framework Development KIRSTEN L. FINDELL Geophysical Fluid Dynamics Laboratory, Princeton University, Princeton, New Jersey ELFATIH A. B. ELTAHIR Parsons Laboratory, Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts (Manuscript received 6 March 2002, in final form 26 November 2002) ABSTRACT This paper investigates the influence of soil moisture on the development and triggering of convection in different early-morning atmospheric conditions. A one-dimensional model of the atmospheric boundary layer (BL) is initialized with atmospheric sounding data from Illinois and with the soil moisture set to either extremely wet (saturated) or extremely dry (20% of saturation) conditions. Two measures are developed to assess the low- level temperature and humidity structure of the early-morning atmosphere. These two measures are used to distinguish between four types of soundings, based on the likely outcome of the model: 1) those soundings favoring deep convection over dry soils, 2) those favoring deep convection over wet soils, 3) those unlikely to convect over any land surface, and 4) those likely to convect over any land surface. Examples of the first two cases are presented in detail. The early-morning atmosphere is characterized in this work by the newly developed convective triggering potential (CTP) and a low-level humidity index, HIlow. The CTP measures the departure from a moist adiabatic temperature lapse rate in the region between 100 and 300 mb (about 1–3 km) above the ground surface (AGS). This region is the critical interface between the near-surface region, which is almost always incorporated into the growing BL, and free atmospheric air, which is almost never incorporated into the BL. Together, these two measures form the CTP-HIlow framework for analyzing atmospheric controls on soil moisture–boundary layer interactions. Results show that in Illinois deep convection is trigged in the model 22% of the time over wet soils and only 13% of the time over dry soils. Additional testing varying the radiative conditions in Illinois and also using the 1D model with soundings from four additional stations confirm that the CTP-HIlow framework is valid for regions far removed from Illinois. 1. Introduction Feedbacks between the land surface and the atmo- sphere have been the focus of much recent inquiry into questions ranging from the maintenance of extreme drought or flood conditions, to the influence of defor- estation on rainfall, to responses to increases in atmo- spheric concentrations of greenhouse gases. Many stud- ies of the midwestern U.S. drought of 1988 and flood of 1993, for example, suggest that the soil moisture condition in these cases helped to sustain the extreme circumstances throughout the summer (Trenberth and Guillemont 1996; Trenberth et al. 1988; Atlas et al. 1993). Others suggest that there is actually a negative Corresponding author address: Dr. Kirsten L. Findell, Geophysical Fluid Dynamics Lab, Princeton University, P.O. Box 308, Princeton, NJ 08542. E-mail: kirsten@alum.mit.edu feedback between soil moisture and drought (flood) con- ditions (Giorgi et al. 1996; Paegle et al. 1996). Ek and Mahrt (1994) caution against extending the results of these studies to all locations and synoptic settings. With a one-dimensional model initialized with data from the Hydrological–Atmospheric Pilot Experiment–Modélis- ation du Bilan Hydrique (HAPEX–MOBILHY) exper- iment, they show that the influence of the land surface on the development of boundary layer (BL) clouds is highly dependent on the initial (early morning) condi- tion of the atmosphere. Furthermore, modeling studies are also dependent on factors such as the convection scheme (Pan et al. 1996; Pal 1997), the domain size (Seth and Giorgi 1998), or the vertical resolution of the model and/or the forcing data. This work is an effort to define the physical mech- anisms controlling the interactions between the land sur- face and the atmospheric boundary layer (BL). Specif- ically, this paper addresses the question of how the ear- JUNE 2003 553F I N D E L L A N D E L T A H I R ly-morning atmospheric thermodynamic structure af- fects the interactions between fluxes from the land surface (and thus the soil moisture state) and the growth and development of the BL, leading to the triggering of convection. There are three main characteristics of the early- morning atmospheric structure that significantly influ- ence the nature and evolution of the boundary layer during the course of the coming day: • the properties of the residual layer, since this air will quite likely be incorporated into the BL (e.g., Chen and Avissar 1994; Rabin et al. 1990; Cutrim et al. 1995; Rabin and Martin 1996); • the depth of the nocturnal stable layer (e.g., Wetzel et al. 1996; Segal et al. 1995), since this will determine the ability of the growing BL to reach beyond the air of this near-surface stable layer and the time at which it does so; and, • the height and strength of the inversion separating the mixed layer from the overlying free atmosphere (e.g., Betts et al. 1996; Ek and Mahrt 1994; Mahrt 1997; Mahrt and Pierce 1980; Segal et al. 1995), since this affects both