1. Introduction
Near-surface (2 m) temperature (NST) forecasts are critical for the protection of life and property, for economic and operational activities, and for routine day-to-day planning but remain a major challenge for numerical weather prediction. Modeling systems in many regions of the world have trouble simulating NSTs and typically underpredict the diurnal NST cycle, which largely reflects a pronounced nighttime NST warm bias (e.g., Steeneveld et al. 2008; Edwards et al. 2011; Kilpelainen et al. 2012; Holtslag et al. 2013; Ngan et al. 2013). These errors are especially prevalent in high-resolution modeling systems (<5-km grid spacing) over many regions of the western United States (e.g., Mass et al. 2002; Cheng and Steenburgh 2005; Hart et al. 2005; Zhang et al. 2013). By influencing low-level stratification, boundary layer depth and mixing, thermally driven flows, and convective initiation, NST forecast errors ultimately affect the prediction of precipitation (amount and type), fog and clouds, air quality, and surface and boundary layer winds (e.g., Hanna and Yang 2001; Rife et al. 2002; Marshall et al. 2003; Holt et al. 2006).
There have been numerous hypotheses concerning the sources of these NST forecast errors ranging from inadequate horizontal or vertical resolution to the inaccurate initialization and parameterization of boundary layer and land surface characteristics and processes (e.g., Hanna and Yang 2001; Mass et al. 2002; Marshall et al. 2003; Cheng and Steenburgh 2005). In this paper, we concentrate on the initialization and parameterization of land surface characteristics and processes, which control the surface energy budget and contribute to NST errors through the inaccurate partitioning of sensible, latent, and ground heat fluxes (e.g., Huang et al. 1996; Davis et al. 1999; Marshall et al. 2003; Reeves et al. 2011). In most land surface models (LSMs), land surface parameters (e.g., albedo, roughness length, and soil porosity) are specified using land-use and soil-type databases, whereas soil moisture and temperature are derived from observational data and/or land surface modeling. In either case, the incorrect specification of these land surface characteristics is at least partly responsible for NST forecast errors (e.g., Huang et al. 1996; Dirmeyer et al. 2000; Rife et al. 2004; Wen et al. 2012).
Soil moisture is an important initialized variable because it strongly influences NSTs, surface and boundary layer winds, and the development of moist convection (e.g., Ookouchi et al. 1984; Avissar and Pielke 1989; Segal et al. 1989; Doran and Zhong 1995; Banta and Gannon 1995; Huang et al. 1996; Holt et al. 2006; Zhou and Geerts 2013), but a lack of in situ observations, combined with instrument and representativeness errors, limits reliable soil-moisture assimilation (e.g., Dirmeyer et al. 2000; Godfrey and Stensrud 2008; Liu et al. 2011). As a result, most soil-moisture analyses are based on either land surface model simulations forced by meteorological data (Dirmeyer et al. 2002) or low-resolution satellite soil-moisture retrievals that can only retrieve shallow soil moisture (Jackson et al. 2010).
Soil moisture has a direct influence on the ratio of surface sensible and latent heat fluxes (i.e., the Bowen ratio; Bowen 1926) but can also affect the surface energy balance by altering the surface albedo and the soil thermal conductivity. An increase in soil moisture tends to decrease the surface albedo, especially for playa land surfaces with dissolvable salt crusts (Idso et al. 1975; Tapper 1988), and to increase the soil thermal conductivity since water has a higher thermal conductivity than the air it replaces (Cosenza et al. 2003). Soil thermal conductivity is difficult to estimate because it is a function of the volume fractions of water, air, and soil; the mineral composition of the soil; and the numerous interactions among these variables (Farouki 1986). Although there are several soil thermal conductivity estimates (e.g., Kersten 1949; de Vries 1963; Johansen 1975; McCumber and Pielke 1981; McInnes 1981; Campbell 1985), the complexity of soil structure and the processes involved preclude a physically accurate and mechanistically based predictive model (Tarnawski et al. 2009). McCumber and Pielke (1981) offer a simple method (hereinafter referred to as MP81) to estimate soil thermal conductivity that was incorporated into several LSMs, including early versions of the National Centers for Environmental Prediction–Oregon State University–U.S. Air Force–Office of Hydrologic Development LSM known as Noah (e.g., Noilhan and Planton 1989; Ek and Mahrt 1991; Viterbo and Beljaars 1995; Chen and Dudhia 2001; Ek et al. 2003). MP81 produces higher-than-observed soil thermal conductivity in very wet conditions and lower-than-observed soil thermal conductivity in dry conditions for some soil textures (Peters-Lidard et al. 1998). For this reason, MP81 was replaced by the Johansen (1975) method (hereinafter referred to as J75) in the version of the Noah LSM that is coupled to the Weather Research and Forecasting (WRF) Model (Skamarock et al. 2008), which is considered to be one of the most accurate parameterizations for land surface modeling (e.g., Farouki 1986; Peters-Lidard et al. 1998). More recent parameterizations have adopted J75 but have modified its variables (Tarnawski and Leong 2000; Balland and Arp 2005; Côté and Konrad 2005; Lu et al. 2007).
In this paper we examine the causes of NST forecast errors produced by WRF over Dugway Proving Ground (DPG) in the Great Salt Lake Desert of northwestern Utah (Fig. 1). This dryland region features an extensive playa that is surrounded by sparsely vegetated desert, which has a lower albedo, lower soil moisture, larger Bowen ratio, and lower soil thermal conductivity than the playa. These properties contribute to a larger diurnal temperature range (DTR) over the sparsely vegetated desert than the playa (Rife et al. 2002), and forecasting the resulting thermally forced flows (Tapper 1988) is paramount to DPG’s wind-sensitive military testing operations (Liu et al. 2008). We will show that operational and retrospective WRF simulations typically underpredict the diurnal temperature cycle over silt loam and sandy loam soils in the sparsely vegetated desert, with a pronounced nighttime warm bias. This bias can be reduced by improving the initialization of soil moisture and by using MP81 for silt loam and sandy loam soils.
Google Earth image of northwestern Utah with relevant landmarks. The perimeter of DPG is outlined (copyright 2013 Google; imagery copyright 2013 TerraMetrics).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
2. Data and methods
a. Operational NST forecasts
Four months of NST forecasts for DPG were examined to identify biases in WRF. We concentrate on September and October of 2011 and 2012 because these months are most relevant to the Mountain Terrain Atmospheric Modeling and Observations (MATERHORN) autumn 2012 field campaign, held at DPG from 25 September to 25 October 2012. We refer to September–October of 2011 and September–October of 2012 as the pre-MATERHORN and MATERHORN periods, respectively.
The source of the forecasts is the U.S. Army Test and Evaluation Command Four-Dimensional Weather System (4DWX), developed by the National Center for Atmospheric Research (NCAR). We used two versions of the 4DWX run for DPG (4DWX-DPG). In 2011 the system was based on version 3.2 of WRF, and in 2012 it was based on version 3.3.1 of WRF, which updated the urban land-use parameters and allows the Noah LSM to take seasonal roughness-length changes into account. The 4DWX-DPG has 30-, 10-, 3.3-, and 1.1-km one-way nested domains centered over DPG (Liu et al. 2008; Fig. 2). The use of one-way nesting reflects its superiority over two-way nesting in unpublished test cases. The vertical spacing of the 36 half-η levels varies from ~30 m near the surface, with the lowest half-η level at ~15 m AGL, to ~1250 m in the upper troposphere and lower stratosphere. The physics packages include the Rapid Radiative Transfer Model longwave radiation parameterization (Mlawer et al. 1997), Dudhia shortwave radiation parameterization (Dudhia 1989), Noah LSM (Chen and Dudhia 2001), Yonsei University planetary boundary layer parameterization (Hong et al. 2006), Lin et al. (1983) microphysics, new Kain–Fritsch cumulus parameterization (Kain 2004), and explicit sixth-order numerical diffusion (Knievel et al. 2007). The atmospheric data assimilation cycle uses Newtonian nudging over a 3-h period to assimilate observations from aviation routine weather report (METAR) stations, rawinsondes, profilers, buoys, aircraft, satellites, and other observing platforms (Liu et al. 2008). 4DWX-DPG is run every 3 h for eight times per day (0200, 0500, 0800, 1100, 1400, 1700, 2000, and 2300 UTC) and produces 48-h forecasts. Liu et al. (2008) provide additional information on the physic packages and data assimilation of 4DWX.
