Introduction
The spatial variability and temporal variability of the atmosphere within the PBL are strongly influenced by the heterogeneity of the earth's surface. Matter and energy exchange at the earth–atmosphere interface occurs nonlinearly across a continuous spectrum of scales in time and space. Nonlinear feedbacks between fixed parameters (e.g., surface topography and soil and vegetation type), slowly changing surface variables (e.g., soil wetness and vegetation greenness), and rapidly changing atmospheric conditions (e.g., air temperature and relative humidity) complicate efforts in measurement and modeling of the land surface. For meteorologists, a major difficulty lies in determining the scale(s) at which neglecting surface heterogeneity becomes significant to weather forecasts.
Observations of the energy processes at the land surface have been restricted historically to microscale or small-scale regions because of limited experimental resources; likewise, climate and mesoscale models have been restrained to large-scale aggregation of land surface properties because of limited computational resources. Large-eddy simulation models treat PBL properties explicitly but are limited by the size of the computational domain. Observations at the microscale often are used to validate “area-averaged” bulk properties from numerical models (Raupach and Finnigan 1995). Furthermore, the regional-scale impact of near-surface properties has not been adequately resolved because of the lack of meso- to synoptic-scale, long-term observations. Although field projects such as the First International Satellite Land Surface Climatology Project Field Experiment (FIFE; Sellers et al. 1988) and the Southern Great Plains Experiment—1997 (SGP-97; Twine et al. 2000) have yielded much information, these experiments were too limited in duration and areal extent to capture seasonal and regional variability.
The purpose of this paper is to document the monthly to annual, regional-scale feedbacks between the land surface and atmosphere as measured by observations. Results are presented from a newly developed network of sensors that monitors ground and surface-layer properties across a regional scale in real time. Observations were collected across all of Oklahoma at 89 sites from 1 January to 31 December 2000, every 5–30 min. This study extends the work begun by Long and Ackerman (1995) and Barnett et al. (1998) by including more surface variables and a full annual cycle of observations.
Data
Measurement systems
The Oklahoma Mesonetwork (Brock et al. 1995) is a permanent system of 115 stations designed to capture mesoscale variability that the federal network of surface stations often undersamples. The mesonet system measures a suite of meteorological and surface components across all of Oklahoma, with an average spacing interval of 30 km between sites. Each site measures solar radiation, air pressure, precipitation, wind speed and direction at 10 m, air temperature and relative humidity at 1.5 m, and bare soil and sod temperature at a 10-cm depth. The mesonet was installed during 1992 and became operational on 1 January 1994.
The Oklahoma Atmospheric Surface-Layer Instrumentation System (OASIS; Brotzge et al. 1999) enhanced 89 of the mesonet sites with new sensors to enable routine measurements of the surface energy budget. Net radiation Rnet, ground heat flux G, sensible heat flux H, and skin temperature are measured. Latent heat flux LE is estimated as the residual of the surface energy balance. In addition, soil water potential is measured at 5-, 25-, 60-, and 75-cm depths. A map of the 115 Oklahoma Mesonet and 89 OASIS sites is shown in Fig. 1.
Measurement methods
Net radiation is measured at the OASIS sites using the Kipp and Zonen, Inc., NR Lite (Brotzge and Duchon 2000). Sensible heat flux is estimated using a profile technique (Brotzge and Crawford 2000) that employs two temperature and two wind sensors and applies Monin–Obukhov similarity theory. Ground heat flux is estimated using a combination approach (Tanner 1960); the soil heat flux and soil heat storage are calculated separately. The soil heat flux is estimated using two HFT3.1 heat flux plates manufactured by Radiation and Energy Balance Systems, Inc. (REBS). The soil heat storage is estimated using two REBS platinum resistance temperature detectors, soil water potential estimated at 5 cm (using the Campbell Scientific, Inc., 229L), and knowledge of the soil properties at each site. Soil water content was calculated from the soil water potential as described by Basara and Crawford (2000). The degree of soil saturation then was estimated from the soil water content and soil porosity estimates (Rawls et al. 1983).
Data quality
All standard mesonet data were collected as described by Brock et al. (1995). All data were archived in 5-min-averaged samples, except for soil moisture data, which were collected as 30-min samples. Quality-control checks, including range testing and persistent testing, were applied to the data as described by Shafer et al. (2000); the data were also visually inspected. All OASIS data were quality assured, as described by Brotzge (2000).