the rate of entrainment of overlying air into the developing BL, and the buildup of moisture and moist static energy in the mixed layer. A few studies have investigated the influence of varying one or more of these properties, notably Ek and Mahrt (1994), Chen and Avissar (1994), and Segal et al. (1995). There is need, however, for a measure that as- sesses the combined effects of these components of the early-morning atmospheric structure on the potential for vegetation and/or soil moisture to influence the devel- opment of convection. We describe such a measure in this paper. The convective triggering potential (CTP) focuses on the temperature lapse rate between 100 and 300 mb (about 1–3 km) above the ground surface (AGS). The CTP was developed from work with a one-dimensional boundary layer model initialized with sounding data from Illinois. It is coupled with a low-level humidity index, HIlow, to better describe early-morning atmo- spheric conditions and help diagnose the likelihood for deep convection during that day. The model is briefly described in section 2. In section 3 we define the CTP and the HIlow, and in section 4 we present two case studies highlighting the advantage of dry soils in high CTP environments and wet soils in intermediate CTP environments. In section 5, we present the full suite of model results generated using initial soundings from the summers of 1997–99 at a station in central Illinois. These results show that deep convection is more likely to occur in the model over wet soils than over dry soils, given the atmospheric environment of central Illinois. This is consistent with the small but significant positive soil moisture–rainfall feedback pre- viously reported for Illinois (Findell and Eltahir 1997, 1999). The results in section 5 are presented in the context of the CTP-HIlow framework. The framework is used to distinguish between four different atmospheric regimes: one where it is easier for high sensible heat flux regions to trigger deep convection, another where it is easier for high latent heat flux regions to trigger deep con- vection, a third where the atmosphere is so dry and/or stable that deep convection is unlikely over any surface, and a fourth where the atmosphere is very humid and marginally unstable so that convection is very likely over any surface. Note that in both the third and fourth conditions, the likelihood of deep convective activity is independent of the surface flux partitioning. To provide further support for the extension of this framework beyond the original development location, section 6 includes analyses of 1D results from four ad- ditional stations. These additional results confirm that the CTP-HIlow framework is a robust indicator of soil moisture–rainfall feedbacks. Sections 7 and 8 include a brief discussion followed by the conclusions of this work. 2. Model description The model used in this work is a modified version of Kim and Entekhabi’s (1998a,b) mixed-layer model of the surface energy budget and the planetary boundary layer (PBL). The heart of the model is comprised of equations for soil temperature (Ts), mixed-layer poten- tial temperature (u), mixed-layer specific humidity (q), and the height of the PBL (h). In order to look at bound- ary layer growth on days with different early-morning atmospheric conditions, alterations to the original model were required: • the growing BL entrains air from a user-input pre- scribed sounding, rather than from constant lapse-rate profiles; • free convection is triggered when the growing BL reaches the level of free convection: at this point, the model assumptions of a well-mixed, cloud-free boundary layer are no longer valid and the simulation is terminated; • cloud fraction is set to zero; • soil saturation is fixed for the duration of the model runs. The first two changes are fundamental changes in the nature of the model. They allow for a melding of data analysis and model simulations. Confining the analysis to clear skies allows us to focus on the impacts of land surface conditions in the triggering of convection—be it deep, precipitating convection or weak convection producing shallow clouds. The model halts whenever either of these conditions occurs, since after free con- vection the model assumptions, including the no-cloud assumption, are no longer valid. We are considering time scales on the order of 12–15 h, during which the as- sumption of constant soil saturation is reasonable. 