Domains used for 4DWX-DPG (30, 10, 3.3, and 1.1 km) and the ARW case studies (12, 4, and 1.3 km); the 30-s-resolution elevation (m) is shaded.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Land surface (e.g., albedo, roughness length, and emissivity) and soil (e.g., porosity, quartz content, and wilting point) parameters are fixed for each land-use category and soil-texture class, respectively. These parameters are found in lookup tables, which were compiled from a variety of studies (e.g., Cosby et al. 1984; Mahfouf et al. 1995; Peters-Lidard et al. 1998). In 2011 4DWX-DPG was initialized with the standard geographic data available with the community version of the WRF Model, modified to include three additional land-cover categories of playa, white sand, and lava. In 2012, the land-cover and terrain elevation were updated on the basis of the newer 33-category National Land Cover Database dataset (Fry et al. 2011), which increased the area defined as playa. The soil-texture class is defined by a 16-category U.S. Geological Survey dataset, which is also modified to include playa, white sand, and lava soil-texture classes. Initial soil-moisture and soil-temperature fields at 5-, 25-, 70-, and 150-cm depths are obtained from a relatively coarse 1.0° Global Forecasting System (GFS) analysis because anecdotal evidence suggests that under some circumstances it outperforms the 12-km North American Model (NAM) analysis at DPG. These fields are interpolated to 4DWX-DPG using the default WRF preprocessing interpolation schemes of 16-point parabolic interpolation away from water bodies and four-point or nearest-neighbor interpolation near water bodies.
b. Case studies
We examine in detail three 36-h periods with quiescent large-scale conditions during the pre-MATERHORN and MATERHORN periods to isolate local land–atmosphere processes. The first, hereinafter called 2011-DRY (1200 UTC 22 September 2011–0000 UTC 24 September 2011), and the second, hereinafter called 2011-WET (1200 UTC 11 October 2011–0000 UTC 13 October 2011), occurred during the pre-MATERHORN period and offer contrasting soil moistures (0.12 vs 0.19 m3 m−3, respectively, at DPG). Skies were clear for both events, but 19–24 mm of rain fell at DPG 3–7 days before 2011-WET, moistening the near-surface soil. The third event is MATERHORN intensive observing period 5 (MATERHORN-IOP5; 1200 UTC 9 October 2012–0000 UTC 11 October 2012). It also features dry soils (0.13 m3 m−3 at DPG), although high cirrus clouds were present, reducing the net radiation at the surface. Surface energy balance and tethersonde observations collected during MATERHORN-IOP5 enable a more thorough verification of the model solutions. We simulate these three cases using version 3.4 of the community Advanced Research version of WRF, with the same physics packages as in the 4DWX-DPG system but with three larger one-way nested domains (12-, 4-, and 1.3-km grid spacing) and cold-start initial conditions (Fig. 2). The initial 6 h of each simulation is excluded from the study to reduce the influence of model spinup of the atmosphere. The large 1.3-km domain allows us to cover the entire playa and to use a broader range of regional surface observations for model validation.
We generate a nine-member ensemble for each of the three cases on the basis of three different parameterizations of soil thermal conductivity and three different top-layer (5 cm) soil-moisture initial analyses in the 1.3-km domain. The three parameterizations of soil thermal conductivity are J75, MP81, and a hybrid that uses MP81 over silt loam and sandy loam soils and J75 elsewhere. The three top-layer soil-moisture analyses are the interpolated 1.0° GFS analysis, an interpolated 12-km NAM analysis, and a soil-moisture analysis that is created using data from the U.S. Department of Agriculture’s Soil Climate Analysis Network (SCAN; Schaefer et al. 2007).
The SCAN-based soil-moisture analysis uses data from five SCAN stations located in the 1.3-km domain. SCAN stations use Stevens Water Monitoring Systems, Inc., Hydra Probes to measure soil moisture, and the probes are calibrated differently for different soil-texture classes (Seyfried et al. 2005). Of the five SCAN stations in our innermost domain, three are in loam soil (Morgan, Nephi, and Grantsville), one is in sandy loam soil (Goshute), and one is in silt loam soil (Dugway; see Fig. 3 for locations). All five stations measure soil moisture hourly at depths of 5.1, 10.2, 20.3, 50.8, and 101.6 cm, but the Noah LSM is configured with depths centered at 5, 25, 70, and 150 cm below the surface, making only the 5.1-cm SCAN measurement directly relevant for initialization and validation without vertical interpolation. Sensitivity studies suggest, however, that NST forecasts are relatively insensitive to deep (25 cm or greater) soil moisture and the vertical gradient in soil during the study period (not shown).
Google Earth image of the innermost 1.3 km ARW domain with Mesowest (categorized by soil-texture class), SCAN, and EFS-sage stations annotated (copyright 2013 Google; imagery copyright 2013 TerraMetrics).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Our SCAN-based soil-moisture analyses begin with GFS soil moisture but replace the 5-cm soil moisture over the areas defined as silt loam with the Dugway 5.1-cm soil moisture, the areas defined as sandy loam with the Goshute 5.1-cm soil moisture, and the areas defined as loam with the mean of the Morgan, Nephi, and Grantsville 5.1-cm soil moistures. SCAN data are used for validation and initialization in other numerical weather prediction studies (e.g., Case et al. 2011; Li et al. 2012), but point soil-moisture observations are usually spatially and temporally interpolated to produce an observed soil-moisture field (e.g., Marshall et al. 2003; Robock et al. 2003; Godfrey and Stensrud 2008). Our approach of delineating soil moisture by soil-texture class is crude, but observed and simulated soil moistures depend critically on soil-texture class (Mostovoy and Anantharaj 2008). Figure 4 compares the 5-cm soil moisture from the three analyses. For all three cases, the GFS has the highest mean 5-cm soil moisture, followed by the NAM and then the SCAN.
GFS-, NAM-, and SCAN-derived 5-cm volumetric soil-moisture content for the three case studies. Grids are interpolated during WRF preprocessing.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
c. Validation data and methods
4DWX-DPG NST forecasts are verified using BEs for seven stations over playa soil (DPG-PLAYA) and 10 stations over silt loam soil (DPG-SL) between 1257 and 1335 m MSL at or near DPG (Fig. 5). One playa and two silt loam stations were not considered because they sit on the playa margin and often had relatively erratic and unrepresentative NSTs in comparison with the other stations. Since the 4DWX-DPG 1.1-km domain includes only two DPG-PLAYA stations, we use forecasts from the 3.3-km domain for verification over both soil-texture classes since there is little difference between the NST forecasts from the 1.1- and 3.3-km domains. For both DPG-PLAYA and DPG-SL stations, the BEs consider all eight 4DWX-DPG forecasts initialized 14–34 h before the forecast hour to avoid contamination from observational nudging.
