1. Introduction
This paper is the second of a two-part series that uses a combination of numerical simulations and observations to explore the relationship of surface heterogeneity and associated fluxes (W m−2) of sensible heat H and latent heat (LE), to potential temperature Θ (K), mixing ratio Q (g kg−1), depth, and convective structure on scales from 1 to 100 km in the fair-weather convective boundary layer (CBL), while evaluating the numerical simulations. The numerical simulations use the Advanced Research Weather Research and Forecasting (ARW-WRF) model (Skamarock et al. 2005), coupled to the Noah land surface model (LSM), which was initialized using the National Center for Atmospheric Research (NCAR) High-Resolution Land Data Assimilation System (HRLDAS; Chen et al. 2007). The data were collected in southeast Kansas using aircraft, surface flux towers, and three radar wind profilers, during May–June, 2002, as part of the International H2O Project (IHOP_2002; Weckwerth et al. 2004, LeMone et al. 2007). The goal of IHOP_2002 was to improve prediction of convective precipitation in numerical weather prediction (NWP) models by improving the measurement and use of water vapor data, and by improving representation of the evolution of water vapor in numerical weather prediction models. Surface processes were emphasized because of their importance in the initiation and evolution of precipitating convection.
In LeMone et al. (2010, hereafter Part I), we focused on comparisons of observed and simulated surface fluxes, CBL thermodynamics, and CBL depth, and found that
-
The Noah LSM produced an east–west gradient in H and LE consistent with the modeled soil moisture pattern, but not the observed variation associated with land cover, due to shortcomings in the input land-use and land-cover table.
-
While modeled mean LE values and horizontal variation in the CBL depth compared reasonably well with observations, H and CBL depth were too high. This is likely due to too low a value of the coefficient C in Zilitinkevich’s formulation used in relating the roughness lengths for heat and momentum in the Noah LSM, with too-low simulated soil moisture playing a secondary role.
The effects of surface heterogeneity on fair-weather CBL structure have been the subject of numerous modeling studies, some of which incorporate observational data, and a few observational studies. Mesoscale models have been applied to demonstrate the potential importance of CBL mesoscale circulations (∼10–100 km) generated by surface heterogeneity and to understand the scales favored for such circulations (e.g., Chen and Avissar 1994; Baidya Roy and Avissar 2000; Baidya Roy et al. 2003). Reen et al. (2006) and Desai et al. (2006) combined observations and a mesoscale model to look at the effects of heterogeneous surfaces on CBL structure in real situations. Larger and faster computers have enabled numerous large-eddy simulations (LESs) to study the structure and evolution of mesoscale CBL circulations (e.g., Avissar and Schmidt 1998; Letzel and Raasch 2003; Patton et al. 2005; Kang and Davis 2008). However, observational documentation of mesoscale circulations is typically sketchy (e.g., Mahrt et al. 1994; LeMone et al. 2002; Kang et al. 2007; Taylor et al. 2007), with the significant exception of circulations along the dryline that lead to precipitating convection (Ogura and Chen 1977 and many subsequent papers, recognizing that factors other than surface heterogeneity are also important for the dryline). More traditional LESs have examined CBL-depth scale (∼few kilometers, hereafter referred to as kilometer scale) structure for decades (e.g., Deardorff 1972; Moeng and Sullivan 1994), and there has been limited attention to kilometer-scale structure generated in high-resolution mesoscale model simulations (Trier et al. 2004; Miao and Chen 2008).
Our goal to compare CBL structure and evolution at the scales observed requires high-resolution runs. The choice of an LES to simulate real-world situations is appealing, since large CBL eddies are resolved. Indeed Conzemius and Fedorovich (2008) used LES to simulate of CBL development near the western track, and Gorska et al. (2008) used LES to simulate of mesoscale circulations along the eastern track. However, these studies were hindered by the need for periodic boundary conditions. Moreover, CBL structure can be influenced by interaction with tropospheric gravity waves (e.g., Clark et al. 1986; Balaji et al. 1993), requiring inclusion of a deeper domain than ordinarily used in LES. These factors, along with the convenience of having an NWP model with a reasonably robust surface model, led to the choice to run ARW-WRF run at 1-km resolution. We recognize the potential problems created by “double counting” associated with partially resolving CBL-scale motions and using a CBL scheme, but note the developing literature of high-resolution ARW-WRF runs that appear successful, even at very high resolution (e.g., Davis et al. 2008, Miao and Chen 2008). Furthermore, actual comparisons to data such as done here can help us understand the limitations of stretching ARW-WRF beyond its expected capabilities.
The paper is outlined as follows. Data collection and analysis are discussed in section 2, with model runs and analysis of results summarized in section 3. Section 4 compares the observed and modeled CBL structure from kilometer-scale to mesoscale (∼10–100 km), drawing from some sensitivity runs to help interpret the model CBL convective structure. The results are summarized in section 5.
2. Data collection and analysis
The four fair-weather days examined had scattered clouds and wind from approximately south-southwest to southeast (Table 1). Winds at 65 m above ground level (AGL) vary from 3.9 m s−1 on 30 May to 9.4 m s−1 on 22 June. Timing of the most recent rainfall ranged from 1.7 to 6.7 days previous. Figure 1 shows a higher density of crops (mainly winter wheat) to the west and mainly grasslands to the east, relative to the eastern track and for at least 100 km to the south. The band of winter wheat extends south to the Oklahoma–Kansas border; south of the border, the winter wheat band extends southwestward into Texas. The dominant soil type along the flight track is silty clay loam (Soil Survey Geographic Database; more information is available online at http://dbwww.essc.psu.edu/dbtop/doc/statsgo/statsgo_info.html; Fig. 2), consistent with the observed near-surface soil type at the IHOP_2002 surface sites.