556 VOLUME 4J O U R N A L O F H Y D R O M E T E O R O L O G Y use for many years in thunderstorm and weather pre- diction. As Mueller et al. (1993) report and the results of this work confirm, these traditional stability indices are helpful in ruling out the possibility of rain in very stable atmospheric conditions, but when instability is indicated, they give no further clues of where and when—or even if—convection might be triggered. Sim- ilarly, traditional humidity indices are helpful in ruling out days where the atmosphere is too dry for rainfall to develop, but are less helpful in more humid situations. The CTP is a measure of atmospheric stability, ruling out convection in stable conditions, as many traditional stability indices do (e.g., Showalter index, Showalter 1953). However, the CTP is also—and perhaps more importantly—a measure of the influence of surface flux partitioning on the likelihood of convection in unstable situations. In the next section, we present two case stud- ies that demonstrate how the CTP effectively discrim- inates between atmospheric conditions favoring the de- velopment of rainfall over wet soils from those favoring the development of rainfall over dry soils. In subsequent sections we will couple the CTP with HIlow to improve on this ability to discriminate between differing atmospheric conditions. The HIlow is a variation on the humidity index of Lytinska et al. (1976), which was defined as the sum of the dewpoint depressions at 850, 700, and 500 mb: HI 5 (T 2 T ) 1 (T 2 T )850 d,850 700 d,700 1 (T 2 T ), (2)500 d,500 where Tp is the temperature at pressure level p and Td,p is the dewpoint temperature at pressure level p. Though this index was indeed somewhat helpful in distinguish- ing between very dry and very humid atmospheres, the 500-mb information included in this index is generally beyond the reach of typical boundary layer growth, and is therefore not relevant for this work. Other combi- nations of dewpoint depressions at levels below 500 mb all prove to be helpful in assessing the convective po- tential of Illinois soundings. The most effective was the sum of the dewpoint depressions at 950 and 850 mb: HI 5 (T 2 T ) 1 (T 2 T ).low 950 d,950 850 d,850 (3) Defined more generally, HIlow is the sum of the dewpoint depressions 50 and 100 mb above the ground surface. This is the definition that will be used throughout this work. 4. Case studies highlighting the relevance of the CTP Convection is triggered in the model when the level of free convection and the boundary layer top meet. In simplified terms, this can occur when the LFC remains constant and the BL grows up to the LFC, or when the BL height remains constant and the LFC drops to the top of the BL. Obviously many combinations of BL growth and LFC descent can also bring these two levels together. The extremes, however, describe the charac- teristic manner in which convection is triggered over very dry and very wet soils, respectively. We will now present two case studies highlighting these different methods for triggering convection. Figure 2 shows two initial 0600 LT soundings with very different CTP values. These soundings are indic- ative of the types of initial atmospheric conditions that lead to rain over wet but not over dry soils (Fig. 2a: 3 July 1999, CTP 5 88 J kg21), and those that lead to rain over dry but not over wet soils (Fig. 2b: 23 July 1999, CTP 5 254 J kg21). The boundary layer height, the level of free convection, and the moist static energy (as described by uE) values for the wet and dry soil model runs for the first day (3 July 1999) are shown in Fig. 3. Model profiles at 1300 LT are shown in Fig. 4. Similar plots for the second day (23 July 1999) are shown in Figs. 5 and 6. In many model runs, the boundary layer height over wet soils grows slowly but steadily until 1200 or 1400 LT, and then remains relatively constant (Fig. 5a). The uE continues to grow due to the continued input of mois- ture from the land surface (Fig. 5c). Over dry soils, on the other hand, the behavior of these two variables is often reversed (Figs. 3b,d): the BL height grows steadily and more rapidly throughout the day, but the uE plateaus or even drops in the afternoon, primarily because of increased entrainment of dry air from above the BL and limited moisture flux contributions from the land sur- face, which are then spread out over a deep BL. In the dry soil case, the BL top and the LFC will meet only if the BL grows high enough to reach the LFC. The critical factors influencing the BL growth are the sen- sible heat flux (determined by the land surface soil mois- ture and/or vegetation) and the temperature lapse rate of the air being entrained. In the wet soil case, the BL top and the LFC will meet only if the uE grows large enough to bring the LFC down to the BL top. The crit- ical factor influencing the fall of the LFC are the BL uE and the temperature lapse rate of the air through which the LFC drops. The BL uE is determined by the latent heat flux (determined by the land surface soil moisture and/or vegetation) and the initial low-level hu- midity (assesed by a humidity index such as HIlow). Note that the temperature lapse rate in the critical CTP region is a central factor in both styles of convective triggering. Consider, for example, the case of 3 July 1999 (initial sounding Fig. 2a). The uE in the BL over dry soils peaks just after 1000 LT, and levels off at 1300 LT (Fig. 3d). The BL height, however (Fig. 3b), continues to increase until almost 1600 LT. For BL deepening to trigger con- vection with no accompanying increase in uE, the BL must grow from 890 to 685 mb (the point where the parcel path crosses the environmental temperature line in Fig. 4b). In contrast, the wet soil boundary layer grows more gradually than that over the dry soil, but the uE is also increasing. The pseudoadiabats in Fig. 4a JUNE 2003 557F I