The Oklahoma City Micronet
Abstract
The Oklahoma City Micronet (OKCNET) is an operational surface observing network designed to improve atmospheric monitoring across the Oklahoma City, Oklahoma, metropolitan area. The 40 station network consists of 4 Oklahoma Mesonet stations and 36 micronet stations mounted on traffic signals at an average station spacing of approximately 3 km. Using several technical innovations as well as existing infrastructure in Oklahoma City, data are collected and quality assured in near real-time at an interval of 1 min for the traffic signal sites and 5 min for the Mesonet sites. Because OKCNET also spans a land use gradient from rural to urban, the spatial and temporal densities of OKCNET observations have shed new insights on atmospheric processes (e.g. the urban heat island, severe thunderstorm evolution) across the Oklahoma City metropolitan area. Copyright © 2010 Royal Meteorological Society
1. Introduction
Founded in 1889, Oklahoma City (OKC), USA is a rapidly growing urban area in the south-central region of the United States. Currently, the areal extent of OKC spans nearly 1610 km2, placing OKC within the top 10 largest cities by land area in the United States. However, only about 630 km2 of OKC was defined as urbanized in 2009 (United States Census Bureau, 2009). Even so, this urbanized area is expanding rapidly.
Since 2000, the population of the OKC metropolitan area (i.e. the geographic area that consists of the City of Oklahoma City and its surrounding communities) has increased by an estimated 10.1% to more than 1.2 million residents (United States Census Bureau, 2009). In 2008, the City of Oklahoma City ranked as the 31st most populated urban area in the United States.
Between 28 June and 31 July 2003, the Joint Urban 2003 (JU2003) field experiment was conducted in OKC to collect data for the improvement and validation of numerical models that simulate dispersion within urban environments (Allwine et al., 2004). During the 6 week field experiment, intensive observing periods occurred using a temporary array of various instrument systems deployed in and around the central business district (CBD) of OKC (Allwine and Flaherty, 2006).
Because of the complex nature of atmospheric processes within the urban boundary layer, field experiments such as JU2003 have been critical to the advancement of research in urban meteorology and urban climatology. Other field programmes have contributed greatly to the understanding of the impacts of urban areas on atmospheric processes, for example see reviews in Arnfield (2003) and Grimmond (2006). Unfortunately, logistical constraints often mean that after researchers have completed the measurement goals of the particular field experiment, the instrumentation typically is decommissioned. Such was the case with JU2003 in Oklahoma City.
At the same time, most real-time, continuous, quality-assured atmospheric observations in North America currently are collected at city or town airports, not within the urbanized centre of a city. Thus, most surface atmosphere observing sites are separated from areas with the densest populations. Recent publications have shown that urbanization has increased and will continue to increase worldwide (United Nations Human Settlements Program, 1997; Dabberdt et al., 2000; United Nations, 2003) and, as such, a growing need exists for improved urban atmospheric observations for a variety of applications, including public health and safety.
In Oklahoma and prior to JU2003, the locations of automated surface observing sites were consistent with most North American cities (i.e. away from or on the periphery of the urban areas). These sites included those of the Oklahoma Mesonet (Brock et al., 1995; McPherson et al., 2007), which has collected research-quality observations of soil and atmospheric variables statewide in near real-time since 1994. The Oklahoma Mesonet has served as the foundation for mesoscale monitoring of surface-layer processes across the state and has been instrumental in providing critical weather and climate information for a variety of end-users. Yet, Oklahoma Mesonet stations were located away from urban areas to maintain the site representativeness requirements of the network.