Google Earth image of DPG and surrounding area with DPG-PLAYA (blue) and DPG-SL (red) stations used for verification. Open circles represent stations on the playa margin that were not considered for the validation (copyright 2013 Google; imagery copyright 2013 TerraMetrics).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
For the case studies, BEs are calculated for stations in each represented soil-texture class to examine model performance over different soil textures. This includes approximately 87 loam, 56 sandy loam, 35 silt loam, 15 silty clay, 9 playa, 8 silty clay loam, 4 water, and 2 sand Mesowest stations below 1750 m MSL in our 1.3-km domain (Fig. 3). The silty clay loam, water, and sand stations are not considered in the analysis. Some stations were not consistently active, and therefore the exact number of stations used for verification varied among the three case studies. Most stations are clustered along the Wasatch Front, and no stations are located over far northwestern Utah. The vast majority of these stations are either over shrubland or urban land cover, except for the nine playa stations, which are all over playa land cover. Although no formal quality control was performed on any of the observational datasets, missing and obviously erroneous observations were removed.
d. MATERHORN observations
During MATERHORN, soil and radiation observations were taken from 5 to 26 October 2012 at an extended flux site located near the DPG-SL stations (EFS-sage; Fig. 3). Soil temperatures were measured with thermocouples (Omega Engineering, Inc.) at depths of 1, 2.5, 5, 7.5, 10, 15, 25, and 70 cm. Thermal-property sensors (model TP01; Hukseflux Thermal Sensors B.V.) were installed at depths of 5, 10, and 25 cm and measured soil thermal conductivity [with a mean uncertainty of 0.01 W m−1 K−1 within the range of 0.3–4.0 W m−1 K−1 (Overduin et al. 2006)], soil thermal diffusivity, and volumetric heat capacity. Two self-calibrating heat flux plates (Hukseflux HFP-SC) measured the subsurface heat flux at 5-cm depths at two locations separated by approximately 1 m. The individual shortwave and longwave components of the surface radiation balance at EFS-sage were measured with Kipp & Zonen B.V. up- and down-facing CMP21 pyranometers and CGR4 pyrgeometers, respectively, mounted 2 m AGL. The surface ground heat flux at EFS-sage was calculated as the sum of the average measured ground heat flux at a 5-cm depth and the heat-storage change in the soil layer between 0 and 5 cm. The heat storage was calculated using the direct measurements of the thermal heat capacity at 5 cm below the surface and the soil temperatures at depths of 1, 2.5, and 5 cm.
A tethered balloon system (Vaisala, Inc., DigiCORA) was also flown up to 400 m AGL at regular intervals during MATERHORN near EFS-sage, collecting atmospheric temperature, humidity and wind profiles. For brevity, we compare observed and model profiles for a time near the peak strength of the nocturnal inversion (~1200 UTC).
3. Results
a. Systematic biases in operational NST forecasts
At 1200 UTC (0500 LST), the 4DWX-DPG BE is 3.4°C at the DPG-SL stations during the pre-MATERHORN and MATERHORN periods. At 0000 UTC (1700 LST), the BE at these sites is −1.1°C. This diurnal variation of BE at DPG-SL stations, especially the 1200 UTC (0500 LST) warm bias, is fairly consistent throughout the validation period and is most pronounced during quiescent large-scale conditions (Fig. 6). For example, when 700-hPa (~1000 m AGL) winds from the Salt Lake City (KSLC) sounding are less than 8 m s−1, the 1200 UTC warm bias at DPG-SL stations increases to 3.9°C and sometimes exceeds 6°C. Rife et al. (2002) documented similar nocturnal warm biases at their nonplaya DPG stations during an event they simulated from July of 1998. Meanwhile, the BEs at the DPG-PLAYA stations show little diurnal variation and are −0.7° and −0.6°C at 1200 and 0000 UTC, respectively, suggesting that the largest errors are confined to the DPG-SL stations.
4DWX-DPG NST BEs for DPG-PLAYA (blue) and DPG-SL (red) stations and the observed KSLC 700-hPa wind speed (black) at (top) 1200 and (bottom) 0000 UTC. The three case studies are highlighted for reference.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
The observed DTR1 is substantially larger at DPG-SL than at DPG-PLAYA stations. As shown by forecasts initialized at 1100 UTC the day prior, the 4DWX-DPG DTR is similar for both station types (Fig. 7). At DPG-SL stations, the mean observed DTR is 19.2°C, with some days exceeding 25°C, but the mean 4DWX-DPG DTR is only 12.7°C. At DPG-PLAYA stations, the mean observed DTR is 13.8°C, and the mean 4DWX-DPG DTR is only slightly smaller at 12.5°C. Therefore, the DTR at DPG-SL stations is underpredicted, resembling that at the DPG-PLAYA stations, where it is relatively well predicted. This result suggests that soil properties at the DPG-SL stations are inaccurately represented in 4DWX-DPG.
Observed (thick lines) and 4DWX-DPG (thin lines) DTRs at DPG-SL and DPG-PLAYA stations. The three case studies are highlighted for reference.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
b. Potential error sources
The analysis above illustrates the existence of a 1200 UTC warm bias and reduced DTR at DPG-SL stations but not at DPG-PLAYA stations. Rife et al. (2002) state that the major differences between the sparsely vegetated desert containing the DPG-SL stations and the playa are associated with the vegetation cover, albedo, soil thermal conductivity, and near-surface soil moisture. Therefore, the error at the DPG-SL stations is likely related to the initialization or parameterization of at least one of these surface properties. The WRF land cover at the DPG-SL stations is shrubland, which has a prescribed albedo of 0.25–0.30 that appears reasonable on the basis of comparisons with the Moderate Resolution Imaging Spectroradiometer (MODIS) white-sky albedo (known as MOD43B3) 16-day 1-km product (not shown). Therefore, the likely sources of the NST errors over DPG-SL are the parameterization and initialization of soil thermal conductivity and soil moisture.
Figure 8 shows κJ75 and κMP81 as a function of unfrozen volumetric soil-moisture content for five different soil-texture classes. Both methods have increasing thermal conductivity with increasing soil moisture since the water has a larger thermal conductivity than the air it replaces (Cosenza et al. 2003). The κMP81 is more variable among the different soil-texture classes than is κJ75, and κMP81 is more sensitive to soil moisture over certain soil-moisture ranges (i.e., 0.15–0.33 m3 m−3 for silt loam). The κMP81 and κJ75 are substantially different over silt loam at low and high soil moistures. For example, when the silt loam soil moisture is relatively moist at 0.33 m3 m−3, κMP81 is 1.9 W m−1 K−1 but κJ75 is only 1.14 W m−1 K−1; when the soil moisture is relatively dry at 0.12 m3 m−3, κMP81 is 0.17 W m−1 K−1 but κJ75 is 0.64 W m−1 K−1. The κJ75 and κMP81 are similar near their intersection point at 0.25 m3 m−3. For playa and silty clay soil textures, κJ75 and κMP81 are even more dissimilar at low soil moistures than for silt loam, and they do not intersect until the soils are much more saturated. The κJ75 and κMP81 are also very sensitive to the WRF-defined soil parameters, such as soil porosity (not shown), but the representativeness of those parameters is not examined in this study.