N D E L L A N D E L T A H I R FIG. 2. Profile of initial conditions for (a) 3 Jul 1999: CTP 5 87 J kg21, HIlow 5 10.68C, rainfall occurs only over wet soils; and (b) 23 Jul 1999: CTP 5 254 J kg21, HIlow 5 11.68C, rainfall occurs only over dry soils. Thick dashed- dotted line is profile temperature; thick dashed line is profile dewpoint temperature; light solid lines are dry adiabats; long-dashed lines are constant temperature; dotted lines are constant mixing ratio. FIG. 3. The boundary layer height and the level of free convection in the (a) wet soil and (b) dry soil model runs for 3 Jul 1999; uE for these same model runs: (c) wet soil case, and (d) dry soil case. 558 VOLUME 4J O U R N A L O F H Y D R O M E T E O R O L O G Y FIG. 4. Profile of model conditions at 1300 LT in (a) the wet soil run and (b) the dry soil run on 3 Jul 1999. Lines as in Fig. 2. Additional line shows the path of a surface parcel: it follows a dry adiabat until reaching its lifting condensation level, then it follows a moist adiabat until reaching the level of neutral buoyancy (off plot). FIG. 5. As in Fig. 3 but for 23 Jul 1999. JUNE 2003 561F I N D E L L A N D E L T A H I R FIG. 9. Values of the CTP and HIlow for days when outcomes of dry soil and wet soil model runs are the same. Filled circles indicates rain over both soil states; triangles: shallow clouds over both; x: no convection over either. at the Lincoln, Illinois, station when both extremely wet and extremely dry soils result in deep convection in the model (Fig. 9). On these days, the early-morning at- mosphere was very humid (HIlow # 10.58C in all 25 cases) and contained no inversions to block deep con- vection (CTP . 0 J kg21). Despite the atmospherically controlled label applied to these 25 cases, Fig. 10 shows that the properties of the boundary layer at the time of convective triggering were significantly different over soils of different moisture content. The anticipated result of higher soil moisture leading to higher boundary layer uE (Betts et al. 1996; Eltahir 1998) is indeed noticable, with a 5.48C difference being significant at the a 5 0.0375 level. Accompanying these higher uE values come larger CAPEs (an 850 J kg21 difference), deeper convection depths (a 1.18-km difference), and smaller dewpoint depressions (a 4.28C difference), all signifi- cant at the a 5 0.0015 level. Each of these differences in the mean properties is a direct result of the higher evaporative fraction (lower Bowen ratio) over wet soils leading to lower boundary layer temperatures, higher specific humidities, lower boundary layer heights, and less entrainment. The differences between the mean trig- gering times and the mean precipitable water in the entire column are not statistically significant. From these results, we conclude that even when the occurrence of rainfall is atmospherically controlled, the land surface moisture condition can indeed impact the depth of rain. This is supported by the studies of Wil- liams and Renno (1993) and Eltahir and Pal (1996). Williams and Renno (1993) demonstrated that CAPE tends to be linear and close to zero below some threshold wet-bulb potential temperature value (uw), while above this threshold there is a ;1000 (J kg21) 8C21 slope of increasing CAPE with increasing uw. Eltahir and Pal (1996) also found this threshold behavior, and further showed that above this threshold, CAPE is linearly cor- related with rainfall depth in the Amazon. This suggests a positive feedback mechanism between soil moisture and the depth of rainfall in Illinois. This result is con- sistent with the work of Findell and Eltahir (1997), who showed that late spring/early summer large-scale mois- ture conditions are positively correlated with the total rainfall depth over the course of the summer in Illinois. b. Soil moisture affects outcome Figure 11 divides the 63 cases when different soil moisture conditions led to different model results into wet soil advantage and dry soil advantage days. Rain occurs over wet soils but not over dry 41% of the time, while the reverse occurs only 8% of the time. Similarly, shallow clouds occur over wet soils but not dry 40% of the time, but only 11% of the time are there shallow clouds over dry but not over wet soils. Figure 12 shows that these data are fairly well stratified in CTP-HIlow space. The wet soil advantage cases with rain (filled circles) all have 0 # CTP # 200 J kg21 and 18 of the 24 have 5 # HIlow # 12 K. All but one of the cases with rain over dry soils but not over wet (filled triangles) have CTP $ 200 J kg21 and HIlow $ 11 K. The case studies presented in section 2 describe behavior typical of the wet soil and dry soil advantage days. c. Summary of Illinois results Figure 13 is a composite sketch of the information provided in Figs. 9 and 12, separated into responses in the model runs with wet soils and those with dry soils. This figure summarizes the predictive capability gained from use of the CTP and HIlow as measures of the early- morning atmospheric setting, according to this 1D mod- eling study using data from Illinois. As shown in this figure, in very dry or very humid atmospheres, the mod- el outcome is determined by the atmosphere alone: • HIlow . 