As in Part I, aircraft data are from the University of Wyoming King Air, which flew straight and level legs along the ∼46-km long eastern track (Fig. 1). While at least 5 passes were flown between 60 and 70 m AGL, there are data collected from at least 1 other altitude range (Table 1). Aircraft-relative winds were measured by a Rosemount 858AJ/1332 differential pressure gust-probe system. Aircraft position and motion relative to the ground were measured by a Honeywell Laseref SM inertial navigation system and corrected using GPS to within 100 m horizontally. Aircraft altitude was based on a King KRA5 radio altimeter for heights below 610 m; and an APN159 radar altimeter for heights above 610 m. Estimates of virtual potential temperature Θυ are based on water vapor densities from a LiCor 6262 gas analyzer,1 flight-level air temperatures from a reverse-flow platinum-resistance thermometer built at the University of Wyoming, and pressure from a Rosemount 1201 sensor. Data were made available at 25 Hz and as 1-s averages.
Statistics on flight-level wind and thermodynamics were based on the 1-s data. Aircraft-based divergences were computed from the along-track wind component using least squares best-fit lines to each flight track. Values at specific times could then be computed from a least squares best fit line to the divergence time series at each level.
To calculate the Obukhov length L, we use leg-averaged flux measurements from the King Air following Strassberg et al. (2008), and fluxes from the three towers along the eastern track, located on grassland sites at positions 7, 8, and 9 in Fig. 1. These measurements are described in detail in Part I and LeMone et al. (2007).
Argonne National Laboratory’s Atmospheric Boundary Layer Experiments (ABLE; ABLE is no longer operating; however, the data are available at http://gonzalo.er.anl.gov/ABLE; Klazura et al. 2006) facility radar wind profilers were used to obtain CBL depth and convergence. The locations of the two profilers used for boundary layer (BL) depth—Beaumont and Oxford, Kansas—are shown in Fig. 1. The BL height was defined as the center height of the gate just below the maximum signal-to-noise dropoff rate with height (Coulter and Holdridge 1998). For convergence estimates, the ABLE Whitewater profiler was combined with those at Beaumont and Oxford. The Whitewater profiler lies ∼63 km north-northwest of Oxford (∼350° azimuth) so that the three profiles form a nearly equilateral triangle. The vertical resolution of the profiler data is of the order of 60 m. Radar wind profilers can in principle measure vertical wind directly, but in the CBL this measurement tends to be negatively biased (Angevine 1997).
3. ARW-WRF runs
The model runs are described in detail in Part I, so we provide only essential detail here. Our control runs use ARW-WRF version 2.1.2 with the Yonsei University (YSU) PBL scheme (Hong et al. 2006), coupled to the Noah LSM (Chen and Dudhia 2001; Ek et al. 2003). The simulations, for the four fair-weather days in Table 1, start at 1200 UTC and run for 24 h, using the nested domains shown in Fig. 3. The vertical grid contains 30 full-sigma levels from the surface to 50 hPa, of which the lowest 8 levels are below 1 km AGL (lowest level 28 m AGL) in order to have finer resolution in the CBL. The horizontal grid spacing (number of points) of the three outer domains are 9 km (237 × 201), 3 km (280 × 229), and 1 km (391 × 289), respectively. Initial and boundary conditions are from the National Centers for Environmental Prediction (NCEP) 6-hourly Eta Data Assimilation System (EDAS) on a 40-km grid. In the control ARW-WRF simulation, the Noah LSM is initialized by volumetric soil moisture and temperature profiles and skin temperature obtained from HRLDAS (Chen et al. 2007), which is run offline but on the same nested WRF grid for an 18-month spinup period ending at the initialization time of each ARW-WRF simulation (here, runs start 1 January 2001). The input data at the surface are default values except for the land-use characteristics, which are based on Moderate resolution Imaging Spectroradiometer (MODIS) data from the Boston University Department of Geography (see online at http://www-modis.bu.edu/landcover/; Fig. 2).
We did several sensitivity runs to see whether the modeled CBL-scale convective structure was robust. Two runs tested the effects of horizontal resolution using data from 22 June; and several idealized runs were conducted using data from 20 June to assess the impact of grid spacing, surface properties, and filtering (Table 2). Two higher-resolution runs for 22 June were run using the same version of ARW-WRF (ARW-WRF V2.1.2) as the control runs, with a fourth domain with 333-m grid spacing (Fig. 3), with initial surface conditions interpolated from the 1-km HRLDAS results.
The idealized runs used the public-release ARW-WRF version 3 (ARW-WRF V3), with all points in the domain initially assigned surface characteristics and meteorological data for a typical point on the eastern track (37.4188°N, 96.6472°W), and in domains centered at this point (Fig. 3). Initial surface properties in the idealized runs were varied horizontally by randomly perturbing the initial volumetric soil moisture profile and/or Ts, using a uniform distribution with ranges of ±0.02 and ±0.01 K, respectively. We varied grid spacings for both the surface and atmosphere grid together and separately, ran simulations with and without the Knievel et al. (2007) filter to damp unresolved motions (<6Δx), and changed the domain depth for some runs. The grid points in the innermost domain varied from 226 × 226 for the 200-m runs to 101 × 101 for the remaining runs. Because input data were based on only one point, winds had to be determined using an option in ARW-WRF that maintains the geostrophic part with an implicit pressure gradient, with only the ageostrophic component responding to the Coriolis acceleration. Table 2 will be discussed in greater detail in section 4b.
A final sensitivity run was conducted to assess the effect of the Knievel filter on mesoscale structure. Using the same model setup as for the control runs, we ran ARW-WRF V3 for 30 May with and without the Knievel filter.
4. Results
a. Horizontal variability in virtual potential temperature
As noted in Part I, Θ increased westward along the eastern track in both model results and observations. The largest east–west difference was for 30 May, and the smallest for 22 June. These trends are consistent with the east–west differences in Θυ, which extends through the CBL (Table 3). The differences between these two days are also consistent with the modeled soil moisture patterns, which show stronger contrast along the flight track on 30 May than for 22 June (Fig. 4).