The infrastructure of the Oklahoma Mesonet and the success of JU2003, including various collaborations with municipal officials across Oklahoma City, provided the opportunity to develop an operational urban weather network with high spatial and temporal resolutions and a well-documented quality assurance programme. With the assistance of the City of Oklahoma City, the Oklahoma City Micronet (OKCNET) was deployed in 2008 as a dense network of robust, automated weather stations designed to collect and disseminate high-quality observations of atmospheric conditions in near real-time across OKC. OKCNET was funded by the Oklahoma State Regents for Higher Education, Office of the Vice President for Research of the University of Oklahoma, the Oklahoma Mesonet, and the Oklahoma Climatological Survey (OCS), with in-kind contributions from the City of Oklahoma City. The project included the deployment of three new Oklahoma Mesonet sites within OKC and the installation of 36 stations mounted on traffic signals. OKCNET began operations on 1 June 2008 and was officially commissioned on 8 November 2008.
The primary focus of this article is to provide an overview of the technical design of OKCNET and the associated innovations that were needed to complete the network. In addition, specific examples are provided that demonstrate the capabilities of OKCNET to observe atmospheric processes that are of critical importance to the scientific community and local end users of the data.
2. Station design
2.1. Oklahoma Mesonet stations
As part of a joint effort between the OKCNET project and the Oklahoma Mesonet (McPherson et al., 2007), three new Oklahoma Mesonet sites were installed within OKC in early 2007 (Figure 1): Oklahoma City North (OKCN), Oklahoma City West (OKCW), and Oklahoma City East (OKCE). A fourth site, close to the rural town of Spencer (SPEN) but within the OKC city limits, was installed in 1994 as one of the original Oklahoma Mesonet sites. Each of these stations is located within a fenced 100 m2 plot of land and measures more than 20 environmental variables, including wind at 2 and 10 m, air temperature at 1.5 and 9 m, relative humidity, rainfall, pressure, solar radiation, and soil temperature and moisture at various depths (McPherson et al., 2007). All sensors are mounted on or near a 10 m tower supported by three guy wires and the sites operate via solar power. The four Oklahoma Mesonet sites in OKC transmit data to OCS every minute via 900 MHz spread-spectrum communications for ingest into the Mesonet processing system. All communications pathways from the station to the central computers are two-way, enabling staff at OCS to set clocks, download datalogger programs and request missed or corrupted observations.
The location of OKCNET stations on 1 October 2009. 
Traffic Signal Stations; 
Oklahoma Mesonet Station
2.2. Traffic signal stations
The OKCNET developers were challenged to deploy sensors throughout OKC while facing several physical and logistical constraints. Standard Oklahoma Mesonet sites were not a feasible option for most locations, especially within the central business district, because of siting constraints and station cost. Sensor packages needed to be located within the urban core and throughout the residential neighbourhoods and industrial zones with minimum risk of vandalism. To conduct operations similar to those of the Oklahoma Mesonet, two-way communications were needed between the station and the central processing facility. Design solutions to these challenges are detailed below.
2.2.1. Communications
The City of Oklahoma City provided an essential resource for the development of OKCNET: one of the largest municipal IEEE 802.11 Wi-Fi (TM) mesh networks in the world. The network, called OKC WIFI, was designed for use by public safety and OKC operations. The restricted-use network includes wireless access points manufactured by Tropos Networks, Inc., mounted on traffic signals across OKC and provides coverage over 1440 km2.
2.2.2. Measurement technology
-
a module for measuring temperature, humidity and pressure (the PTU module defined by Vaisala);
-
a WINDCAP® sensor for measuring wind speed and direction, and,
-
a RAINCAP® sensor for measuring precipitation.
The WXT510 sensor also included a heating unit to reduce the accumulation of frozen precipitation on the sensor during winter weather conditions.
Along with the WXT510 for meteorological observations, a Campbell Scientific Inc., model CR200 data logger was used to collect, store and process data at each site. Because the logger used serial communications and the Tropos node was based on internet protocol (IP), a Lantronix UDS1100 was used to communicate with the CR200 via IP. Additionally, a PowerDsine Power Over Ethernet (POE) unit separated the power and communications from each ethernet line. It also converted power from 48 to 18 V to charge the station battery, which provided temporary power during the loss of POE. Finally, a 7.5 Ah, 12 V battery was installed as a secondary power supply.