From its commissioning in March of 1994 through May of 1999, the Oklahoma Mesonet collected and archived 99.9% of over 75 million possible observations (Shafer et al. 2000). The quality of the standard mesonet data has been described by Brock et al. (1995), Crawford et al. (1995), Arndt et al. (1998), and Shafer et al. (2000). Air temperature has a sensor accuracy of 0.35°C, relative humidity RH has a sensor accuracy of 3%, and wind speed estimates are accurate to within 2%. Rainfall accuracy is within 1%, and pyranometer measurements are estimated to have an accuracy of 5%. Brotzge (2000) did an extensive review of the accuracy of the mesonet flux estimates. He found that, when compared with other independent systems, net radiation and sensible heat flux were measured within approximately 5% and latent heat and ground heat fluxes were within 10%. Likewise, Basara (2000) and Basara and Crawford (2000) have investigated sensor performance of the mesonet soil moisture measurements. The soil temperature sensors used to estimate the soil moisture are calibrated to an accuracy of 0.5°C.
Climate of Oklahoma
The climate and topography of Oklahoma are characterized by its diversity. Elevations increase from 300 m above sea level in southeastern Oklahoma to over 1200 m in the northwest panhandle (Johnson and Duchon 1995). Annual rainfall patterns reflect a similar trend across the state, with a decrease from 1300 mm in the more humid southeast to less than 450 mm in the northwest. Forests dominate the eastern one-third of the state; the western one-half of the state is primarily agricultural and grassland. Known as the “tree line,” an abrupt east–west transition from forests to grasslands exists near the center of the state and is found at approximately 97.5°W longitude (Johnson and Duchon 1995). The wide range in climate and topography across the state provides a unique location for examination of land surface–atmosphere interactions.
The weather across Oklahoma during 2000 was characterized by several extreme events. Much-above-average precipitation during March, May, and June was followed by a severe statewide drought during August–September. October–December were characterized by much-above-average rainfall. Temperatures during the spring and summer months were mild, with much-above-normal temperatures during the drought period of August and September. The combined months of November and December were the coldest in the 109-yr period of record.
Temporal variability
Observations of air temperature, relative humidity, rainfall, surface fluxes, and soil moisture (degree of soil saturation) were averaged across all sites over 24-h periods to yield daily, statewide means (Figs. 2a–d). Data were averaged between 0000 and 2355 UTC. Eighty-nine sites were included in the study. Data collected from 2000 show two distinct climatological patterns. The spring and the early summer were characterized by an active weather pattern with numerous frontal systems and precipitation. In contrast, the autumn period (August and September) was very dry and sunny. The atmospheric and surface variables shown in Fig. 2 both reflect these two climatological patterns. The hottest period during the year coincided with the driest period (Fig. 2a); skies also remained mostly sunny statewide as indicated by the observed net radiation (Fig. 2b) and solar radiation (not shown). Drying of the soil occurred at all depths, with the lowest levels lagging behind the drier near-surface levels (Fig. 2c). The decreased relative humidity and increased sensible heat flux reflect the increased moisture stress (Figs. 2b,c). Bowen ratio is the ratio of sensible to latent heat flux and is a simple yet direct way to measure the effect of surface wetness upon surface fluxes. Statewide averages of Bowen ratio were estimated daily and reflect the changing states of soil moisture and vegetation (Fig. 2d).
To quantify the rate of response of the atmospheric and soil variables to rainfall, each variable was averaged over the number of days since rainfall was last measured at the site (Figs. 3a–d). Note that data plotted for day 15 include all data collected during periods of 15 days or greater without precipitation. Results indicated a steady increase in air and skin temperatures, increased sensible heat fluxes, and a rapid decline in atmospheric humidity and soil wetness (Figs. 3a–d). A general increase in shortwave radiation with increased drying shows increasingly sunny skies across Oklahoma. This trend may indicate some feedback of the lower, drier boundary layer upon cloud production.
A typical response of the land surface to synoptic-scale forcing was shown by Figs. 3a–d. A semiregular, synoptic-scale frontal passage was observed every 7–11 days as indicated by spectral analysis of the pressure data, as shown in Fig. 4c. These synoptic systems were associated with an increase in cloudiness between days 6 and 11 (Fig. 3b) and a drop in atmospheric pressure between days 10 and 11 (not shown). These trends in cloudiness and pressure were accompanied by a marked increase in surface relative humidity of over 10% (Fig. 3c).