158C: 562 VOLUME 4J O U R N A L O F H Y D R O M E T E O R O L O G Y FIG. 10. Average 6 one standard deviation of (a) free convection triggering time, (b) dewpoint depression, (c) precipitable water, (d) CAPE, (e) uE, and (f ) depth of convection, for the 25 instances when deep convection (likely to rain) is triggered in both the saturated and the dry soil runs. Differences between the wet soil and dry soil values of the dewpoint depression, CAPE, uE, and depth of convection are highly statistically significant (see text). - Any CTP: no convection will result over any soil condition. • HIlow , 58C: - CTP . 0 J kg21: rain will occur over any soil con- dition; - CTP , 0 J kg21: shallow clouds will result over any soil condition; At intermediate humidity levels, the land surface moisture condition can significantly impact the likeli- hood of rain, and the CTP can help to determine what that impact will be: • 58C , HIlow , 108C: - CTP , 0 J kg21: Shallow clouds over wet soils. No convection over dry soils. - CTP . 0 J kg21: Wet soils favored. Rain over wet soils, rain likely (but not certain) over dry soils. • 108C , HIlow , 158C: - CTP , 50 J kg21: Shallow clouds likely (but not certain) over wet soils. No convection over dry soils. - 50 J kg21 , CTP , 200 J kg21: Transition zone: Any outcome possible. Convection of either kind is more likely over dry than wet soils, but no convec- tion is the most likely result over either. - CTP . 200 J kg21: Dry soils favored. No convection over wet soils, rain or shallow clouds possible over dry. 6. One-dimensional BL results from other stations In order to determine if the CTP-HIlow approach used to classify atmospheric conditions yielding a wet soil or a dry soil convective advantage was valid outside the region of Illinois, the methodology of section 5 was applied to four additional stations from other parts of the United States. For each of these four additional sta- tions, [Wilmington, OH (station ILN, latitude 39.48N), Shreveport, LA (station SHV, latitude 32.58N), Charles- ton, SC (station CHS, latitude 32.98N), and Albuquer- que, NM (station ABQ, latitude 35.08N)], 1D model runs were performed for each day from the summer of 1998 with radiative conditions determined for the actual latitude of the station on 29 July. At the four stations, 70, 75, 73, and 86 days, respectively, were used for model initialization. Composite plots of the results from these four stations are given in Fig. 14. These results show consistency with the CTP-HIlow framework de- veloped from Illinois soundings, suggesting that the framework is applicable in a wide range of atmospheric and geographic settings. Similarly, sensitivity tests with varing radiative conditions at the Illinois station (not shown) showed expected changes with more (less) deep convection with increased (decreased) solar radiation, but a continued agreement with the wet and dry soil advantages described by the CTP-HIlow framework. The results of Fig. 14 also add information to a portion of JUNE 2003 563F I N D E L L A N D E L T A H I R FIG. 11. Division of the outcome combinations for the 63 days in which the model resulted in different outcomes over wet and dry soils. CTP-HIlow space that was not covered by data from Il- linois. In the HIlow range of 58–108C, there were no Illinois soundings with CTP . 225 J kg21. Data from these additional stations show that the wet soil advan- tage regime can be extended to CTPs of up to 300 J kg21 in this HIlow range (Fig. 14). This emphasizes the point that both the temperature structure and the hu- midity of the low-level air are critical factors determin- ing the nature of interactions between the land surface and the boundary layer. Experiments with various modifications on the pres- sure levels included in the definition of HIlow indicated that most low-level humidity measures could be used to create a robust CTP-HIlow framework. The value HIlow as defined in this work (the sum of the dewpoint de- pressions 50 and 150 mb above the ground surface, usually near 950 and 850 mb) was the best performer, but all had a few outliers caused by sharp humidity drops just below one of the levels included in the definition. This happens in two cases in Fig. 14. Both show rain over wet soils only with intermediate CTPs and very high HIlows. Both are cases from station ILN with very humid near-surface layers and a sharp humidity drop below 850 mb (one of the two levels included in the HIlow value), but above the level at which convection is triggered. The more extreme of the two cases, with an HIlow of about 248C, has a specific humidity drop of 6 g kg21 between 860 and 840 mb (8–2 g kg21) while convection was triggered in the model at about 910 mb. Thus, the value at 850 mb is not representative of the conditions in the mixed layer at the time of triggering. Future work will include the development of an index that describes the humidity throughout the lower tro- posphere, rather than at a few distinct levels. This should not be as sensitive to extreme humidity changes. Table 1 separates the data points of Fig. 14 into the outcomes at the four individual stations