Power consumption with and without the WXT510 heating element active was a critical constraint in the traffic signal station design. Through laboratory testing, all station components (i.e. WXT510, UDS1100, CR200, POE and battery) consumed 2.5 W of power (194 mA at 12.86 V) when the WXT510 heater was not in use. With an average POE power supply of 7 W from the Tropos unit, a surplus of 4.5 W remained to charge the station's battery. When the WXT510 heating element was operating at full capacity, the sensor's power consumption increased an additional 13.2 W, requiring a total of 15.7 W for station operation. Hence, when the heating element was active, the battery compensated for the power deficit of 8.7 W.
2.2.3. Additional components
The main exterior components of each traffic signal site were comprised aluminium and stainless steel to provide durability and overall aesthetics. The logger/battery enclosures were cast aluminium traffic cabinets manufactured by Pelco Inc., and were similar to hardware already used by the City of Oklahoma City. The WXT510 sensor connected to the top of the enclosure through a threaded aluminium pipe. Once assembled, the design allowed for quick installation or replacement of units as well as meeting the aesthetic requirements requested by city officials.
Stainless steel mounting brackets, clamps and ties secured the hardware and cables to the traffic pole and provided the low visual profile requested by the City of Oklahoma City (Figure 2). Once each traffic signal station was securely mounted, a single ethernet cable led from the Tropos unit to the enclosure, supplying both power and communications.
The KCB103 site mounted to the top of a traffic signal in Oklahoma City and connected to a Tropos access point. This figure is available in colour online at wileyonlinelibrary.com/journal/met
3. Network design
3.1. Site selection
The Oklahoma Mesonet sites deployed in OKC (OKCW, OKCN, OKCE, SPEN) followed that network's standard protocols for siting and installation. In February 2007, OKCN was installed 11 km north of the central business district; in April 2007, OKCW and OKCE were deployed 6 km west and 6 km east of the central business district, respectively. The sites were established to provide relevant observations of soil and atmospheric variables close to the CBD while still sampling the mesoscale environment.
Site selection for OKCNET traffic signal stations was constrained to locations with OKC WIFI access points. To make provision for various atmospheric scales of motion throughout the metropolitan area, OKCNET sites were installed at an increased spatial density in the CBD as compared to the suburban and rural areas (Figure 1). Sites were selected to sample the atmosphere in a location representative of the surrounding area (e.g. Oke, 2006a). The resultant design provided the opportunity to quantify the transition of the atmosphere from rural to urban. The installation of OKCNET stations at 9 m yields sites within the CBD that are deeply embedded within the urban canopy layer with reduced overall exposure due to the influence of nearby buildings. Conversely, the sites on the periphery of OKCNET observe conditions near the top or slightly above the urban canopy layer with much greater overall site exposure. Such differences impact not only surface winds measured at each site but also temperature, whereby vertical conditions (i.e. 2 vs 9 m) are nearly isothermal in the urban canopy layer within the CBD (Basara et al., 2009b) while vertical temperature profiles yield increasing gradients away from the urban centre.
Other considerations for station siting included providing increased weather measurements within different communities of Oklahoma City, deploying stations in the watersheds most greatly affected by heavy rainfall events, and obtaining temperature, wind and precipitation observations at critical roadway locations for winter weather decision making. Special attention was also given to installing stations near important OKC community centres or landmarks including schools, major airports (e.g. Will Rogers World Airport, Tinker Air Force Base), entertainment complexes (e.g. Bricktown, the Ford Center), shopping malls (e.g. Quail Springs Mall, Penn Square Mall), and university campuses (e.g. University of Oklahoma Health Sciences Center, Oklahoma State University-OKC). For site maintenance safety, locations with overhead power lines were avoided.
In December 2007, an initial traffic signal station was installed in downtown OKC for testing purposes. Once completed, 35 additional OKCNET traffic signal stations were installed during May 2008. The average spacing of the 40 stations of OKCNET (i.e. traffic signal and Mesonet stations) is 3.3 km.