Calculated J75 (κJ75; solid lines) and MP81 (κMP81; dashed lines) thermal conductivity as a function of volumetric soil moisture for silt loam, playa, sandy loam, loam, and silty clay soil textures.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
The κJ75 and κMP81 are sensitive to soil moisture, and there is considerable soil-moisture disagreement among the GFS and NAM analyses and the Dugway SCAN soil-moisture observations. For example, at the Dugway SCAN station during the pre-MATERHORN and MATERHORN periods, the GFS, NAM, and observed 5-cm soil moistures average 0.22, 0.19, and 0.16 m3 m−3, respectively (Fig. 9).2 All three values are between the WRF prescribed vegetation wilting point of 0.084 m3 m−3 and porosity of 0.476 m3 m−3 assigned to silt loam soil. The moist bias in the analyzed fields is more pronounced during the pre-MATERHORN period, especially in the GFS, which averages 70% more soil moisture than is observed at the Dugway SCAN station. The GFS soil moisture is nudged toward a soil-moisture “climatology” that may be too wet over this area. During MATERHORN, after a significant rain event on 2 September 2012 there is fair agreement among all three soil moistures, with the GFS slightly wetter, until the precipitation event on 12–13 October 2012. The NAM did not respond significantly, the GFS responded suddenly, and the Dugway SCAN station responded slowly. In general, the accumulated daily precipitation at the Dugway SCAN station helps to explain many of the observed soil-moisture increases but only partially explains the NAM and GFS soil-moisture increases. The GFS analyzed soil moisture is a function of its precipitation output, whereas the NAM analyzed soil moisture is a function of radar-derived and gauge-based precipitation. These differing inputs help to explain why the soil-moisture analyses react differently after a rain event (Liu et al. 2011).
Time series of 0000 and 1200 UTC 5-cm volumetric soil moisture with accumulated 24-h precipitation (0000–0000 UTC; bars) at the Dugway SCAN station. The three case studies are highlighted for reference.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
c. Case studies
Sensitivity to the initialization of soil moisture and the parameterization of soil thermal conductivity is illustrated through simulations of 2011-DRY, 2011-WET, and MATERHORN-IOP5. We simulate these events using a nine-member ensemble with varying 5-cm soil-moisture initialization among the GFS-, NAM-, and SCAN-derived soil-moisture analyses and varying soil thermal conductivity parameterizations among J75, MP81, and a hybrid approach that uses MP81 over silt loam and sandy loam soil-texture classes and J75 over the other texture classes. All three cases occurred during quiescent large-scale conditions, and we only focus on NST BEs for silt loam, playa, sandy loam, loam, and silty clay soil-texture stations.
1) 2011-DRY
The 2011-DRY case featured an amplifying upper-level ridge centered over Utah (not shown). In fact, the observed 700-hPa winds from the 1200 UTC 22 September KSLC sounding were only 4.6 m s−1. Soils were generally dry in the SCAN network (e.g., 0.12 m3 m−3 at the Dugway SCAN station) but relatively moist in the GFS and NAM analyses (e.g., 0.24 and 0.21 m3 m−3, respectively, over silt loam; Table 1). The quiescent conditions and dry soils contributed to a difference of 6.1°C between the mean observed temperature at the DPG-PLAYA and DPG-SL stations at 1200 UTC 23 September and a large mean observed DTR at the DPG-SL stations of 24.7°C (Fig. 10). The control ensemble member (J75-GFS) for this case produces a 1200 UTC warm bias of 6.8°C at the DPG-SL stations, which results in a mean difference of only 0.2°C between the DPG-PLAYA and DPG-SL stations at 1200 UTC 23 September and a mean DTR at the DPG-SL stations of only 13.0°C.
Mean initialized 5-cm volumetric soil moisture (m3 m−3).
Mean observed and simulated DTRs for DPG-SL stations, mean observed and simulated temperature difference between DPG-PLAYA and DPG-SL stations at 1200 UTC, and the BE at DPG-SL stations at 1200 UTC for each case study.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Figures 11a–c show the BEs over the five different soil textures for the ensemble members that use J75 with varying soil moisture. In J75-GFS, silt loam, playa, sandy loam, loam, and silty clay stations have consistent 0000 UTC cool biases of 3.1°–3.5°C but varying 1200 UTC warm biases of 0.8°–4.9°C on 23 September (Fig. 11a). Silt loam stations have the greatest 1200 UTC warm bias, with a mean late-night (0600–1200 UTC) warm bias of 4.9°C (Fig. 12). Reducing the soil moisture reduces the late-night warm biases modestly. For example, J75-NAM and J75-SCAN initialize silt loam soil moisture to 12% and 50% lower than that for J75-GFS, respectively, but these changes only reduce the late-night BEs at silt loam stations by 0.1° and 1.1°C, respectively (Fig. 12).
The BE over each soil-texture class during 2011-DRY (1800 UTC 22 Sep–0000 UTC 24 Sep 2011) for (a) J75-GFS, (b) J75-NAM, (c) J75-SCAN, (d) MP81-GFS, (e) MP81-NAM, (f) MP81-SCAN, (g) hybrid-GFS, (h) hybrid-NAM, and (i) hybrid-SCAN. Yellow shading and gray shading represent daylight and nighttime hours, respectively.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
2011-DRY, 2011-WET, and MATERHORN-IOP5 mean 0600–1200 UTC BEs (°C) over each soil type for each ensemble member.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Figures 11d–f show the BEs for the ensemble members that use MP81 with varying soil moisture. The BEs for silt loam, sandy loam, and loam stations in MP81-GFS (Fig. 11d) are very similar to J75-GFS (Fig. 11a), but MP81-GFS introduces a large nighttime cool bias at playa and silty clay stations. Nighttime NST forecasts from MP81 members are also more sensitive to soil moisture than the J75 members. For example, mean late-night BEs at the loam soil stations decrease by 3.1°C between MP81-GFS and MP81-SCAN, respectively, but they only decrease by 0.8°C between J75-GFS and J75-SCAN (Fig. 12). The greatest mean late-night BE improvement occurs at silt loam stations where mean late-night BEs are 4.6°C in MP81-GFS but are reduced to 0.3°C in MP81-SCAN. MP81 introduces an unfortunate evening-transition cool bias at 0100 UTC, especially in MP81-SCAN (Fig. 11f). Figure 13 shows mean observed, J75-SCAN, and MP81-SCAN NSTs for the DPG-SL stations and how MP81-SCAN begins to cool rapidly after 0000 UTC, an hour earlier than observed, which introduces the 0100 UTC cool bias. Perhaps MP81-SCAN predicts the onset of the evening transition too early. Figure 13 also illustrates how MP81-SCAN slightly reduces the afternoon cool bias but significantly reduces the nighttime warm bias in comparison with J75-SCAN. At night, after the initial rapid cooling after sunset, the temperature tendency of J75-SCAN and MP81-SCAN is nearly identical to the observed tendency, suggesting that κ and G have the greatest influence on NST immediately following sunset.
Time series of mean observed and simulated NST at DPG-SL stations from 1800 UTC 22 Sep to 0000 UTC 24 Sep 2011. Yellow shading and gray shading represent daylight and nighttime hours, respectively.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Figures 11g–i show the BEs for the hybrid ensemble members that use κMP81 for silt loam and sandy loam soils, and κJ75 for all other soil textures. The BEs in hybrid-GFS (Fig. 11g) and hybrid-NAM (Fig. 11h) are similar to the BEs in J75-GFS and J75-NAM, respectively, but the large nighttime warm bias at silt loam stations nearly disappears in hybrid-SCAN (Fig. 11i). Hybrid-SCAN also has less variance in BEs among the different soil-texture stations. Of interest is that the hybrid ensemble members introduce additional errors over some soil textures that are unrelated to soil moisture and κ. For example, MP81-GFS and hybrid-GFS yield a mean late-night BE of 1.9° and 2.9°C, respectively, at sandy loam stations even though sandy loam uses MP81 for both members (Fig. 12). This may be related to temperature advection from areas with other soil textures that have an adjusted κ, or a sensitive land–atmosphere feedback process. Also of interest, the significant NST changes in hybrid-SCAN did not greatly affect the 10-m winds in 2011-DRY or the other case studies. For example, silt loam mean late-night wind speed BEs are −0.4 m s−1 in J75-GFS and −0.2 m s−1 in hybrid-SCAN (not shown).