To see whether these along-track Θυ changes can result from horizontal heterogeneity alone (i.e., no synoptic-scale differential advection), we consider two idealized trajectories that start out with the same Θυ, one headed northward toward the western end of the flight track over dormant winter wheat and dry soils, and the second headed northward toward the eastern end over green grasses and wet soils, consistent with southerly winds passing over the vegetation and soil moisture depicted in Figs. 1 and 4. For simplicity, we assume that the surface virtual temperature fluxes
Setting f = 0.2 and applying (3) to the four IHOP_2002 days, with S, Δ〈Θυ〉, zi, and
b. Convective-scale structure and clouds
On 3 of the 4 days, the ARW-WRF runs produced roll-type structures aligned with the CBL wind, with structure and orientation similar to that documented in satellite images (Fig. 5). From the figure, the model replicates the observed north–south cloud streets on 22 June, and south-southeast–north-northwest cloud streets that vary longitudinally on 20 June. For 17 June, ARW-WRF produces south-southwest–north-northeast linear roll structures with the same orientation of the cloud streets depicted in the satellite image. The radar composites in Fig. 6 suggest roll alignment corresponding to the cloud streets. The observed and modeled roll wavelength, about 4 km, is consistent with that associated with CBL rolls (∼3zi, e.g., LeMone 1973, land values from Fig. 4 in Young et al. 2002). However, this result could be fortuitous: for these 1-km grid simulations, 4 km is less than the ∼6–10Δx quoted in Skamarock (2004) as the lower limit for properly resolving structure and amplitude.
The ARW-WRF did not produce clear-cut convective patterns on 30 May, with patches of irregular cellular convection and artificial-looking linear structures 2Δx wide near the flight track and profiler triangle (not well resolved in Fig. 5a). The Wichita Weather Surveillance Radar-1988 Doppler (WSR-88D) radar pattern in Fig. 6 also shows a cellular pattern. The 2Δx features probably result from the shallow CBL depth (Table 1), which reduces the expected convection scale (order CBL depth); and a combination of light winds (Table 1) and the numerical diffusion scheme in ARW-WRF version 2.1.2, which is less effective at reducing 2Δx features in light winds (Knievel et al. 2007).
The presence of CBL rolls (as opposed to cellular convection) corresponds to what occurs in nature. In Figs. 5 and 6, the rolls occur for values of −zi/L less than around 25, where L is the Obukhov length, as expected from previous work by Weckwerth et al. (1999) and others. Similarly, ARW-WRF simulations by Miao and Chen (2008) showed rolls for smaller −zi/L ratios and random convection for larger ratios, and Méso-NH (Lafore et al. 1998) simulations by Lothon et al. (2007) showed rolls transitioning to cells as −zi/L increased beyond 25.
For the observations, the “regional”3 −zi/L ratios based on aircraft 46-km averages (Strassberg et al. 2008) seem to have a better correspondence to rolls than those based on the local flux tower data (Table 4). As shown by Strassberg et al. using data from 22 June, these regional values correspond to a momentum roughness length z0 of 0.14 m, close to the value of z0 = 0.12 prescribed for grassland in ARW-WRF. Note that the −zi/L threshold for roll occurrence is significantly exceeded for 20 June (Table 4) when we use u* measured at the surface stations or use model fluxes with the local (0.02 cm) z0 value; thus, additional evidence for the need to use regional values.
The good correspondence between modeled and observed type of convective structure was surprising, given the use of a PBL scheme designed for grid spacings much larger than 1 km (e.g., Wyngaard 2004) and the associated use of grid boxes with smaller width-to-height ratios (Wyngaard et al. 1998). To test the robustness of the convection type and scale in the ARW-WRF simulations, we ran a series of tests to assess the impact of surface perturbations and grid sizes on the characteristics of the CBL convection, summarized in Table 2. As seen from the table, the grid spacings for both the surface and the atmosphere were varied, sometimes separately.4 Volumetric soil moisture (surface temperature Ts) was perturbed randomly using a uniform distribution with a range of ±0.02 (0.01 K) relative to the mean. On some runs, Ts was perturbed and soil moisture left constant. We also varied the depth of the model and horizontal diffusion (not shown). Each run is identified by what is perturbed and at what scale. For example, T1 has Ts at 1-km spacing and a 1-km atmosphere grid, while TS1 has 1-km grids for Ts and soil moisture perturbations and the atmosphere. When the surface and atmosphere grid spacing differ, the suffix “mix” is used. Additional letters indicate the presence or absence of the Knievel et al. (2007) filter and the simulation depth. Thus, TSmix1_0.33NFs has 1-km soil moisture and Ts perturbations and 0.33-km atmosphere grid spacing, no Knievel filter (NF) and a shallow (s) domain.
Roll wavelengths in Table 2 were determined by manually counting vertical-velocity maxima or minima in the x direction (east–west) for direct comparison to grid spacing. However, roll aspect ratios (roll spacing divided by zi) were calculated using the true roll spacing (measured normal to the roll axis). In nested runs, rolls were counted and convective type evaluated far enough downstream of the outer grid to mitigate boundary effects. For example, the convection in Fig. 7b would be considered “cellular.”
The data in Table 2 suggest four general tendencies.
First, roll spacing in the x direction decreased with smaller grid size. For 22 June, the 1-km run (control) had rolls ranging from 2 km (2Δx) to ∼4.5 km across, while the 333-m runs (HiResE and HiResW) typically had 2-km rolls. For the idealized runs, the reduction of roll size in the x-direction with smaller grids was clearly associated with the Knievel filter, which damps out perturbations smaller than 6Δx (6.7 km for TS1 versus 3.5 km for TS0.5). As in the case of 22 June, this trend was also evident without the filter (4.0, 2.4, and 1.8 km for TS1NF, TS0.5NFs, and TSmix1_0.33NFs, respectively). At the smallest atmosphere grid spacing with rolls (TSmix1_0.33NFs), their aspect ratio was 1.4, smaller than the expected land value (∼2–3 or slightly greater). There also appear to be secondary effects related to domain top and the presence or absence of nesting. As expected, deactivating the Knievel filter led to smaller-scale rolls (6.7 km for TS1 versus 4.0 km for TS1NF).