3.2. Site classification
Stewart and Oke (2009) reviewed over 180 urban heat island studies published between 1950 and 2007 and found that one-third of the articles failed to provide either a qualitative or quantitative description of the sites used: the remaining two-thirds included only qualitative site descriptions. Further, Oke (2006) noted that urban network operators should develop standardized site information so that data users would be able to understand and appreciate the nuances of each station without having to visit each site. While no set standard exists, four published methodologies previously have been used to classify urban meteorological stations (Auer, 1978; Ellefsen, 1990–1991; Oke, 2004; Stewart and Oke, 2009). Thus, to meet the need for increased metadata for OKCNET, Schroeder and Basara (2010) analyzed and classified each OKCNET site according to the four published classification schemes. Table I displays a modified summary of the Schroeder and Basara (2010) analysis.
| Site | Stewart and Oke (2009) | Oke (2004) | Auer (1978) | Ellefsen (1990–1991) |
|---|---|---|---|---|
| KCB101 | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KCB102 | Old Core | UCZ2 | C1 | Dc1 |
| KCB103 | Modern Core | UCZ1 | C1 | Dc1 |
| KCB104 | Modern Core | UCZ1 | C1 | Dc1 |
| KCB105 | Modern Core | UCZ1 | C1 | Dc1 |
| KCB106 | Modern Core | UCZ1 | C1 | Dc1 |
| KCB107 | Old Core | UCZ2 | I2 | Dc4 |
| KCB108 | Modern Core | UCZ1 | C1 | Dc1 |
| KCB109 | Extensive Lowrise | UCZ2 | I1 | Dc1 |
| KCB110 | Open Grounds | UCZ6 | I2, A3 | Do6 |
| KSW101 | Grassland | UCZ7 | A3 | N/A |
| KSW102 | Open Grounds | UCZ6 | A1 | Do6 |
| KSW103 | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KSW104 | Open Grounds | UCZ6 | A1 | Do4, Do6 |
| KSW105 | Dispersed Settlement | UCZ4, UCZ7 | I2, C1 | Do4 |
| KSW106a | Extensive Lowrise/House and Garden | UCZ4, UCZ5 | I2, A3 | Do3 |
| KSW107 | Extensive Lowrise | UCZ4 | I2, R4 | Do4 |
| KSW108 | Open Grounds | UCZ6 | A1 | Do6 |
| KSW109 | Extensive Lowrise | UCZ5 | I2, R4 | Do4 |
| KSW110 | Dispersed Settlement | UCZ7 | A3 | N/A |
| KSW111 | House and Garden/Open Grounds | UCZ5, UCZ6 | R1, A1 | Do3, Do6 |
| KSW112 | House and Garden/Open Grounds | UCZ5, UCZ6 | R1, A1 | Do3, Do6 |
| KSE101 | Extensive Lowrise/Open Grounds | UCZ4, UCZ6 | I2, R4 | Do4 |
| KSE102 | Dispersed Settlement | UCZ7 | R4 | Do3 |
| KNW101a | Extensive Lowrise | UCZ4 | I2, R4 | Do5 |
| KNW102 | Dispersed Settlement | UCZ7 | I2, A3 | Do4 |
| KNW103 | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KNW104 | Dispersed Settlement | UCZ7 | R4 | Do3 |
| KNW105 | Extensive Lowrise | UCZ4, UCZ6 | I2, R4 | Do5 |
| KNW106 | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KNW107 | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KNW108 | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KNW201 | Extensive Lowrise/Open Grounds | UCZ4/UCZ6 | I2, R4 | Do1 |
| KNW202 | Extensive Lowrise | UCZ4 | I2 | Do5 |
| KNE101 | Forest | UCZ7 | A4 | N/A |
| KNE102a | Extensive Lowrise | UCZ4 | I2 | Do4 |
| KNE103 | Dispersed Settlement | UCZ7 | I2, A3 | Do4 |
| KNE104 | Extensive Lowrise | UCZ4 | I2 | Do5 |
| KNE105 | Extensive Lowrise | UCZ4 | NA | NA |
| KNE202 | Open Grounds | UCZ6 | A1 | Do6 |
| OKCN | Grassland | UCZ7 | A1 | N/A |
| OKCE | Dispersed Settlement | UCZ6 | A1 | Do4 |
| OKCW | Open Grounds | UCZ7 | A1 | Do4 |
| SPEN | Grassland | UCZ7 | A3 | N/A |
- a Site was moved or decommissioned.