2) 2011-WET
The 2011-WET case occurred during a quiescent period that followed 19–24 mm of rainfall over DPG from 4 to 8 October 2011, with a corresponding increase in area-averaged GFS, NAM, and SCAN 5-cm soil moisture of 11%, 16%, and 31% relative to 2011-DRY, respectively (Fig. 4). The increase in soil moisture likely helped to reduce the difference in mean observed temperature at the DPG-PLAYA and DPG-SL stations at 1200 UTC to 3.9°C, which is 2.2°C lower than in 2011-DRY, and to lower the mean observed DTR at DPG-SL stations to 18.4°C, which is 6.3°C lower than in 2011-DRY (Fig. 10). Our control J75-GFS forecast produced a 1200 UTC warm bias of 4.2°C at the DPG-SL stations as compared with 6.8°C in 2011-DRY. Forecast DTRs are similar between 2011-DRY and 2011-WET, but because the observed DTR is 6.3°C smaller during 2011-WET the DTR underprediction is also reduced.
Figures 14a–c show the BEs for the J75 members. Relative to 2011-DRY, J75-GFS has similar 0000 UTC cool biases of 2.4°–3.0°C over all soil textures but a smaller range in the 1200 UTC warm biases of 0.5°–2.9°C (Fig. 14a). Silt loam stations again have the highest mean late-night BE of 1.9°C (Fig. 12). Silt loam soil moisture reduces by 0.09 m3 m−3 between the GFS and SCAN analyses from 0.28 m3 m−3 to a value of 0.19 m3 m−3 (Table 1), but this change only translates to a mean late-night BE reduction of 0.3°C (Fig. 12). For comparison, silt loam soil moisture also reduces by 0.09 m3 m−3 in 2011-DRY, but from 0.21 m3 m−3 to a value of 0.12 m3 m−3 between the NAM and SCAN analyses (Table 1), and mean late-night BEs reduce by 1.0°C under this drier regime. Therefore, NSTs at silt loam stations are less sensitive to the higher soil moisture in 2011-WET than in 2011-DRY, which is to be expected from Fig. 8.
As in Fig. 11, but for 2011-WET (1800 UTC 11 Oct–0000 UTC 13 Oct 2011).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Figures 14d–f and 14g–i show the BEs for the MP81 and hybrid members, respectively, during 2011-WET. Similar to 2011-DRY, MP81 members introduce a significant cool bias at silty clay and playa stations and also make nighttime BEs more sensitive to soil moisture than do J75 members. The 0100 UTC cool bias in the MP81 members of the 2011-DRY case is no longer apparent, however. The hybrid-SCAN member has improved late-night BEs over all soil textures relative to those in J75-GFS, and the BE at silt loam stations is reduced to 1.0°C (Fig. 12).
3) MATERHORN-IOP5
MATERHORN-IOP5 was also characterized by quiescent large-scale conditions, with 700-hPa winds of only 5.2 m s−1 in the 0000 UTC 10 October KSLC sounding. An upper-level cutoff low off the Californian coast drove high cirrus clouds over the region throughout the period, but precipitation was not observed (not shown). In fact, only 1–7 mm of rain was observed at DPG-SL stations since 2 September 2012. As a result, MATERHORN-IOP5 had the lowest 5-cm soil moisture of the three cases (Fig. 4). The SCAN soil moisture over loam soil was especially low at 0.05 m3 m−3 (Table 1), which is the mean of 0.12 m3 m−3 at Nephi, 0.03 m3 m−3 at Morgan, and 0.01 m3 m−3 at Grantsville. Cloud cover was likely responsible for a reduced 1200 UTC temperature difference of 4.2°C between DPG-PLAYA and DPG-SL stations and for a DTR of 22.7°C at DPG-SL stations on 10 October 2012, which is 2°C lower than in 2011-DRY and 4.3°C lower than 2 days prior on 8 October 2012 (i.e., Fig. 7). J75-GFS forecast a 1200 UTC BE of 6.0°C at DPG-SL stations and underpredicted the mean DTR at these stations by 9°C (Fig. 10).
Figure 15 shows a very similar BE pattern in comparison with 2011-DRY except that the daytime cool bias is not as large and the mean nighttime warm bias at silt loam stations is not as pronounced. The mean late-night warm bias at silt loam stations decreases from 4.3°C in J75-GFS to 1.6°C in hybrid-SCAN (Fig. 12). The mean late-night BEs for the rest of the soil textures, except for loam, are very similar (≥1.3°C) in hybrid-SCAN. This suggests the presence of a nighttime warm bias unrelated to the LSM. The BEs at loam stations in J75-SCAN, MP81-SCAN, and hybrid-SCAN are relatively lower than at the other stations because the SCAN soil moisture for loam is very low and may not be representative.
As in Fig. 11, but for MATERHORN-IOP5 (1800 UTC 9 Oct–0000 UTC 11 Oct 2012).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
MATERHORN observations taken at EFS-sage during MATERHORN-IOP5 indicate that hybrid-SCAN also improves forecasts of other near-surface variables. A comparison of the potential temperature profile of a tethersonde ascending between 1201 and 1221 UTC 10 October with the 1200 UTC Hybrid-SCAN and J75-GFS profiles at EFS-sage illustrates how Hybrid-SCAN is colder than J75-GFS and closer to the observations below ~150 m AGL (Fig. 16). Figure 17 illustrates how hybrid-SCAN better predicts G at EFS-sage than does J75-GFS. J75-GFS overpredicts the diurnal amplitude of G, leading to more subsurface heat storage during the day and thus warmer surface temperatures and larger longwave radiation emission at night. For example, J75-GFS predicts a G of −167 W m−2 at 1800 UTC 10 October, which is 50 W m−2 lower than observed, and 93 W m−2 at 0100 UTC, which is 42 W m−2 higher than observed. Hybrid-SCAN slightly underpredicts the magnitude of G during the day but nearly matches the nighttime observations. Hybrid-SCAN also forecasts upwelling longwave radiation better than J75-GFS does, especially at night (Fig. 18a). There are only minor differences between the hybrid-SCAN and J75-GFS downwelling longwave radiation (Fig. 18b), and there are no discernable upwelling and downwelling shortwave radiation differences (Figs. 18c,d). Latent and sensible heat flux data were not available during this IOP and could not be verified.
Simulated 1200 UTC 10 Oct 2012 potential temperature (K) profiles at EFS-sage and an observed potential temperature profile taken between 1201 and 1221 UTC 10 Oct 2012 from an ascending tethersonde.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
Observed and simulated ground heat flux at EFS-sage from 1800 UTC 9 Oct to 0000 UTC 11 Oct 2012. Yellow shading and gray shading represent daylight and nighttime hours, respectively.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
As in Fig. 17, but for the longwave and shortwave components of the surface energy balance.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
d. Observed soil thermal conductivity
Figure 19 shows 3 weeks of observed TP01 κ in comparison with κJ75 and κMP81 calculated using interpolated GFS, NAM, and SCAN 5-cm soil moisture at EFS-sage. Before the rain event on 12–13 October 2012, κJ75 is nearly double the observed TP01 κ, but κMP81 produces soil thermal conductivities that are within 0.21 W m−1 K−1 of the observed TP01 κ. After the rain event, there is considerable variability among the soil-moisture analyses (e.g., Fig. 9), and this variability is also evident in κMP81, but little variability is seen in κJ75. The soil-moisture sensor at EFS-sage was not working properly during MATERHORN, and therefore the soil moisture at this site is unknown.