The increased RH and decreased radiation were likely caused by the return flow of moisture ahead of synoptic frontal systems. Prior to frontal passage, cloudiness and humidity increased. A decrease in radiation reduced surface heating and soil evaporation. By day 12, radiation increased again, accompanied by increased atmospheric temperature and lowered humidity levels.
During days 6–11, the degree of soil saturation at the 60- and 75-cm depths remained nearly constant or increased (Fig. 3d), despite no rainfall in over a week. The slight increases in soil moisture most likely resulted from moisture being drawn up from lower depths. Dew and light precipitation, too light to be detected by the rain gauge network, also could have moistened the upper layers. Decreased radiation and increased atmospheric moisture reduced evapotranspiration and the need for root-zone soil moisture during this period as well. By day 14, moisture levels at both 60 and 75 cm had resumed drying.
Note that, in interpreting Figs. 3a–d, it is important to realize that the sample size decreased with each passing day of data. For example, data that were rain free for only 8 days were excluded from the diagram after day 8.
Spectral analysis of the data highlights the regularity of daily, synoptic, and seasonal variations. The diurnal cycle accounts for much of the total temporal variance of air temperature, relative humidity, wind speed, and solar radiation. The annual variance of solar radiation is dominated by the diurnal cycle because both day and nighttime observations are included in the dataset. The influence of synoptic-scale features is most evident by the atmospheric pressure variance, with a distinct period of about 10 days. This 10-day cycle also is evident from the spectral analysis of the relative humidity and wind speed data.
Long-term variability is characterized by the annual cycle and a secondary maximum at about day 40–50. The variance of sensible heat flux is found exclusively in the monthly to annual period because of the seasonal variations in land surface properties. Likewise, air temperature variability is also dominant at the longer timescales because of the annual cycle of the changing seasons. The secondary maximum of variance at the 40–50-day period results from the approximate 40–50-day period of drought during August and September. The drought represented a distinctive surface feature during 2000 and would likely be absent during a nondrought year.
Spatial variability
Given the density of the observational network across Oklahoma, site-to-site correlation coefficients were determined for all measured parameters (Figs. 5–6). For each site, the annual time series of a variable was correlated with that same variable time series from all other 88 sites. Thus, for every site, 88 correlation values were produced for each variable. The data were then plotted, putting each site at the center of the plot and displaying the 88 other site correlation coefficients in the appropriate geographic configuration. This process was repeated for each site, yielding a total 7921 correlation coefficients for each variable. Data from these 89 sites then were interpolated to a single, uniform 20 × 20 grid using linear Delaunay triangulation. Delaunay triangulation returns a set of triangles such that no data points are contained in any triangle's circumcircle. The triangulation method is based on sweepline Voronoi code by Fortune (1987). Other Delaunay-based interpolation schemes (cubic and nearest neighbor) were tested, and the results did not vary significantly. The spatial plots were summarized quantitatively as shown in Figs. 7 and 8. These figures represent linear fits to the correlation-versus-distance data that are contoured in Figs. 5 and 6. Because no stations are spaced closer than 20 km, correlation results are not plotted for distances less than 20 km (Figs. 7–8).
Spatial correlations of greater than 0.5 were observed between sites up to 100 km apart for atmospheric variables such as solar radiation, air temperature, and relative humidity (Figs. 5a–c). Solar radiation exhibited the greatest spatial coherence with little directional dependence. Air temperature and RH were better correlated southeast–northwest and south–north, respectively, along predominant surface flow patterns.
The precipitation correlation gradient also is oriented northwest–southeast and reflects the general movement of fronts across the state from northwest to southeast. However, precipitation has a much lower correlation of less than 0.3 at a radius of 100 km (Fig. 5d); as compared with other atmospheric variables, the spatial correlation of rainfall drops much more quickly with distance (Fig. 7a). This lower correlation is due to the convective nature of rainfall. Note, however, that the nature of precipitation changes dramatically with season across the Midwest. The warm season is characterized by convective rainfall, and the cool season is predominantly stratiform. This distinct change in weather ultimately affects surface heterogeneity. When the spatial correlation of precipitation is stratified by season (Fig. 8), the much higher correlation of stratiform precipitation is observed.