During the installation of the traffic signal stations, eight digital panoramic photographs were taken 2 m north of each site at the height of the WXT510 sensor, representing the view toward the four cardinal and four ordinal directions (in some cases, the pictures were taken from another side of the site due to logistical limitations). The photos and other detailed site information (e.g. latitude, longitude, elevation) formed the foundational metadata for each site. Supplemental site photos were collected during subsequent site visitations.
4. Network operations
Through a combination of sensor calibration, routine maintenance and both automated and manual quality assurance (QA) processes, OKCNET personnel strive to provide the highest quality observations to users in near real-time and as complete a data archive as possible.
4.1. Sensor calibration
McPherson et al. (2007) detailed the sensor calibration procedures for the Oklahoma Mesonet. As for the traffic signal stations, the Vaisala WXT510 sensor was too large for extended calibration in the environmental chambers of the Fred V. Brock Standards Laboratory of the Oklahoma Mesonet. Thus, in June 2006, the OKCNET inter-comparison facility was erected in a field in Norman, Oklahoma, USA, where other atmospheric instruments were tested and calibrated (Basara et al., 2009). The inter-comparison facility was outfitted such that up to 33 WXT510 sensors could be mounted with laboratory-calibrated reference instruments, including a Vaisala HMP45C, a Thermometrics Fastherm, an R. M. Young cup anemometer, a two-dimensional sonic anemometer manufactured by Vaisala, and four WaterLog H-340 tipping bucket rain gauges (Basara et al., 2009).
From 1 July 2006 to 30 September 2007, personnel conducted a systematic evaluation of the WXT510 sensors at the OKCNET inter-comparison facility. Of the variables collected, the only measurement to demonstrate a bias was rainfall. Hence, Basara et al. (2009) developed a bias correction that was applied to OKCNET rainfall measurements. Table II displays the calibration criteria that all WXT510 sensors must pass prior to deployment in OKCNET.
| Variable | Range | Resolution | Accuracy | Verification test | Exp. field accuracy |
|---|---|---|---|---|---|
| Air temperature | − 52 °C to 60 °C | 0.1 °C | 0.3 °C | ± 0.75 °C, root mean squared error (RMS) ≤0.35 °C | Up to 1 °C in light winds and strong radiation |
| Pressure | 600–1100 hPa | 1 hPa (at − 52 to 60 °C) | 0.5 hPa (at 0–30 °C) | ± 1 hPa, RMS ≤0.75 hPa | Same as verification test. |
| Relative humidity | 0–100% | 0.1% | 3% (at 0–90%) 5% (at 90–100%) | ± 4%, RMS ≤2% | Same as verification test. |
| Wind speed | 0–60 m s−1 | 0.1 m s−1 | 0.3 m s−1 or 2% (whichever is greater) | ± 1 m s−1 from 0 to 10 m s−1 RMS ≤0.50 m s−1 | Same as verification test. |
| Wind direction | 0–360° | 1° | 2° | ± 20° | Same as verification test. |
| Rainfall | NA | 0.01 mm | Within 5% | Rain detected on rain event days | Within 15% with the correction algorithm applied. |
4.2. Maintenance
As needed, an electronics technician performs a routine maintenance pass at each OKCNET station and could replace sensors or equipment (specifically the PTU). Because the sites include no moving parts and do not require modifications to nearby vegetation, the temporal period needed between routine site visits is driven by sensor calibration. As such, each site is visited within 18 months as part of the routine schedule. During the visit, the technician opens the enclosure door upon arrival and closes it on departure, signalling the automated QA system to mark all data as erroneous during the maintenance period (e.g. Shafer et al., 2000). In addition, specific maintenance is performed at a station when a QA meteorologist suspects erroneous observations from the station. Before leaving the site, digital photos are taken from each of the eight directions noted in Section 3.2 and added to the metadata inventory. Fiebrich et al. (2006) detailed maintenance procedures for the Oklahoma Mesonet stations.