Time series of 0000 and 1200 UTC observed 5-cm soil thermal conductivity at EFS-sage (solid black line) and calculated κJ75 (solid lines as in color key) and κMP81 (dashed lines as in color key) using interpolated or SCAN observed 5-cm soil moisture. Accumulated 24-h precipitation at the Dugway SCAN station is also shown (green bars).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-13-0362.1
4. Conclusions
Four months of operational WRF near-surface temperature forecasts in September and October of 2011 and 2012 show a pronounced warm bias of 3.4°C at 1200 UTC over the sparsely vegetated silt loam soil area of Dugway Proving Ground. The WRF forecasts also underpredict the magnitude of the diurnal temperature range at DPG-SL stations, producing a DTR similar to that over the adjacent playa region. Because NST forecasts over DPG-PLAYA are relatively accurate, the DPG-SL errors are likely related to the specification or parameterization of soil characteristics and properties in the Noah LSM.
Case studies reveal that BEs at DPG-SL stations, as well as at other stations in the surrounding region with silt or sandy loam soils, are highly sensitive to the soil thermal conductivity κ, which controls the ground heat flux G in the Noah LSM. The methods introduced by Johansen (1975) and by McCumber and Pielke (1981) have become the two most widely used κ parameterizations in LSMs, and both depend on soil moisture. We mitigate the nighttime warm bias in silt loam and sandy loam soil regions by switching from J75 to MP81 for these soils only and by using a soil-texture-class-dependent soil-moisture analysis created from SCAN observations at 5 cm. The κMP81 is more sensitive to soil moisture than κJ75, and, since SCAN observations are mostly drier than the analyzed soil-moisture fields of the NAM and GFS, the reduced soil moisture combined with MP81 greatly reduces κ and the nighttime NST warm bias. The greatest NST improvements occurred over areas with silt loam soil textures, like the area near the DPG-SL stations, which relies on accurate near-surface forecasts for military operations. Mean late-night BEs decreased by more than 4°C during 2011-DRY at all silt loam stations and decreased from 2.1°C to a value of −0.4°C at sandy loam stations. The variance in bias error among the different soil textures also reduces when these changes are incorporated. The BE improvement was substantially less during 2011-WET, which had higher soil moistures. The soil moistures in 2011-WET were still considerably below saturation, preventing us from validating MP81 over silt loam and sandy loam during very wet conditions. A daytime cool bias was evident in each case study before and after the changes were made to soil moisture and κ, suggesting that the cool bias depends on more than just the Noah LSM and may be related to errors in the surface-layer parameterization.
MP81 used in conjunction with 5-cm SCAN soil moisture over silt loam and sandy loam soils not only improved NST forecasts but also improved the predicted κ, G, longwave and shortwave radiation, and low-level thermal profile. The κMP81 more closely matched observations of κ taken from a TP01 sensor at EFS-sage during dry periods than did κJ75. The κMP81 variability also more closely matched the observed TP01 κ variability after a rain event than did the κJ75 variability. The G was observed during MATERHORN-IOP5 at EFS-sage, and the run that used MP81 and SCAN soil moisture more closely matched observed G than did the run that used J75 and GFS soil moisture. Upwelling longwave emission also improved at night, with no detrimental effects to the other surface energy balance components, and the low-level 1200 UTC potential temperature profile improved in the lowest 150-m AGL.
Peters-Lidard et al. (1998) provided the rationale for replacing MP81 with J75 in the Noah LSM on the basis of a comparison with data collected in Kansas during the first International Satellite Land Surface Climatology Project. They verified κMP81 and κJ75 against observed κ from sand, clay, and peat soil textures and found that κJ75 better predicted the κ values and associated surface fluxes. In their study, MP81 underestimated κ at low soil moisture and overestimated it at high soil moisture. The change to J75 in the WRF version of the Noah LSM occurred in 2001, which may explain why Rife et al. (2002) and Davis et al. (1999) noted a pronounced cold bias over the playa but Reeves et al. (2011) discovered only a slight morning warm bias. Peters-Lidard et al. (1998) never mentioned any simulations involving silt loam and sandy loam soils textures, which we have shown to perform better using MP81.
We expect other dryland regions with silt loam and sandy loam soil textures to have improved NST forecasts when MP81 is used in conjunction with improved soil-moisture initialization. Therefore, we recommend that the Noah LSM incorporate an optional MP81 soil thermal conductivity parameterization for use over dryland regions with silt loam and sandy loam soils.
Acknowledgments
This research was funded by Office of Naval Research Award N00014-11-1-0709, Mountain Terrain Atmospheric Modeling and Observations Program, and by the U.S. Army Test and Evaluation Command through an interagency agreement with the National Science Foundation. We thank our fellow MATERHORN participants, and we thank specifically Silvana Di Sabatino and Laura Leo of Notre Dame for collecting the tethersonde data; Michael Barlage, Fei Chen, Michael Duda, Pedro Jimenez, Branko Kosovic, Rebecca Ruttenberg, Mukul Tawari, and Steve Oncley of NCAR; Micheal Ek of NCEP; and John Pace (program manager for operational meteorology, U.S. Army RDT&E) for their input and contributions. During research for this article, visits to NCAR by authors Massey and Steenburgh were made possible by NCAR’s Advanced Study Program.
REFERENCES
Al Nakshabandi, G., and H. Kohnke, 1965: Thermal conductivity and diffusivity of soils as related to moisture tension and other physical properties. Agric. Meteor., 2, 271–279, doi:10.1016/0002-1571(65)90013-0.
Avissar, R., and R. A. Pielke, 1989: A parameterization of heterogeneous land surfaces for atmospheric numerical models and its impact on regional meteorology. Mon. Wea. Rev., 117, 2113–2136, doi:10.1175/1520-0493(1989)117<2113:APOHLS>2.0.CO;2.
Balland, V., and P. A. Arp, 2005: Modeling soil thermal conductivities over a wide range of conditions. J. Eng. Environ. Sci., 4, 549–558, doi:10.1139/s05-007.
Banta, R. M., and P. T. Gannon, 1995: Influence of soil moisture on simulations of katabatic flow. Theor. Appl. Climatol., 52, 85–94, doi:10.1007/BF00865509.
Bowen, I. S., 1926: The ratio of heat losses by conduction and by evaporation from any water surface. Phys. Rev., 27, 779–787, doi:10.1103/PhysRev.27.779.
Campbell, G. S., 1985: Soil Physics with BASIC: Transport Models for Soil–Plant Systems. Developments in Soil Science, Vol. 14, Elsevier, 150 pp.
Case, J. L., S. V. Kumar, J. Srikishen, and G. J. Jedlovec, 2011: Improving numerical weather predictions of summertime precipitation over the southeastern United States through a high-resolution initialization of the surface state. Wea. Forecasting, 26, 785–807, doi:10.1175/2011WAF2222455.1.
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, doi:10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2.
Cheng, W. Y. Y., and W. J. Steenburgh, 2005: Evaluation of surface sensible weather forecasts by the WRF and the Eta Models over the western United States. Wea. Forecasting, 20, 812–821, doi:10.1175/WAF885.1.
Clapp, R. B., and G. M. Hornberger, 1978: Empirical equations for some soil hydraulic properties. Water Resour. Res., 14, 601–604, doi:10.1029/WR014i004p00601.
Cosby, B. J., G. M. Hornberger, R. B. Clapp, and T. R. Ginn, 1984: A statistical exploration of the relationship of moisture characteristics to the physical properties of soils. Water Resour. Res., 20, 682–690, doi:10.1029/WR020i006p00682.