Local topography and vegetation modify surface wind flow at individual weather stations, creating much lower site-to-site correlations in wind speed (Fig. 5e). Nevertheless, a similar northwest–southeast gradient in wind speed correlation is observed. In addition, note that at greater distances the correlation of wind speed is similar to that of solar radiation, temperature, and humidity (Fig. 7a).
Surface parameters that vary as a function of vegetation and soil properties reflect much greater heterogeneity (Fig. 7b). For example, skin temperature (Fig. 5f) is a strong function of radiative forcing, surface cover, and vegetation and soil type. The spatial correlation of skin temperature drops to less than 0.3 at a distance of 100 km (Fig. 5f).
The influence of the surface reduces the spatial correlation of net radiation (Fig. 6a) as compared with that observed by solar radiation (Fig. 5a). The spatial correlation of ground heat flux (Fig. 6b) is weak and isotropic. Sensible heat flux also has only a weak correlation among sites up to 50 km apart (Fig. 6c). The low spatial correlation of the sensible flux likely is caused by local fetch and topographical features that affect the measurement. The problem of measuring winds is evident by the low spatial correlation of wind speeds (Fig. 5e). Note that all correlations are computed after the statewide mean has been removed.
The means and standard deviations of the spatial correlations were estimated from each variable among sites spaced less than 100 km apart (Figs. 9a,b). In addition, correlations were calculated using different averaging intervals of 15, 30, and 60 min and 24 h.
As found by Long and Ackerman (1995), spatial correlations generally improved with increased averaging interval. Results improved most for estimates of solar radiation (Fig. 9a); results are indeterminant for much lower mean correlations (Fig. 9b). As shown by Figs. 5–9, the greater the influence of the land surface is, the lower the mean spatial correlation is.
Spatial–temporal relationships during an annual cycle
Many surface parameters (such as soil wetness and vegetation greenness) change slowly with season, and such long-term changes can cause feedback to the climate system. A number of studies have demonstrated the effect of surface heterogeneity upon weather and climate through cloud formation (Zhong and Doran 1997), the development of mesoscale circulations (Mahfouf et al. 1987; Segal and Arritt 1992), and thunderstorm initiation (Rabin et al. 1990; Chang and Wetzel 1991; Clark and Arritt 1995). As a result of previous work, the authors hypothesize that organized, homogeneous patches of land surface features may have a much greater effect upon the mesoscale and synoptic-scale environment than does more random, microscale heterogeneity.
To explore this hypothesis, Hovmoeller diagrams of atmospheric and surface variables were plotted for an annual cycle. Hovmoeller diagrams are a common technique for displaying isopleths of atmospheric variation as averaged across latitudinal or longitudinal bands (Glickman 2000). All 5- and 30-min data were averaged over 24-h periods to form single daily means for each variable. The 24-h period was defined to be from 0000 to 2355 UTC. Because most surface parameters vary greatest from east to west, data from the 89 OASIS sites were collapsed onto a single east–west line. A five-point triangular filter was applied to the data to smooth the spatial inhomogeneities. Data were computed for each day of 2000. A second five-point triangular filter was applied in the same manner to smooth data temporally. Note that data from the Oklahoma Panhandle were excluded from this study because of the limited geographical area represented.
Atmospheric and vegetation data
Daily rainfall totals across the state (Fig. 10a) indicate three major rainy periods during the year. The first event was focused primarily in western Oklahoma during days of year (DOY) 100–125. The second and third events were statewide and occurred during DOY 150–200 and DOY 250–325. Little to no rainfall was reported statewide during DOY 200–250.
Biweekly normalized-difference vegetation index (NDVI) data were extracted for each mesonet site during 2000. NDVI data are estimated routinely from satellite imagery and provide a near-real-time measure of vegetation “greenness.” Daily values of NDVI were linearly interpolated from biweekly estimates and are plotted in Fig. 10b. The tree line is observed easily at −97.5° longitude and is present throughout much of the year. Much “greener” vegetation is observed to the east of the line, with much more sparse vegetation observed to the west. The primary area of vegetation is observed throughout eastern Oklahoma and extends from early spring (DOY 75) to late summer (DOY 200). A secondary maximum is observed in western Oklahoma between −98.5° and −99.5° longitude between DOY 75–150. This region is the winter wheat belt found primarily in northwestern and west–central Oklahoma. The crop is harvested typically about 1 June, but the wheat reaches senescence during early May.