4.3. Automated quality assurance
All OKCNET data are automatically checked by software developed by the Oklahoma Mesonet. Similar to the Oklahoma Mesonet data, OKCNET observations are never altered. Instead, each datum is flagged as either ‘good’, ‘suspect’, ‘warning’, or ‘failure’ (Shafer et al., 2000; Fiebrich and Crawford, 2001; McPherson et al., 2007).
McPherson et al. (2007) described the QA procedures at each Oklahoma Mesonet site. Similarly, data from the OKCNET traffic signal stations are automatically quality assured by range and temporal tests (see Table III), as well as spatial tests by which neighbouring stations are compared using a Barnes (1964) objective analysis scheme. In some instances, an adjustment test identifies data that failed the temporal check but agree spatially with neighbouring data. Because of the urban canyons of downtown OKC, the wind direction data are challenging to test via the spatial test, and the measurements often do not agree with data from neighbouring sites due to obstructions (e.g. buildings and trees).
| Variable | Range allowed | Maximum step allowed between consecutive 5 min observations |
|---|---|---|
| Air temperature | − 30 to 50 °C | + 6 °C or − 9 °C |
| Pressure | 800–1050 hPa | + 5 hPa or − 4 hPa |
| Relative humidity | 3–100% | + 22% or − 23% |
| Wind speed | 0–50 m s−1 | ± 40 m s−1 |
| Wind direction | 0–360° | N/A |
| Rainfall | 0–625 mm | + 40 mm |
OKCNET applies the Mesonet spatial, step, and persistence tests (McPherson et al., 2007) every minute, and only good and suspect data are delivered in real-time to users. As a result, millions of QA computations are completed on nearly 640 000 unique observations collected daily by OKCNET. To date 97.85% of all OKCNET observations have passed the QA routines as ‘good’ while the remaining 2.15% have been categorized as ‘suspect’, ‘warning’ or ‘failure’.
4.4. Manual quality assurance
Full-time QA meteorologists analyse the results of the automated quality assurance tests and perform numerous manual techniques to ensure proper QA flags are applied to each datum (McPherson et al., 2007). For data deemed erroneous, the start date/time of the problem is determined and observations are flagged manually in the QA database until the problem was resolved.
Monthly data analyses supplement daily manual QA to detect biases or drift in sensors. Statistics for the month (e.g. averages, accumulations) are computed, plotted and analyzed spatially and temporally for each variable. After the analysis of monthly data is completed, QA meteorologists prepare a QA report that documents the problems, repairs and activities performed by field technicians during the previous month that affected data quality.
Several variables measured at the traffic signal sites have specific manual QA requirements resulting from problems with the WXT510 sensor in an urban environment. For example, wind and rainfall data require special attention due to the presence of birds that randomly land on top of the WXT510 sensor, interfering with the sonic anemometer and rain plate. Birds landing on the sensor cause false wind spikes and rainfall totals.
- 1.
Wind speed: occasionally, wind speeds spike when birds land on the sensors. Such bird-induced wind gusts often happen at sunrise or sunset on days with light winds and are often 5 to 30 m s−1 greater than expected wind gusts for the ambient conditions. Each day, a QA meteorologist checks wind data for the erroneous gusts caused by birds because neither the WXT510 firmware nor the automated QA algorithms can differentiate the erroneous spikes from real wind gusts. Unfortunately, due to the spatial and temporal variability of wind, especially in the presence of severe convective storms, the employment of a step or spatial tests results in true wind gusts being flagged as ‘suspect’, ‘warning’ or ‘failure’. When a bird-induced wind gust is identified, the data are manually flagged as erroneous.