analyzed. With- in the framework of this model, the numbers suggest that during the summer of 1998 there was the potential for a positive feedback between soil moisture and deep convection at station ILN (Wilmington, OH), negative feedbacks at stations CHS (Charleston, SC) and ABQ (Albuquerque, NM), and a neutral response at station SHV (Shreveport, LA). The high frequency of modeled rainfall events and the likelihood of deeper rainfall depths when convection is triggered over wet soils as opposed to dry soils is expected to dampen the negative feedback signals. This dampening effect should be stronger at station CHS than at station ABQ because of the high percentage of days with rainfall expected over both soil types at CHS (32.9%; only 11.6% at ABQ). These feedback signals are discussed in detail in Findell and Eltahir (2003, hereafter Part II), with particular em- phasis placed on the negative feedback signals seen in some years in the southwest (the region influenced by the North American monsoon system and the dryline region of the Texas and Oklahoma panhandles). Figure 14 shows that, as predicted by the model, the CTP-HIlow framework is valid for a wide range of lo- cations and atmospheric settings. It suggests that the CTP and HIlow values marking the transition from wet soil to dry soil advantage regimes are independent of location, although the range of circumstances must be further expanded to fully cover CTP-HIlow space. More significantly, it suggests that for matters of convective triggering and response to land surface conditions, the degree of departure from moist adiabatic conditions be- tween approximately 1 and 3 km AGS is important in all the locations studied. These results, coupled with those from Illinois summarized by Fig. 13, were used to generate the full CTP-HIlow framework for analyzing soil moisture–rainfall feedbacks presented in Fig. 15. 7. Discussion The case studies presented in section 4 highlight the significance of the convective triggering potential within the context of a one-dimensional boundary layer model. In Findell and Eltahir (2003a) we apply the framework developed here to three-dimensional simulations with the fifth-generation Pennsylvania State University–Na- tional Center for Atmospheric Research Mesoscale Model (MM5; Grell et al. 1995) and to an analysis of observations of soil moisture, rainfall, and BL properties from the FIFE experiment in Kansas (Sellers et al. 1992). The work in Findell and Eltahir (2003a) high- lights the importance of the vertical profile of the winds in influencing the triggering of deep convection; it shows that the winds form a crucial third dimension to the CTP-HIlow framework. This study, however, focuses on the results of the one-dimensional BL model and the physical reasons behind the differing behavior seen in 566 VOLUME 4J O U R N A L O F H Y D R O M E T E O R O L O G Y FIG. 14. Composites of the 1D model results from the four additional stations described in the text and in the previous four figures. The wet soil advantage region is approximately bound by the blue ellipse, and the dry soil advantage region by the red ellipse. The results from these four stations are consistent with the framework developed with data from Illinois, and add additional information in areas unpopulated by data from Illinois. Symbols for the left plot as is Fig. 9; those for the right plot as in Fig. 12. TABLE 1. Results of 1D model runs from four additional stations (occurences in %). The numbers suggest that during the summer of 1998 there was a positive feedback between soil moisture and rainfall at Station ILN (Wilmington, OH), negative feedbacks at stations CHS (Charleston, SC), and ABQ (Albuquerque, NM), and a neutral response at station SHV (Shreveport, LA). Station Atmospherically controlled cases (%) Both rain Both SC Neither convective Nonatmospherically controlled cases (%) Wet Ad: Rain Wet Ad: SC Dry Ad: Rain Dry Ad: SC ILN SHV CHS ABQ 23.2 16.0 32.9 11.6 8.7 2.7 1.4 1.2 40.6 68.0 43.8 72.1 13.0 2.7 2.7 1.2 13.0 4.0 1.4 0.0 1.4 4.0 12.3 8.1 0.0 2.7 5.5 5.8 SC 5 shallow clouds; Ad 5 advantage. JUNE 2003 567F I N D E L L A N D E L T A H I R FIG. 15. The CTP-HIlow framework for describing atmospheric con- trols on soil moisture–rainfall feedbacks. Only when the early-morn- ing atmosphere has CTP . 0 J kg21 and 5 , HIlow , 158C can flux partioning at the surface influence the triggering of convection. a high CTP case where a negative feedback is expected between soil moisture and rainfall, and a negative CTP case where wet soils are more likely to lead to shallow clouds, as long as the low-level humidity deficit is not too large. Betts and Ball (1995) found similar evidence for pos- itive soil moisture–rainfall feedbacks in data from the FIFE site in Kansas. They found that increased soil moisture led to an increased diurnal uE range, and was accompanied by a decrease in the peak depth to the LCL from ;230 mb over dry soils to ;130 mb over wet soils. Note that the 100 mb between these two LCL depths is captured by the critical CTP region. This dif- ference in uE behavior over soils of different moisture content is important, Betts and Ball