Cosenza, P., R. Guerin, and A. Tabbagh, 2003: Relationship between thermal conductivity and water content of soils using numerical modeling. Eur. J. Soil Sci., 54, 581–587, doi:10.1046/j.1365-2389.2003.00539.x.
Côté, J., and J. M. Konrad, 2005: A generalized thermal conductivity model for soils and construction materials. Can. Geotech. J., 42, 443–458, doi:10.1139/t04-106.
Davis, C., T. Warner, E. Astling, and J. Bowers, 1999: Development and application of an operational, relocatable, mesogamma-scale weather analysis and forecasting system. Tellus, 51A, 710–727, doi:10.1034/j.1600-0870.1999.00013.x.
de Vries, D. A., 1963: Thermal properties of soils. Physics of Plant Environment, W. R. van Wijk, Ed., North-Holland, 210–235.
Dirmeyer, P. A., F. J. Zeng, A. Ducharne, J. C. Morrill, and R. D. Koster, 2000: The sensitivity of surface fluxes to soil water content in three land surface schemes. J. Hydrometeor., 1, 121–134, doi:10.1175/1525-7541(2000)001<0121:TSOSFT>2.0.CO;2.
Dirmeyer, P. A., X. Gao, and T. Oki, 2002: GSWP-2: The Second Global Soil Wetness Project science and implementation plan. International GEWEX Project Office Publication Series 37, 65 pp. [Available online at http://www.iges.org/gswp/GSWP2_SIplan.pdf.]
Doran, J. C., and S. Zhong, 1995: Variations in mixed-layer depths arising from inhomogeneous surface conditions. J. Climate, 8, 1965–1973, doi:10.1175/1520-0442(1995)008<1965:VIMLDA>2.0.CO;2.
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.
Edwards, J. M., J. R. McGregor, M. R. Bush, and F. J. Bornemann, 2011: Assessment of numerical weather forecasts against observations from Cardington: Seasonal diurnal cycles of screen-level and surface temperatures and surface fluxes. Quart. J. Roy. Meteor. Soc., 137, 656–672, doi:10.1002/qj.742.
Ek, M., and L. Mahrt, 1991: OSU 1-D PBL model user’s guide. Version 1.04, 120 pp. [Available from Department of Atmospheric Sciences, Oregon State University, Corvallis, OR 97331-2209.]
Ek, M., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, V. Koren, G. Gayno, and J. D. Tarplay, 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.
Farouki, O. T., 1986: Thermal Properties of Soils. Series on Rock and Soil Mechanics, Vol. 11, Trans Tech, 136 pp.
Fry, J., and Coauthors, 2011: Completion of the 2006 National Land Cover Database for the conterminous United States. Photogramm. Eng. Remote Sens., 77, 858–864. [Available online at http://digital.ipcprintservices.com/publication/?i=78634.]
Godfrey, C. M., and D. J. Stensrud, 2008: Soil temperature and moisture errors in operational Eta Model analyses. J. Hydrometeor., 9, 367–387, doi:10.1175/2007JHM942.1.
Hanna, S. R., and R. Yang, 2001: Evaluations of mesoscale models’ simulations of near-surface winds, temperature gradients, and mixing depths. J. Appl. Meteor., 40, 1095–1104, doi:10.1175/1520-0450(2001)040<1095:EOMMSO>2.0.CO;2.
Hart, K. A., W. J. Steenburgh, and D. J. Onton, 2005: Model forecast improvements with decreased horizontal grid spacing over finescale intermountain orography during the 2002 Olympic Winter Games. Wea. Forecasting, 20, 558–576, doi:10.1175/WAF865.1.
Holt, T. R., D. Niyogi, F. Chen, K. Manning, M. A. LeMone, and A. Qureshi, 2006: Effect of land–atmosphere interactions on the IHOP 24–25 May 2002 convection case. Mon. Wea. Rev., 134, 113–133, doi:10.1175/MWR3057.1.
Holtslag, A. A. M., and Coauthors, 2013: Stable atmospheric boundary layers and diurnal cycles: Challenges for weather and climate models. Bull. Amer. Meteor. Soc., 94, 1691–1706, doi:10.1175/BAMS-D-11-00187.1.
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, doi:10.1175/MWR3199.1.
Horel, J., and Coauthors, 2002: MesoWest: Cooperative mesonets in the western United States. Bull. Amer. Meteor. Soc., 83, 211–225, doi:10.1175/1520-0477(2002)083<0211:MCMITW>2.3.CO;2.
Huang, J., H. M. van den Dool, and K. P. Georgarakos, 1996: Analysis of model-calculated soil moisture over the United States (1931–1993) and applications to long-range temperature forecasts. J. Climate, 9, 1350–1362, doi:10.1175/1520-0442(1996)009<1350:AOMCSM>2.0.CO;2.
Idso, S., R. Jackson, B. Kimball, and F. Nakayama, 1975: The dependence of bare soil albedo on soil water content. J. Appl. Meteor., 14, 109–113, doi:10.1175/1520-0450(1975)014<0109:TDOBSA>2.0.CO;2.
Jackson, T. J., and Coauthors, 2010: Validation of advanced microwave scanning radiometer soil moisture products. IEEE Trans. Geosci. Remote Sens., 48, 4256–4272, doi:10.1109/TGRS.2010.2051035.
Johansen, O., 1975: Thermal conductivity of soils. Ph.D. thesis, University of Trondheim, 236 pp. [Available from Universitetsbiblioteket i Trondheim, Høgskoleringen 1, 7034 Trondheim, Norway.]
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, doi:10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Kersten, M. S., 1949: Thermal properties of soils. University of Minnesota Engineering Experiment Station Bulletin 28, 227 pp. [Available from University of Minnesota Agricultural Experiment Station, St. Paul, MN 55108.]
Kilpelainen, T., T. Vihma, M. Manninen, A. Sjoblom, E. Jakobson, T. Palo, and M. Maturilli, 2012: Modelling the vertical structure of the atmospheric boundary layer over Arctic fjords in Svalbard. Quart. J. Roy. Meteor. Soc., 138, 1867–1883, doi:10.1002/qj.1914.
Knievel, J. C., G. H. Bryan, and J. P. Hacker, 2007: Explicit diffusion in the WRF Model. Mon. Wea. Rev., 135, 3808–3824, doi:10.1175/2007MWR2100.1.
Li, B., D. Toll, X. Zhan, and B. Cosgrove, 2012: Improving estimated soil moisture fields through assimilation of AMSR-E soil moisture retrievals with an ensemble Kalman filter and a mass conservation constraint. Hydrol. Earth Syst. Sci., 16, 105–119, doi:10.5194/hess-16-105-2012.
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.
Liu, Q., and Coauthors, 2011: The contributions of precipitation and soil moisture observations to the skill of soil moisture estimates in a land data assimilation system. J. Hydrometeor., 12, 750–765, doi:10.1175/JHM-D-10-05000.1.
Liu, Y., and Coauthors, 2008: The operational mesogamma-scale analysis and forecast system of the U.S. Army Test and Evaluation Command. Part I: Overview of the modeling system, the forecast products, and how the products are! used. J. Appl. Meteor. Climatol., 47, 1077–1092, doi:10.1175/2007JAMC1653.1.
Lu, S., T. S. Ren, Y. S. Gong, and R. Horton, 2007: An improved model for predicting soil thermal conductivity from water content at room temperature. Soil Sci. Soc. Amer. J., 71, 8–14, doi:10.2136/sssaj2006.0041.