The seasonal cycles in rainfall and vegetation affect many of the observed surface and atmospheric patterns. For easier interpretation, site values were subtracted from a statewide mean to produce daily anomalies for each variable. Daily anomalies are plotted in Fig. 11 for air temperature, relative humidity, shortwave radiation, and wind speed.
Anomalies of solar radiation showed a relative increase in cloudiness to the east across Oklahoma during much of the year (Fig. 11a). This likely results from the proximity to return southeasterly flow from the Gulf of Mexico. In addition, more abundant vegetation across the eastern half of Oklahoma leads to higher evapotranspiration (ET) and soil moisture values and results in greater LE flux from the surface. This increased surface wetness lowers lifting condensation level heights and increases the likelihood of cloud formation.
Temperature anomalies are distinctly different between eastern and western Oklahoma (Fig. 11b). During the winter season, temperatures are cooler (warmer) in western (eastern) Oklahoma than the statewide mean; during the summer season, temperatures are warmer (cooler) in western (eastern) Oklahoma. Much higher elevations in western Oklahoma account for some of this temperature difference. However, sparse vegetation across much of western Oklahoma leads to greater radiational cooling, whereas across much of eastern Oklahoma more dense vegetation tends to moderate temperatures (Fig. 10b). Eastern Oklahoma is also much cloudier (Fig. 11a), keeping temperatures cooler during the summer and warmer during the winter season, whereas the much clearer skies across western portions of the state would tend to enhance seasonal changes. Thus, topography, vegetation, and radiational forcing each appear to reinforce the seasonal temperature pattern. Radiational forcing appears to be the dominant factor, with cloud anomalies coincident with temperature anomalies. A positive feedback between clouds and surface temperature is found during the winter months, and a negative feedback is observed during the summer months.
The relative humidity (Fig. 11c) and dewpoint (not shown) reflect a similar gradient as temperature, with values increasing from west to east. However, the moisture gradient remains throughout the year. During the early spring months, average to above-average moisture is found as far west as −99.0° longitude but retracts eastward as the summer progresses. It is surmised that the strong dynamics during the spring and early summer pull return flow moisture from the Gulf of Mexico much farther west; as the dominant flow shifts north, synoptic systems become weaker and return moisture is no longer advected as far west. By midsummer (DOY 180), the gradient has aligned itself along the tree line in central Oklahoma.
Wind speeds across Oklahoma remain steady throughout the year and increase from east to west (Fig. 11d). As with temperature and moisture, the strongest gradient is aligned north–south along the tree line in central Oklahoma. The more dense vegetation across eastern Oklahoma is expected to decrease wind speeds. The maximum wind speeds in Oklahoma are found between −98.5° and −99.25° longitude. These maximum wind speeds have been found to be coincident with the greatest increases in slope (K. Crawford 2000, personal communication).
Radiative and surface fluxes
Net radiation, sensible, and ground heat fluxes were estimated directly from the 89 sites enhanced by OASIS. Results from the spatial analysis reduced the number of sites used for each flux diagram. Those sites with a maximum correlation of less than 0.7 were not included in their respective plotting. These sites were assumed to be dominated by microscale effects and thus were nonrepresentative of the regional area. When the analysis was not limited to sites with a reasonable correlation (in this case, >0.7), the Hovmoeller diagrams were extremely noisy and regional patterns could not be observed.
Daily estimates of net radiation anomalies yielded few spatial or seasonal trends (Fig. 12a). In general, nocturnal radiative cooling offset daytime heating variations. Twenty-four-hour means revealed limited dependence upon surface features. Eighty-six sites were used to produce Fig. 12a.
Measurements of ground heat flux (Fig. 12b) are sensitive to the microscale conditions surrounding the sensors. Data from 19 sites were excluded because of missing or bad data, sensor malfunctions, or nonrepresentative, microscale effects. Nevertheless, results from the spatial analysis presented in section 4 revealed that correlations of greater than 0.6 extended as far as 50–100 km from a site (Fig. 6b). Although anomalies of ground flux indicated a few variations, 24-h averages generally summed to zero. Only the dry period of DOY 200–250 showed a significant homogeneous pattern statewide.