Second, reducing the grid size led to more irregular convection. From the table, the two idealized runs with 200-m atmosphere grid spacing (T1mix1_0.2NFs and TS1mix1_0.2NFs) are dominated by cellular convection, rather than rolls; and the high-resolution 22 June runs had a mix of rolls and more three-dimensional structure in the 333-m grid inner domains. This tendency is illustrated by the change in structure between the 1800 UTC mid-CBL vertical-velocity fields for the idealized runs shown in Figs. 7 and 8, for a 1-km grid (TS1 in Fig. 7a, TS1NF in Fig. 8a) versus a 200-m atmosphere grid (TSmix1_0.2NFs in Fig. 7b). The behavior for TSmix1_0.2NFs was not changed significantly when we removed the Smagorinsky diffusion (not shown in Table 2). Trier et al. (2004) similarly found that convection changed from linear to more cellular when their grid spacing was changed from 3.3 to 1.1 km in their ARW-WRF simulations using the Mellor–Yamada–Janjic (MYJ) PBL scheme.
Third, CBL convective structures with Ts perturbations alone developed later than CBL convection with both Ts and soil moisture perturbations, or failed to develop at all, leaving the YSU PBL scheme to do the vertical transport. This is illustrated by the 1-km simulations with and without soil moisture perturbations in Fig. 8, which at 1800 UTC show well-developed rolls for TS1NF but only infinitesimal resolved motions at for T1NF. The onset time for resolved convection in the temperature-perturbation runs is scale dependent. Of the runs at 1, 2, 4, and 10 km (T1NF, T1, T2, T4, T10), temperature perturbations generated CBL convection only for the 1-km unfiltered run (T1NF), which began to form roll-like convection by 2100 UTC (Fig. 8d). For the 200-m atmosphere grid (T1mix1_0.2NFs), convection formed earlier, but was cellular. The increase in convection growth rate with decreasing horizontal scale is consistent with the results of Weisman et al. (1997).
Fourth, CBL rolls tended to grow with time along with the CBL depth (21-UTC roll size under “remarks” in Table 2), indicating ARW-WRF was able to follow this behavior at all resolutions for which rolls were captured.
c. Mesoscale motions
Given the rarity of direct observations of CBL mesoscale circulations, it is important to consider the conditions under which they can be observed. Once they are set up, mesoscale circulations should evolve slowly if they are to be observed using aircraft or even radar wind profilers. Under what conditions does this occur? A rapid dynamic response like that for 30 May will weaken the 〈Θυ〉 gradient and lead to a weaker or unsteady mesoscale circulation, unless the horizontal change of vertical flux divergence along the wind is sufficient to maintain the gradient against horizontal advection. Consider an air parcel in a two-dimensional circulation moving from point A on the cool side, with surface flux
These arguments about maintenance of a constant 〈Θυ〉 gradient of course apply only under the assumptions that
Kang (2009) suggests that mesoscale circulations become unsteady because of interaction with turbulent eddies when their circulation velocity reaches VC ≈ ut(λC/2λt)0.5, where λC is the wavelength of the circulation (twice our DAB), and ut and λt are the characteristic velocity and scale, respectively, of the turbulent (our kilometer scale) eddies. For typical values of λt ≈ 1.5zi, zi ≈ 1000 m, ut ≈ 1 m s−1, and λC = 100 km, VC = 5.8 m s−1, much greater the 1 m s−1 used to estimate the wavelength for steady state. For the modeled horizontal variability in
d. Observational evidence of mesoscale circulations
Figure 9 shows that estimates of the along-track divergence of the along-track wind component are consistent with the development of along-track mesoscale circulations in the CBL on 30 May (and possibly 17 June), with convergence (divergence) in the lower CBL and divergence (convergence) in the upper CBL. The divergence pattern on 22 June indicates subsidence. The 20 June divergences are less than the standard error.
We do not expect exact correspondence between the divergence-derived mean vertical velocities6 (W) from the ABLE radar wind profiler triangle in Fig. 10 and the aircraft-derived divergences in Fig. 9; since the latter are only for the wind component parallel to the flight track, and the two measurements are displaced horizontally (Fig. 1). However, the two sets of measurements together reveal some interesting patterns. Figures 9 and 10 both indicate strong subsidence on 22 June. There is evidence for circulations largely confined to the CBL on 30 May, with convergence in the lower CBL and divergence in the upper CBL in Fig. 9 and W extrema within the CBL at 1600 UTC and possibly 2000 UTC in Fig. 10, but not at other times. On 20 June at 1600 UTC, there is subsidence that reaches a maximum at around 1300 m AGL (almost 2zi), suggesting a circulation that extends above the CBL, but aircraft-based convergence is smaller than the expected error and available only below 450 m AGL. On 17 June, the W extremum at 2000 UTC is within the CBL in Fig. 10 and corresponds roughly to the top of the divergence layer at 1930 UTC in Fig. 9; but there are W extrema above the CBL at 1800 and 1900 UTC. We interpret these figures as suggestive of mesoscale circulations on 30 May and possibly 17 June, possible mesoscale features centered above the CBL on 20 June, and no detectable mesoscale circulations in the CBL on 22 June.
To see whether the mesoscale features are replicated in ARW-WRF, we produced horizontal fields in domain d03 of temperature at ARW-WRF level 3 (about 300 m AGL) and the vertical velocity at ARW-WRF level 5 (about 500 m AGL) and applied a bidirectional 121-point box filter (11 km in each direction). The resulting plots appear in Figs. 11a and 11b. At scales of tens of kilometers, we see circulations based in the CBL on 30 May and circulations with maximum amplitude above the CBL on 20 June. Mesoscale structures on 22 June are overwhelmed by the rolls. On 17 June, there are some weak warm upwelling regions and cool downwelling regions in the CBL; but little mesoscale circulation evident in the cross section in Fig. 11a, with maxima and minima above the CBL at 1800 UTC, consistent with the ambiguous observations in Figs. 9 and 10. There is a large response to the Wichita urban heat island (Fig. 12) on 30 May (distance = 20 km in cross section, Fig. 11a), but only a small response on the remaining three days (Figs. 11a,b).