Mahfouf, J. F., A. O. Manzi, J. Noilhan, H. Giordani, and M. Deque, 1995: The land surface scheme ISBA within the Météo-France climate model ARPEGE. Part I: Implementation and preliminary results. J. Climate, 8, 2039–2057, doi:10.1175/1520-0442(1995)008<2039:TLSSIW>2.0.CO;2.
Marshall, C. H., K. C. Crawford, K. E. Mitchell, and D. J. Stensrud, 2003: The impact of the land surface physics in the operational NCEP Eta Model on simulating the diurnal cycle: Evaluation and testing using Oklahoma Mesonet data. Wea. Forecasting, 18, 748–768, doi:10.1175/1520-0434(2003)018<0748:TIOTLS>2.0.CO;2.
Mass, C. F., D. Ovens, K. Westrick, and B. A. Colle, 2002: Does increasing horizontal resolution produce more skillful forecasts? The results of two years of real-time numerical weather prediction over the Pacific Northwest. Bull. Amer. Meteor. Soc., 83, 407–430, doi:10.1175/1520-0477(2002)083<0407:DIHRPM>2.3.CO;2.
McCumber, M. C., and R. A. Pielke, 1981: Simulation of the effects of surface fluxes of heat and moisture in a mesoscale numerical model. J. Geophys. Res., 86, 9929–9938, doi:10.1029/JC086iC10p09929.
McInnes, K. J., 1981: Thermal conductivities of soils from dryland wheat regions of eastern Washington. M.S. thesis, Dept. of Agronomy and Soils, Washington State University, Pullman, WA, 51 pp. [Available from Owen Science Library, Washington State University, Pullman, WA 99163-3200.]
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.
Mostovoy, G. V., and V. G. Anantharaj, 2008: Observed and simulated soil moisture variability over the lower Mississippi Delta region. J. Hydrometeor., 9, 1125–1150, doi:10.1175/2008JHM999.1.
Ngan, F., H. Kim, P. Lee, B. Dornblaser, and K. Al-Wali, 2013: A study of nocturnal surface wind speed overprediction by the WRF-ARW model in southeastern Texas. J. Appl. Meteor. Climatol., 52, 2638–2653, doi:10.1175/JAMC-D-13-060.1.
Noilhan, J., and S. Planton, 1989: A simple parameterization of land surface processes for meteorological models. Mon. Wea. Rev., 117, 536–549, doi:10.1175/1520-0493(1989)117<0536:ASPOLS>2.0.CO;2.
Ookouchi, Y., M. Segal, R. C. Kessler, and R. A. Pielke, 1984: Evaluation of soil moisture effects on the generation and modification of mesoscale circulations. Mon. Wea. Rev., 112, 2281–2292, doi:10.1175/1520-0493(1984)112<2281:EOSMEO>2.0.CO;2.
Overduin, P. P., D. L. Kane, and W. K. P. van Loon, 2006: Measuring thermal conductivity in freezing and thawing soil using the soil temperature response to heating. Cold Reg. Sci. Technol., 45, 8–22, doi:10.1016/j.coldregions.2005.12.003.
Peters-Lidard, C. D., E. Blackburn, X. Liang, and E. F. Wood, 1998: The effect of soil thermal conductivity parameterization on surface energy fluxes and temperatures. J. Atmos. Sci., 55, 1209–1224, doi:10.1175/1520-0469(1998)055<1209:TEOSTC>2.0.CO;2.
Reeves, H. D., K. L. Elmore, G. S. Manikin, and D. J. Stensrud, 2011: Assessment of forecasts during persistent valley cold pools in the Bonneville basin by the North American Mesoscale Model. Wea. Forecasting, 26, 447–467, doi:10.1175/WAF-D-10-05014.1.
Rife, D. L., T. T. Warner, F. Chen, and E. G. Astling, 2002: Mechanisms for diurnal boundary layer circulations in the Great Basin Desert. Mon. Wea. Rev., 130, 921–938, doi:10.1175/1520-0493(2002)130<0921:MFDBLC>2.0.CO;2.
Rife, D. L., C. A. Davis, Y. Liu, and T. T. Warner, 2004: Predictability of low-level winds by mesoscale meteorological models. Mon. Wea. Rev., 132, 2553–2569, doi:10.1175/MWR2801.1.
Robock, A., and Coauthors, 2003: Validation of North American Land Data Assimilation Systems retrospective forcing over the southern Great Plains. J. Geophys. Res., 108, 8846, doi:10.1029/2002JD003245.
Schaefer, G. L., M. H. Cosh, and T. J. Jackson, 2007: The USDA Natural Resources Conservation Service Soil Climate Analysis Network (SCAN). J. Atmos. Oceanic Technol., 24, 2073–2077, doi:10.1175/2007JTECHA930.1.
Segal, M., J. R. Garratt, G. Kallos, and R. A. Pielke, 1989: The impact of wet soil and canopy temperatures on daytime boundary-layer growth. J. Atmos. Sci., 46, 3673–3684, doi:10.1175/1520-0469(1989)046<3673:TIOWSA>2.0.CO;2.
Seyfried, M. S., L. E. Grant, E. Du, and K. Humes, 2005: Dielectric loss and calibration of the Hydra Probe soil water sensor. Vadose Zone J., 4, 1070–1079, doi:10.2136/vzj2004.0148.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3_bw.pdf.]
Steeneveld, G. J., T. Mauritsen, E. I. F. de Bruijn, J. Vilà-Guerau de Arellano, G. Svensson, and A. A. M. Holtslag, 2008: Evaluation of limited area models for the representation of the diurnal cycle and contrasting nights in CASES99. J. Appl. Meteor. Climatol., 47, 869–887, doi:10.1175/2007JAMC1702.1.
Tapper, N. J., 1988: Some evidence for a mesoscale thermal circulation at The Salt Lake, New South Wales. Aust. Meteor. Mag., 36, 101–102.
Tarnawski, V. R., and W. H. Leong, 2000: Thermal conductivity of soils at very low moisture content and moderate temperatures. Transp. Porous Media, 41, 137–147, doi:10.1023/A:1006738727206.
Tarnawski, V. R., T. Momose, and W. H. Leong, 2009: Assessing the impact of quartz content on the prediction of soil thermal conductivity. Geotechnique, 59, 331–338, doi:10.1680/geot.2009.59.4.331.
Viterbo, P., and A. C. M. Beljaars, 1995: An improved land surface parameterization scheme in the ECMWF model and its validation. J. Climate, 8, 2716–2748, doi:10.1175/1520-0442(1995)008<2716:AILSPS>2.0.CO;2.
Wen, X., S. Lu, and J. Jin, 2012: Integrating remote sensing data with WRF for improved simulations of oasis effects on local weather processes over an arid region in northwestern China. J. Hydrometeor., 13, 573–587, doi:10.1175/JHM-D-10-05001.1.
Zhang, H., Z. Pu, and X. Zhang, 2013: Examination of errors in near-surface temperature and wind from WRF numerical simulations in regions of complex terrain. Wea. Forecasting, 28, 893–914, doi:10.1175/WAF-D-12-00109.1.
Zhou, X., and B. Geerts, 2013: The Influence of soil moisture on the planetary boundary layer and on cumulus convection over an isolated mountain. Part I: Observations. Mon. Wea. Rev., 141, 1061–1078, doi:10.1175/MWR-D-12-00150.1.
Calculated by subtracting the maximum daily (0000–0000 UTC) hourly temperature from the minimum hourly temperature at each station and then averaging for all DPG-PLAYA and DPG-SL stations.
The small (~0.01 m3 m−3) diurnal cycle in the observed soil moisture at the Dugway SCAN station may either be an upward diffusion of soil moisture during the day or instrument error since the diurnal fluctuations are in phase with the 5.1-cm soil temperatures (not shown).