Sensible heat fluxes (Fig. 12c) were affected by radiative forcing and atmospheric conditions as well as by surface wetness and roughness. Flux estimates are easily influenced by nearby fetch (Brotzge and Crawford 2000), and only 48 of the 89 sites were found to be spatially representative enough to be included in this study. Nevertheless, an examination of the annual variations in flux anomalies across the state revealed a strong dependence upon radiative forcing, rainfall events, and vegetation. The flux anomalies increased from east to west, similar to the pattern shown by solar radiation, humidity, and wind speed. Fluxes generally west of the tree line were 20–40 W m−2 greater than average; areas to the east were up to 40 W m−2 less than the statewide mean. The statewide pattern remained constant during the year and was particularly enhanced during the warm season. The east–west gradient was reduced during rainy periods such as during DOY 150–175 and DOY 250–300.
An exception to the east–west increase in flux is found along −96.5° longitude. This higher-than-average flux is found east of the tree line. An examination of the NDVI showed a minimum along this same longitude, indicating the response of the fluxes to changing surface conditions.
Latent heat fluxes were calculated as the residual of the surface energy budget (LE = Rnet − G − H). The same 48 sites were included in calculation of LE as were used in estimating H. Daily anomalies of latent heat flux (Fig. 12d) reflect an inverse relationship to sensible heat flux. Values increase from west to east, particularly during the warm season. The highest values of LE are observed east of the tree line.
Daily anomalies of skin temperature (not shown) strongly reflected air temperature trends. However, the skin temperature estimates were directly modified by the vegetation state and, in some cases, the vegetation appeared to have a stronger influence than did the radiative forcing. For instance, the wheat fields between −98.75° and −99.25° (as shown in Fig. 10b) have a daily cooling of up to 1°C despite above-average radiative heating. The effect upon surface temperatures decreases as the wheat reaches maturity and is harvested.
Soil moisture
The degree of soil saturation was estimated at the 89 sites at the four depths of 5, 25, 60, and 75 cm (Figs. 13a–d). The soil moisture data from 2000 had not been quality controlled by the Oklahoma Climatological Survey at the time of this study; limited quality control and appropriate calibrations were applied by the authors. Soil moisture estimates have been found to be very sensitive to calibration errors as well (Basara et al. 2000). Confidence in the measurements is increased when multiple depths (i.e., multiple sensors) show similar patterns statewide.
The degrees of soil saturation are above (below) the statewide average east (west) of approximately −98° longitude. Estimates remain nearly constant throughout the year, with some oscillations because of rainfall events. Statewide variability decreases with depth, particularly at the 75-cm level. Values were most homogeneous during the dry period of DOY 200–250.
Conclusions
Annual cycles of atmospheric and near-surface variables were shown to vary significantly both spatially and temporally. Monthly-to-seasonal variations in atmospheric variables are observed statewide. Daily estimates of Bowen ratio and soil moisture reflect the long-term, large-scale atmospheric forcing. These surface parameters provide positive feedback to the climate system, perhaps prolonging dry and wet periods.
The principal results of our study are as follows:
During periods of drying, temporal changes of atmospheric and surface parameters are not steady state. The time between most large-scale systems was between 7 and 12 days. These synoptic systems increased atmospheric humidity and cloudiness and reduced evapotranspiration, which maintained soil wetness levels longer. Only after 12 or more days of precipitation-free weather was further drying expected to continue.
Atmospheric and near-surface variables are most homogeneous during very wet or very dry periods. During the transitional phase from a wet surface to a dry surface, variations in topography and vegetation likely cause patches of wet and dry soils and high and low areas of evapotranspiration. Once the moisture in the soils has reached a critical wilting point, evapotranspiration is minimal and variability caused by vegetation diminishes.
The spatial correlation patterns of shortwave and net radiations are generally isotropic. However, the spatial correlation patterns of air temperature, relative humidity, wind speed, and precipitation are anisotropic. Each favor a northwest–southeast correlation gradient, similar to a typical spring-to-summer regime.