determine, ‘‘If soils are moist enough over large enough horizontal scales, then the associated higher equilibrium uE and the lower cloud-base can be expected to organize mesoscale con- vective systems, just as warmer sea surface temperatures do over the ocean,’’ (Betts and Ball 1995, p. 25 692). With a 1D PBL box model, de Ridder (1997) cal- culated the dependence of uE on the evaporative frac- tion, a, and determined that the potential for moist con- vection increases with a, except in very dry atmo- spheres. This is consistent with the results here regard- ing the lack of convection above a threshold humidity decificit, independent of land surface conditions. Haiden (1997), however, found that ‘‘static stability and tem- perature determine the sign of the Bowen ratio effect, with atmospheric humidity merely affecting its mag- nitude.’’ Our results indicate that within a particular range of humidity, Haiden’s assessment holds, but when the humidity deficit is sufficiently large or sufficiently small, the stability and temperature characteristics do not determine the sign of the Bowen ratio effect. In fact, in these circumstances, the likelihood for convection is independent of the land surface fluxes. It is relevant to note the relationship between these results and the work that originally inspired this inves- tigation of atmospheric controls on soil moisture–rain- fall interactions. Findell and Eltahir (1997) found a small but significant positive feedback between soil moisture and rainfall in Illinois. Expanding on this work, Findell and Eltahir (1999) used near-surface atmospher- ic data and found a significant correlation between soil moisture and wet-bulb depression, Tdpr, and then be- tween Tdpr and subsequent rainfall. They did not, how- ever, find a significant correlation between soil moisture and wet-bulb temperature, Tw, or between Tw and sub- sequent rainfall. The current results seem to be consistent with these findings, though further work with observations is nec- essary. The value HIlow should be closely correlated with Tdpr, since it considers the dewpoint depression at rel- atively low levels. Given the importance of HIlow in the current results, it is not surprising that the surface wet- bulb depression is also a helpful indicator of the link between the land and the atmosphere. The wet-bulb tem- perature, on the other hand, is a measure of the surface energy, much like uE. The current work shows that the surface energy alone is not enough to determine either the potential for rainfall or the impact of the surface moisture on this potential. The CTP is helpful in both of these determinations because it considers the tem- perature profile well above the surface, and because it focuses on the portion of the atmosphere that is between the region that is almost always incorporated into the growing boundary layer and the portion of the free at- mosphere that is almost never incorporated into the growing BL. 8. Conclusions and future work A one-dimensional model of the planetary boundary layer (BL) and surface energy budget has been modified to allow the growing BL to entrain air from an observed atmospheric sounding, rather than from profiles pro- duced by idealized potential temperature and humidity lapse rates. The model is used to analyze the impact of soil saturation on BL development and the triggering of convection in different atmospheric settings. Using ear- ly-morning atmospheric soundings from Illinois to ini- tialize the model, a small positive feedback was seen between soil moisture and rainfall from three summers worth of data from central Illinois, consistent with the work of Findell and Eltahir (1997, 1999). The newly developed convective triggering potential (CTP) is a measure of the early-morning atmospheric thermodynamic structure in the region between 100 and 300 mb (approximately 1 and 3 km) above the ground surface (AGS). The great influence of this region results from its location between the lowest ;1 km, which is almost always incorporated into the boundary layer, and the free atmospheric air above ;3 km, which is almost never incorporated into the BL. 568 VOLUME 4J O U R N A L O F H Y D R O M E T E O R O L O G Y The CTP is coupled with a low-level humidity index, HIlow, to help distinguish between different types of ear- ly-morning soundings based on model response to these differing initial states. The sounding classes are those favoring rainfall over dry soils, those favoring rainfall over wet soils, and those whose convective potential is unaffected by surface fluxes. Together, these two mea- sures form the CTP-HIlow framework for analyzing at- mospheric controls on soil moisture–boundary layer in- teractions (Fig. 15). This framework was initially de- veloped initializing the model with data from Illinois, but additional testing using soundings from Ohio, Lou- isiana, South Carolina, and New Mexico suggest that it is valid for locations far removed from Illinois. This work demonstrates that the early-morning tem- perature and humidity structure must be considered in order to determine how the growing boundary layer will respond to fluxes from the land surface. It shows that within the 1D model, the land surface moisture or veg- etative condition can influence the potential for rainfall only in a limited range of early-morning atmospheric conditions. When the atmosphere is very dry (HIlow . 