The circulations on 30 May appear to be examples of the classic “inland sea breeze,” as studied by Ogura and Chen (1977), Sun and Ogura (1979), Anthes (1984), Segal et al. (1988), Pielke et al. (1991), and many others, with upwelling (downwelling) areas correspond to warmer (cooler) air in the CBL (Fig. 11a). Indeed, Gorska et al. (2008) simulated mesoscale circulations using an LES with the lower boundary conditions idealized from HRLDAS output for 30 May for the eastern track region. The general subsidence over the profiler triangle corresponds roughly to observations (Fig. 10), but the pattern along the flight track is less clear. Exact correspondence is not expected given differences in the observed and modeled fluxes, discussed in Part I. From Fig. 8 in Part I, the modeled PBL top is around 850–830 hPa; thus, the 30 May cross section in Fig. 11a indicates maximum vertical velocities within the CBL, with the largest updraft–downdraft couplet associated with the Wichita urban heat island and its flux anomaly (Fig. 12). All of these features were preserved when this day was rerun using ARW-WRF version 3.0 with the Knievel filter. From the previous discussion, the upstream flux contrast results in strong temperature gradients (Fig. 11a; Table 3), which favor mesoscale circulations. Moreover, the temperature gradient, scale, and flux differences are roughly consistent with near steady state, a condition required for being well observed in a ∼4-h aircraft mission.
While there are localized mesoscale circulations apparently driven by CBL temperature contrast (Fig. 11b, top), there are well-defined linear structures normal to the convective rolls across the region on 20 June (Figs. 5b and 11b) that appear unrelated to the CBL temperature. The vertical-velocity amplitudes in the 20 June cross section in Fig. 11b reach a maximum above the CBL, at about 700 mb. We speculate that the 700-mb extrema are associated with the northeast–southwest linear structures, and that both are due to gravity waves. As obvious from the figures, ARW-WRF replicates both the wavelength and the orientation. However, there is a mismatch in phase speed (14 m s−1 for ARW-WRF, nearly stationary in the observations), so this is speculation at this point.
It is interesting that even on 30 May, when mesoscale circulations were favored; they were not clearly related to the Walnut River Watershed, as was the case in LeMone et al. (2002). We believe that the explanation has to do with the association of vegetation type with elevation in the watershed. The circulation described in LeMone et al. (2002) was in early May, so that the warmth associated with the higher elevations (Walko et al. 1992; Krettenauer and Schumann 1992) was reinforced by higher surface air temperatures associated with dormant grasses there, with lower surface air temperatures above the rapidly growing winter wheat in the lower elevations. However, in the present summer case, the elevated terrain has its surface air temperature cooled due to evapotranspiration from the green grasses, while the lowlands have higher surface air temperatures due to dormant or harvested winter wheat.
5. Conclusions
Comparisons of satellite images and aircraft and surface observations centered along the IHOP_2002 eastern track to ARW-WRF–Noah modeling results on four fair-weather days with southerly winds and nearly clear skies, show that the model captures both the type of convective-scale structure (a few kilometers across) and influence of surface heterogeneity on CBL mesoscale structure (tens of kilometers across) reasonably well. This is in spite of the model deficiencies discussed in Part I, namely significant overestimates of H and CBL depth and the model’s inability to capture variability associated with vegetation because of the model input land-use land-cover table not including enough winter wheat pixels along and upwind the flight track.
While surprising given the probable mismatch of the PBL parameterization to the 1-km grid spacing, modeled two-dimensional CBL structures corresponded well to the cloud streets evident in the 17, 20, and 22 June satellite images, in terms of orientation, spacing, and on 2 out of 3 days, timing. Moreover, evidence from radar as well as satellite images shows that observed as well as modeled rolls occurred for the range of −zi/L < 25 associated with roll occurrence by Weckwerth et al. (1999) and others. For both model and observations, the L values corresponded better to “regional” values of z0 (≈0.1 m), which implicitly account for larger roughness elements (e.g., trees and houses) as well as the vegetation (crops or grassland) specified. However, the “good” roll spacing and aspect ratios were fortuitous: sensitivity studies showed both were influenced by the grid spacing; and with grid spacing less than about 300 m, the rolls broke down into cells. While realistic model convective type is surprising, the change in model behavior at smaller grid sizes is not, given the combined effects of poor horizontal resolution, the potential for double counting created by resolved CBL motions in a PBL scheme designed for larger grid sizes, and the reduction in the aspect ratio of the grid boxes (Wyngaard et al. 1998; Wyngaard 2004).
On scales ≥∼30 km, ARW-WRF again replicates the type of mesoscale structure expected from observations. Mesoscale circulations associated with CBL temperature differences are modeled on 30 May, the day for which they were most likely, given the large along-track Θυ difference and light winds. A sensitivity run using ARW-WRF 3.0 and the Knievel filter produced nearly identical mesoscale structure. On this day, both along-track divergence of the along-track aircraft wind and some radar wind profiler vertical velocity estimates are consistent with CBL circulations, with extrema within the CBL. Idealized trajectory analysis based on model results suggested there was enough horizontal contrast on a large enough scale to create the observed along-track Θυ difference. Idealized analysis of the circulation and effects of horizontal advection suggested that mesoscale circulations could develop fairly rapidly on this day, and be sufficiently steady state to be observable using aircraft, for wavelengths of the order of 100 km.