The greater the influence of surface topography and fetch is, the lower the spatial correlation of a variable is. The mean spatial correlation of wind speed and sensible heat flux was much lower than that of other variables; the spatial correlation of sensible heat flux also decreased quickly with distance because of natural soil and vegetation heterogeneity. The spatial correlation of ground heat flux was larger than expected and also favored a northwest–southeast pattern.
Spatial correlation patterns may strengthen or weaken with season. For example, the spatial correlation of rainfall was much greater during periods of stratiform precipitation than during periods of convective activity (Fig. 8).
Seasonal variability in radiative forcing remains nearly steady as a function of atmospheric moisture, topography, and vegetation (Figs. 10–11). The atmospheric moisture reflects general flow patterns across the state, with a southeasterly flow across much of the state during the spring-to-summer period and drier, southwesterly winds across western Oklahoma during the late summer-to-autumn period. Wind speeds generally increase from east to west across the state as a function of surface roughness via topography and vegetation.
Sensible and latent heat fluxes vary seasonally with surface wetness. In general, sensible fluxes dominated across western Oklahoma while latent heat flux was greater across eastern Oklahoma. Both varied as a function of rainfall, soil wetness, and vegetation state.
Soil moisture generally decreased from east to west across the state. Most areas of high and low anomalies of soil moisture remained similar with depth. As with most other observations, estimates of soil moisture became more homogeneous during the drought period of DOY 200–250.
The observations archived from the OASIS program during 2000 document seasonal, regional-scale changes and interactions within the atmospheric, surface, and soil layers. Because of the scale at which these data have been collected, this work could help to improve and/or to validate climate model parameterizations currently in use, which should in turn help to improve the overall accuracy of climate models. The same holds true for forecast models, with the added benefit that they have resolution on the order of the spacing of Oklahoma Mesonet stations. Therefore, spatial and temporal correlations from forecast models can be compared with the correlations found using Oklahoma Mesonet data. However, as noted earlier, the results from this paper may only be valid in Oklahoma, and results are expected to differ across other regions and during other periods of study.
Acknowledgments
This research was made possible, in part, by NSF Grant 125-5645 and the Williams Energy Corporation Grant 125-4909. The authors thank Dr. Claude Duchon, Dr. Jeff Basara, and the technical support staff of the Oklahoma Climatological Survey for their professional assistance in maintenance, ingest, and quality control of the mesonet data. A special thanks is given to the taxpayers of Oklahoma for their continued support and funding of the Oklahoma Mesonet Project. Last, this work was supported in part by the ARM Program of the U.S. Department of Energy through Battelle PNNL Contract 144880-A-Q1 and DOE Subcontract 354047-AQ5 to the Cooperative Institute for Mesoscale Meteorological Studies.
REFERENCES
Arndt, D. S., M. A. Shafer, S. E. Fredrickson, and J. P. Bostic. 1998. Quality assurance at the Oklahoma Mesonet: A systems viewpoint. Preprints, 10th Symp. on Meteorological Observations and Instrumentation, Phoenix, AZ, Amer. Meteor. Soc., 349–354.
Barnett, T. P., J. Ritchie, J. Foat, and G. Stokes. 1998. On the space–time scales of the surface solar radiation field. J. Climate 11:88–96.
Basara, J. B. 2000. The value of point-scale measurements of soil moisture in planetary boundary layer simulations. Ph.D. dissertation, School of Meteorology, University of Oklahoma, Norman, OK, 225 pp.
Basara, J. B. and T. M. Crawford. 2000. Improved installation procedures for deep-layer soil moisture measurements. J. Atmos. Oceanic Technol. 17:879–884.
Basara, J. B., K. S. Humes, and K. C. Crawford. 2000. The spatial and temporal variability of soil moisture conditions at the Norman mesonet site: Comparison with near-surface atmospheric conditions. Preprints, 15th Conf. on Hydrology, Long Beach, CA, Amer. Meteor. Soc., 111–114.
Brock, F. V., K. C. Crawford, R. L. Elliott, G. W. Cuperus, S. J. Stadler, H. L. Johnson, and M. D. Eilts. 1995. The Oklahoma Mesonet: A technical overview. J. Atmos. Oceanic Technol. 12:5–19.
Brotzge, J. A. 2000. Closure of the surface energy budget. Ph.D. dissertation, School of Meteorology, University of Oklahoma, Norman, OK, 208 pp.