158C) or very stable (CTP , 0 J kg21), rainfall cannot occur, independent of flux partitioning at the surface. When the atmosphere is humid and unstable (HIlow , 58C and CTP . 0 J kg21), then rainfall should occur over both wet and dry soils, with deeper rainfall depths expected over wet soils. In the remaining circumstances (HIlow between 58 and 158C, and CTP . 0 J kg21), then the land surface can significantly influence the likeli- hood of rainfall, with dry soils more likely to trigger rainfall in the high CTP–high HIlow section of this range, and wet soils more likely in the low CTP–low HIlow section. The power of this framework lies in the ability to determine from a simple analysis of early-morning soundings whether a geographical region is likely to see climate-scale feedbacks between soil moisture and rain- fall, and what the nature of those feedbacks are likely to be. In Part II, the framework is used with data from all of the contiguous 48 United States to locate regions of potential positive and negative feedbacks between the land surface and rainfall, and regions where the land surface conditions cannot play a large role in triggering convection. Research into the CTP-HIlow characteristics of other parts of the world is currently underway. A third paper (Findell and Eltahir 2003a) highlights the effects of the vertical profile of the winds through experiments with the three-dimensional, fifth-generation Mesoscale Model (MM5; Grell et al. 1995) and an anal- ysis of data from the FIFE experiment in Kansas (Sellers et al. 1992) in the context of the CTP-HIlow framework. These analyses show that the winds form a crucial third dimension to the CTP-HIlow framework. The identifi- cation of potentially coherent feedback regions (e.g., Part II) will help determine where to locate future ob- servational missions to test the validity of the CTP-HIlow framework beyond a 1D model and to improve under- standing of soil moisture–boundary layer interactions. Given the rarity of study sites with a statistically sig- nificant number of days with soil moisture observations, precipitation observations, and early-morning radio- sondes, it is useful to gleen as much from models and currently available data as is possible. More work with available observational datasets is underway and will continue in the future. The work presented here has strong implications re- garding the importance of high-resolution data and mod- el levels throughout the critical CTP region: it is critical that the vertical resolution in this region be sufficient to distinguish between a moist adiabatic and a dry adi- abatic temperature lapse rate. As discussed in the intro- duction, previous observational and modeling studies have shown evidence of both positive and negative feed- backs. This could be a result of the individual study locations, since the five stations presented in this paper and the nationwide analysis of Findell and Eltahir (2003b, Part II) reveal highly variable CTP-HIlow char- acteristics throughout the United States. However, this could also result from different model and/or forcing- data resolution in the critical CTP region. Further re- search is needed to determine the vertical resolution required to adequately represent this region and its con- trol on land surface–boundary layer interactions. Acknowledgments. This research has been supported by NASA under Agreement NAG5-7525 and NAG5- 8617. The views, opinions, and/or findings contained in this paper are those of the authors and should not be constructed as an official NASA position, policy or de- cision unless so designated by other documentation. The authors would like to thank Wayne Angevine for gra- ciously sharing his data from the Flatland Boundary Layer Experiment. Thanks to Wayne, Alison Grimsdell, and Tony Delany for transferring both the data and some of their expertise on the data to us. These data were used extensively during early work with the 1D model, though only briefly reported here. Thanks to Chris Milly and three anonymous reviewers for helping to improve the quality and clarity of the manuscript, and thanks to Cathy Raphael for help with the figures. Also, thanks to Bob Hart for making his GrADs script to plot a skew T/logp diagram available to all GrADs users. REFERENCES Angevine, W. M., A. W. Grimsdell, L. M. Hartten, and A. C. Delany, 1998: The Flatland boundary layer experiments. Bull. Amer. Me- teor. Soc., 79, 419–431. Atlas, R., N. Wolfson, and J. Terry, 1993: The effect of SST and soil moisture anomalies on the GLA model simulations of the 1988 U.S. summer drought. J. Climate, 6, 2034–2048. Battan, L. J., 1973: Radar Observations of the Atmosphere. University of Chicago Press, 324 pp. Betts, A. K., and J. Ball, 1995: The FIFE surface diurnal cycle climate. J. Geophys. Res., 100 (D12), 25 679–25 693. ——, ——, A. C. Beljaars, M. J. Miller, and P. A. Viterbo, 1996: The land surface–atmosphere interaction: A review based on