There is mesoscale variability in both the observations and the model on 20 June, with patches of surface-driven circulations, but, more interestingly, intensity changes in the north-northwest–south-southeast cloud streets marked by northeast–southwest bands of nearly-clear skies and enhanced cloudiness. This modulation is not associated with temperature changes in the CBL and the amplitude of the vertical-velocity variation peaks above the CBL, suggesting gravity waves. A mismatch between observed and modeled phase speeds indicates more work is needed to confirm this hypothesis. No mesoscale CBL variation is evident either in the satellite image or in the ARW-WRF simulations for 22 June, the day with uniform observed subsidence, the least horizontal contrast, and the strongest winds. None of the mesoscale circulations were clearly linked to the topography sampled by the aircraft and profilers, as in the case described in LeMone et al. (2002) for the same region, perhaps because the thermal effects of land cover and elevation were in opposite directions during IHOP_2002, while they reinforced one another in the former case.
In this paper, we were careful to compare the type rather than the details of CBL and lower-tropospheric structure on scales from kilometers to tens of kilometers, because either the model or the data were not adequate for a closer comparison. The major model shortcomings—the inability to represent the impact of the dormant/harvested winter wheat in the mixed grass-winter wheat regions, and the overestimate of surface sensible heat flux—can be dealt with in a straightforward way. Indeed, a new formulation for C in the Noah LSM based on Chen and Zhang (2009) should improve modeled surface fluxes. The impact on this change on ARW-WRF is currently being tested (S. Trier 2009, personal communication). To deal with observational shortcomings, we will simulate the well-documented mesoscale circulations observed during the Cooperative Atmosphere-Surface Exchange Study-97 (CASES-97) and documented in LeMone et al. (2002). Like the present work, however, the future work will require finer grid spacing than optimum for present ARW-WRF PBL schemes.
Acknowledgments
This work would not have been possible without the commitment, enthusiasm, and expertise of the Wyoming King Air aircrew, the staff maintaining the Wyoming Cloud Radar, the National Center for Atmospheric Research (NCAR) staff maintaining the surface flux instruments, and the students taking the manual observations at the flux sites. We gratefully acknowledge the four reviewers whose thorough reviews improved the manuscript and Stan Trier of NCAR for supplying filtering software for the ARW-WRF data, and for reviewing the manuscript, as well as discussions with George Bryan and Chin-Hoh Moeng in designing and interpreting our sensitivity studies. The NCAR portion of this research was supported by USWRP Grant NSF 01, the NCAR Water Cycle Initiative, and NASA-THP (Grant NNG06GH17G) and NOAA/JCSDA (Grant NA06NES4400013), as well as NCAR base funding. RLG’s participation in IHOP_2002 was supported by NSF Grant ATM-0296159. RC’s work was supported by the U.S. Department of Energy, Office of Biological and Environmental Research, under Contract W-31-109-Eng-38.
REFERENCES
Angevine, W. M. , 1997: Errors in mean vertical velocities measured by boundary layer wind profilers. J. Atmos. Oceanic Technol., 14 , 565–569.
Anthes, R. A. , 1984: Enhancement of convective precipitation by mesoscale variations in vegetative covering in semiarid regions. J. Climate Appl. Meteor., 23 , 541–554.
Avissar, R. , and T. Schmidt , 1998: An evaluation of the scale at which ground surface heat flux patchiness affects the convective boundary layer using large-eddy simulations. J. Atmos. Sci., 55 , 2666–2689.
Baidya Roy, S. , and R. Avissar , 2000: Scales of response of the convective boundary layer to land-surface heterogeneity. Geophys. Res. Lett., 27 , 533–536.
Baidya Roy, S. , C. P. Weaver , D. S. Nolan , and R. Avissar , 2003: A preferred scale for landscape forced mesoscale circulations? J. Geophys. Res., 108 , 8854. doi:10.1029/2002JD003097.
Balaji, V. , J-L. Redelsperger , and G. P. Klaasen , 1993: Mechanisms for the mesoscale organization of tropical cloud clusters in GATE Phase III. Part I: Shallow cloud bands. J. Atmos. Sci., 50 , 3571–3589.
Chen, F. , and R. Avissar , 1994: The impact of land-surface wetness heterogeneity on mesoscale heat fluxes. J. Appl. Meteor., 33 , 1323–1340.
Chen, F. , and J. Dudhia , 2001: Coupling an advanced land surface hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129 , 569–585.
Chen, F. , and Y. Zhang , 2009: On the coupling strength between the land surface and the atmosphere. Geophys. Res. Lett., 36 , L10404. doi:10.1029/2009GL037980.
Chen, F. , and Coauthors , 2007: Description and evaluation of the characteristics of the NCAR high-resolution land data assimilation system. J. Appl. Meteor. Climatol., 46 , 694–713.
Clark, T. L. , T. Hauf , and J. P. Kuettner , 1986: Convectively forced internal gravity waves: Results from two-dimensional experiments. Quart. J. Roy. Meteor. Soc., 112 , 899–926.
Conzemius, R. J. , and E. Fedorovich , 2006: Dynamics of sheared convective boundary layer entrainment. Part I: Methodological background and large-eddy simulations. J. Atmos. Sci., 63 , 1151–1178.
Conzemius, R. J. , and E. Fedorovich , 2008: A case study of convective boundary layer development during IHOP_2002: Numerical simulations compared to observations. Mon. Wea. Rev., 136 , 2305–2320.
Coulter, R. L. , and D. Holdridge , 1998: A procedure for the automatic estimation of mixed layer height. Proc. Eighth Atmospheric Radiation Measurement (ARM) Program Science Team Meeting, Tucson, AZ, Dept. of Energy Office of Energy Research, 177–180.
Davis, C. , and Coauthors , 2008: Prediction of landfalling hurricanes with the advanced hurricane WRF model. Mon. Wea. Rev., 136 , 1990–2005.
Deardorff, J. W. , 1972: Numerical investigation of neutral and unstable planetary boundary layers. J. Atmos. Sci., 29 , 91–115.