Brotzge, J. A. and K. C. Crawford. 2000. Estimating sensible heat flux from the Oklahoma Mesonet. J. Appl. Meteor. 39:102–116.
Brotzge, J. A. and C. E. Duchon. 2000. A field comparison among a domeless net radiometer, two four-component net radiometers, and a domed net radiometer. J. Atmos. Oceanic Technol. 17:1569–1582.
Brotzge, J. A., S. J. Richardson, K. C. Crawford, T. W. Horst, F. V. Brock, K. S. Humes, Z. Sorbjan, and R. L. Elliott. 1999. The Oklahoma Atmospheric Surface-layer Instrumentation System (OASIS) Project. Preprints, 13th Conf. on Boundary Layers and Turbulence, Dallas, TX, Amer. Meteor. Soc., 612–615.
Chang, J-T. and P. J. Wetzel. 1991. Effects of spatial variations of soil moisture and vegetation on the evolution of a prestorm environment: A numerical case study. Mon. Wea. Rev. 119:1368–1390.
Clark, C. A. and R. W. Arritt. 1995. Numerical simulations of the effect of soil moisture and vegetation cover on the development of deep convection. J. Appl. Meteor. 34:2029–2045.
Crawford, K. C., D. S. Arndt, D. J. Shellberg, and B. M. Brus. 1995. A comparison of differences in meteorological measurements made by the Oklahoma Mesonetwork at two co-located ASOS sites. Preprints, 11th Int. Conf. on Interactive Information and Processing Systems for Meteorology, Oceanography, and Hydrology, Dallas, TX, Amer. Meteor. Soc., 299–303.
Fortune, S. 1987. A sweepline algorithm for Voronoi diagrams. Algorithmica 2:153–174.
Glickman, T. S. Ed.,. 2000. Glossary of Meteorology. 2d ed. Amer. Meteor. Soc., 855 pp.
Johnson, H. L. and C. E. Duchon. 1995. Atlas of Oklahoma Climate. University of Oklahoma Press, 136 pp.
Long, C. N. and T. P. Ackerman. 1995. Surface measurements of solar irradiance: A study of the spatial correlation between simultaneous measurements at separated sites. J. Appl. Meteor. 34:1039–1046.
Mahfouf, J-F., E. Richard, and P. Mascart. 1987. The influence of soil and vegetation on the development of mesoscale circulations. J. Climate Appl. Meteor. 26:1483–1495.
Rabin, R. M., S. Stadler, P. J. Wetzel, D. J. Stensrud, and M. Gregory. 1990. Observed effects of landscape variability on convective clouds. Bull. Amer. Meteor. Soc. 71:272–280.
Raupach, M. R. and J. J. Finnigan. 1995. Scale issues in boundary-layer meteorology: Surface energy balances in heterogeneous terrain. Hydrol. Processes 9:589–612.
Rawls, W. J., D. L. Brakensiek, and N. Miller. 1983. Green–Ampt infiltration parameters from soils data. J. Hydraul. Eng. 109:62–70.
Segal, M. and R. W. Arritt. 1992. Nonclassical mesoscale circulations caused by surface sensible heat-flux gradients. Bull. Amer. Meteor. Soc. 73:1593–1604.
Sellers, P. J., F. G. Hall, G. Asrar, D. E. Strebel, and R. E. Murphy. 1988. The First ISLSCP Field Experiment (FIFE). Bull. Amer. Meteor. Soc. 69:22–27.
Shafer, M. A., C. A. Fiebrich, D. S. Arndt, S. E. Fridrickson, and T. W. Hughes. 2000. Quality assurance procedures in the Oklahoma Mesonetwork. J. Atmos. Oceanic Technol. 17:474–494.
Tanner, C. B. 1960. Energy balance approach to evapotranspiration from crops. Soil Sci. Soc. Amer. Proc. 24:1–9.
Twine, T. E., W. P. Kustas, J. M. Norman, D. R. Cook, P. R. Houser, T. P. Meyers, J. H. Prueger, and P. J. Starks. 2000. Correcting eddy covariance flux underestimates over a grassland. Agric. For. Meteor. 103:279–300.
Zhong, S. and J. C. Doran. 1997. A study of the effects of spatially varying fluxes on cloud formation and boundary layer properties using data from the Southern Great Plains Cloud and Radiation Testbed. J. Climate 10:327–341.