Desai, A. R. , K. J. Davis , C. J. Senff , S. Ismail , E. V. Browell , D. R. Stauffer , and B. P. Reen , 2006: A case study on the effects of heterogeneous soil moisture on mesoscale boundary-layer structure in the southern Great Plains, USA. Part I: Simple prognostic model. Bound.-Layer Meteor., 119 , 195–238.
Ek, M. B. , K. E. Mitchell , Y. Lin , E. Rogers , P. Grummann , V. Koren , G. Gayno , and J. D. Tarpley , 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108 , 8851. doi:10.1029/2002JD003296.
Gorska, M. , J. Vila-Guerau de Arellano , M. A. LeMone , and C. C. van Heerwaarden , 2008: Mean and flux horizontal variability of virtual potential temperature, moisture, and carbon dioxide: Aircraft observations and LES study. Mon. Wea. Rev., 136 , 4435–4451.
Grossman, R. L. , D. Yates , M. A. LeMone , M. L. Wesely , and J. Song , 2005: Observed effects of horizontal radiative surface temperature variations on the atmosphere over a Midwest watershed during CASES 97. J. Geophys. Res., 110 , D06117. doi:10.1029/2004JD004542.
Holton, J. R. , 1972: An Introduction to Dynamic Meteorology. Academic Press, 319 pp.
Hong, S-Y. , Y. Noh , and J. Dudhia , 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134 , 2318–2341.
Kang, S. , 2009: Temporal oscillations in the convective boundary layer forced by mesoscale surface heat-flux variations. Bound.-Layer Meteor., 132 , 59–81. doi:10.1007/s10546-009-9391-5.
Kang, S. , and K. J. Davis , 2008: The effects of mesoscale surface heterogeneity on the fair-weather convective atmospheric boundary layer. J. Atmos. Sci., 65 , 3197–3213.
Kang, S. , K. J. Davis , and M. A. LeMone , 2007: Observations of the ABL structures over a heterogeneous land surface during IHOP_2002. J. Hydrometeor., 8 , 221–244.
Klazura, G. , and Coauthors , 2006: Atmospheric boundary layer measurements in south-central Kansas, 1997–2004. Bull. Amer. Meteor. Soc., 87 , 1319–1324.
Knievel, J. C. , G. H. Bryan , and J. P. Hacker , 2007: Explicit numerical diffusion in the WRF model. Mon. Wea. Rev., 135 , 3808–3824.
Krettenauer, K. , and U. Schumann , 1992: Numerical simulation of turbulent convection over wavy terrain. J. Fluid Mech., 237 , 261–299.
Lafore, J-P. , and Coauthors , 1998: The Meso-NH atmospheric simulation system. Part I: Adiabatic formulation and control simulations. Ann. Geophys., 16 , 90–109.
LeMone, M. A. , 1973: The structure and dynamics of horizontal roll vortices in the planetary boundary layer. J. Atmos. Sci., 30 , 1077–1091.
LeMone, M. A. , and Coauthors , 2002: CASES-97: Late-morning warming and moistening of the convective boundary layer over the Walnut River watershed. Bound.-Layer Meteor., 104 , 1–52.
LeMone, M. A. , and Coauthors , 2007: NCAR/CU surface, soil, and vegetation observations during the International H2O Project 2002 field campaign. Bull. Amer. Meteor. Soc., 88 , 65–81.
LeMone, M. A. , F. Chen , M. Tewari , J. Dudhia , B. Geerts , Q. Miao , R. L. Coulter , and R. L. Grossman , 2010: Simulating the IHOP_2002 fair-weather CBL with the WRF-ARW–Noah modeling system. Part I: Surface fluxes and CBL structure and evolution along the eastern track. Mon. Wea. Rev., 138 , 722–744.
Lenschow, D. H. , J. C. Wyngaard , and W. T. Pennell , 1980: Mean-field and second-moment budgets in a baroclinic, convective boundary layer. J. Atmos. Sci., 37 , 1313–1326.
Letzel, M. O. , and S. Raasch , 2003: Large eddy simulation of thermally induced oscillations in the convective boundary layer. J. Atmos. Sci., 60 , 2328–2341.
Lothon, M. , F. Couvreux , S. Donier , F. Guichard , P. Lacarrere , D. H. Lenschow , J. Noilhan , and F. Said , 2007: Impact of coherent eddies on airborne measurements of vertical turbulent fluxes. Bound.-Layer Meteor., 124 , 425–447. doi:10.1007/s10546-007-9182-9.
Mahrt, L. , J. Sun , D. Vickers , J. I. MacPherson , J. R. Pederson , and R. L. Desjardins , 1994: Observations of fluxes and inland breezes over a heterogeneous surface. J. Atmos. Sci., 51 , 2484–2499.
Miao, Q. , and B. Geerts , 2007: Finescale vertical structure and dynamics of some dryline boundaries observed in IHOP. Mon. Wea. Rev., 135 , 4161–4184.
Miao, S. , and F. Chen , 2008: Formation of horizontal convective rolls in urban areas. Atmos. Res., 89 , 298–304. doi:10.1016/j.atmosres.2008.02.013.
Moeng, C-H. , and P. P. Sullivan , 1994: A comparison of shear- and buoyancy-driven planetary boundary layer flows. J. Atmos. Sci., 51 , 999–1022.
Ogura, Y. , and Y. Chen , 1977: A life history of an intense mesoscale convective system in Oklahoma. J. Atmos. Sci., 34 , 1458–1476.
Patton, E. G. , P. P. Sullivan , and C-H. Moeng , 2005: The influence of idealized heterogeneity on wet and dry planetary boundary layers coupled to the land surface. J. Atmos. Sci., 62 , 2078–2097.
Pielke, R. A. , G. Dalu , J. Snook , T. Lee , and T. Kittell , 1991: Nonlinear influence of mesoscale land use on weather and climate. J. Climate, 4 , 1053–1069.
Reen, B. P. , D. R. Stauffer , K. J. Davis , and A. R. Desai , 2006: A case study on the effects of heterogeneous soil moisture on mesoscale boundary-layer structure in the southern Great Plains, USA. Part II: Mesoscale modelling. Bound.-Layer Meteor., 120 , 275–314.
Segal, M. , R. Avissar , M. McCumber , and R. Pielke , 1988: Evaluation of vegetation effects on the generation and modification of mesoscale circulations. J. Atmos. Sci., 45 , 2268–2292.
Skamarock, W. C. , 2004: Evaluating mesoscale NWP models using kinetic energy spectra. Mon. Wea. Rev., 132 , 3019–3032.
Skamarock, W. C. , J. B. Klemp , J. Dudhia , D. O. Gill , D. M. Barker , W. Wang , and J. G. Powers , 2005: A description of the Advanced Research WRF version 2. NCAR Tech. Note TN-468+STR, 88 pp. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307].
Souza, E. P. , N. O. Renno , and M. A. F. Silva Dias , 2000: Convective circulations induced by surface heterogeneities. J. Atmos. Sci., 57 , 2915–2922.
Strassberg, D. , M. LeMone , T. Warner , and J. Alfieri , 2008: Comparison of observed 10-m wind speeds to those based on Monin–Obukhov similarity theory using IHOP_2002 aircraft and surface data. Mon. Wea. Rev., 136 , 964–972.
Sun, W-Y. , and Y. Ogura , 1979: Boundary-layer forcing as a possible trigger to a squall-line formation. J. Atmos. Sci., 36 , 235–254.
Taylor, C. M. , D. J. Parker , and P. P. Harris , 2007: An observational case study of mesoscale atmospheric circulations induced by soil moisture. Geophys. Res. Lett., 34 , L15801. doi:10.1029/2007GL030572.
Trier, S. B. , F. Chen , and K. W. Manning , 2004: A study of convection initiation in a mesoscale model using high-resolution land-surface initial conditions. Mon. Wea. Rev., 132 , 2954–2976.
Walko, R. L. , W. R. Cotton , and R. A. Pielke Sr. , 1992: Large-eddy simulations of the effects of hilly terrain on the convective boundary layer. Bound.-Layer Meteor., 58 , 133–150.
Weckwerth, T. M. , T. W. Horst , and J. W. Wilson , 1999: An observational study of the evolution of horizontal convective rolls. Mon. Wea. Rev., 127 , 2160–2179.
Weckwerth, T. M. , and Coauthors , 2004: An overview of the International H2O Project (IHOP_2002) and some preliminary highlights. Bull. Amer. Meteor. Soc., 85 , 253–277.
Weisman, M. L. , W. C. Skamarock , and J. B. Klemp , 1997: The resolution dependence of explicitly modeled convective systems. Mon. Wea. Rev., 125 , 527–548.
Wyngaard, J. C. , 2004: Toward numerical modeling in the “Terra Incognita”. J. Atmos. Sci., 61 , 1816–1826.
Wyngaard, J. C. , L. J. Peltier , and S. Khanna , 1998: LES in the surface layer: Surface fluxes, scaling, and SGS modeling. J. Atmos. Sci., 55 , 1733–1754.
Young, G. S. , D. A. R. Kristovich , M. R. Hjelmfelt , and R. C. Foster , 2002: Rolls, streets, waves, and more: A review of quasi-two-dimensional structures in the atmospheric boundary layer. Bull. Amer. Meteor. Soc., 83 , 997–1001.
Boundary layer King Air flights and environmental conditions along the eastern track (zi = CBL depth).
Effects of resolution on CBL structure. All idealized runs use ARW-WRF V3, with uniform vegetation (grassland) and soil texture (silty clay loam), and surface, subsurface, and meteorological data from 37.4188°N, 96.6472°W on 20 June, with initial Ts and/or volumetric soil moisture varied using random perturbations. 22 Jun runs use ARW-WRF V2.1.2. Y = “yes,” N = “no,” hWRF = CBL depth from ARW-WRF, averaged over domain unless otherwise indicated. “Filt” refers to the Knievel et al. (2007) filter used in ARW-WRF to damp out unresolved motions; “Nest” refers to whether the finest resolution was a result of nesting.
Observed parameters related to development of mesoscale circulations, interpolated to 1830 UTC unless otherwise indicated.
Criterion −zi/L for convective-roll occurrence from model and observations at 1830 UTC.
There is a small bias (∼1 g m−3) between the LiCor, and the Lyman α, which was used in Part I because it was referenced to an EG and G chilled-mirror dewpointer. Since only horizontal changes in Θυ are considered, the bias should not influence the results.
Conzemius and Fedorovich (2006) suggest the relationship
The “regional” fluxes, being based on momentum fluxes computed from fluctuations relative to a 46-km flight-leg linear trend, include the effects of trees, buildings, etc.; while the “local” fluxes, being based on surface flux sites that are in open areas, represent the surface sampled (in this case, grasses).
The two grid sizes were always such that no interpolation was necessary. For example, for the larger grid 1 km, the smaller grid might be 500 m, so that the four 500-m squares exactly overlap with the 1-km squares. In this case, the 1-km grid square responds to the average of the values for the four 500-m squares; and the four 500-m squares respond to the 1-km square values.
Accounting for the retarding effects of friction would increase the time required to generate a circulation in (5). However, as pointed out in Holton and elsewhere, the frictional force increases as the square of the velocity. We neglect friction and Coriolis accelerations since our arguments are qualitative and we allow uℓ to increase to only 1 m s−1. For more intense circulations, such as those considered in Miao and Geerts (2007) and Souza et al. (2000) or in considering how a steady state is reached, these effects are important.
The unusually large negative values are related to the ground rising northward by about 200 m within the profiler triangle in Fig. 1. To obtain a true vertical wind relative to the center of the earth, a correction related to the wind vector and the gradient in terrain along that vector would be needed.