- 2.
Rainfall: to detect errors in the rainfall dataset, the QA meteorologist compares each station's accumulated rainfall to the storm-total precipitation estimated from the nearest Weather Surveillance Radar 88 Doppler (WSR-88D) radar (McPherson et al., 2007). When a large bird lands on the impact plate, the sensor can falsely register rainfall up to 1 mm. Thus, a QA meteorologist checks rainfall estimates daily to detect possible contamination of the rainfall measurements due to the presence of birds. In addition, the WXT510 rain sensor was found to be sensitive to vibrations caused by nearby train whistles, tornado sirens and even fireworks. If erroneous rainfall data are reported at a site, both the rainfall intensity and the accumulated rainfall (until 0000 UTC the following day) are manually flagged as erroneous.
5. Research and applications
The primary goal of OKCNET is the rapid collection of high-quality observations across OKC. Because OKCNET is centrally nested within the spatial domain of the Oklahoma Mesonet, observations of atmospheric variables are consistently measured across the land use gradients spanning urban to rural. Thus, numerous opportunities exist for research and applications focused on urban-atmosphere interactions in and around OKC.
For example, the observations collected by OKCNET yield new opportunities to quantify the spatial and temporal variability of the urban heat island (UHI) and its associated impacts. The physical processes responsible for the development of the UHI have been documented in numerous studies, and generally the greatest magnitudes occurred at night with calm conditions and clear skies (e.g. Bornstein, 1968; Kopec, 1970; Vukovich, 1975; Nkemdirim, 1980; Oke et al., 1999; Comrie, 2000; Hawkins et al., 2004; Fast et al., 2005). In OKC, Basara et al. (2008) previously demonstrated that a consistent urban heat island (UHI) was present from July 2002 through July 2003 using data collected from temporary stations deployed within the CBD of OKC during JU2003. Using OKCNET observations and a modified form of the methodology described in Basara et al. (2008), an intense heat wave that affected OKC from 28 July through 6 August 2008 was analyzed to quantify the combined impact of the UHI and the heat wave on temperatures in OKC (Basara et al., 2009b). The results demonstrated that the UHI exacerbated the heat wave in the OKC urban areas and provided new insights into the human toll due to enhanced heat in urban areas during heat wave events. As a result, in 2009, the National Weather Service's Weather Forecast Office (NWS WFO) in Norman, Oklahoma, used OKCNET observations along with the results of Basara et al. (2008) and Basara et al. (2009b) to more closely monitor atmospheric conditions in OKC during heat wave events.
While the methodologies employed in Basara et al. (2008) and Basara et al. (2009b) have quantified the temporal behaviour of the UHI in OKC, analysis of OKCNET data yielded additional information about the spatial extent, intensity, and temporal evolution of the UHI. Consistent with prior research, Figures 3(a) and (b) display examples of air temperature during calm conditions and reveal that the most intense UHI signature was located over OKC's central business district, decreasing in intensity from urban to rural zones. However, calm conditions in OKC are rare, even at night. As a result, the intensity, shape and orientation of the UHI changed as a function of wind speed and direction. Figure 4 displays air temperature for a case with mean southerly winds of approximately 7.5 m s−1 on 5 August 2008 at 0900 UTC (0400 LST). In this example, the UHI signature did not represent the concentric feature that was typical during calm conditions, but rather an elongated plume downwind of the CBD.
(a) The urban heat island in Oklahoma City on 11 December 2008 at 1300 UTC (0700 CDT). The observations include 9 m air temperature at OKCNET traffic signal stations and Oklahoma Mesonet stations along with interpolated (a) air temperature (black contours every 1 °C) and (b) wind vectors. Shading has also been added to enhance the visibility of the air temperature gradient. (b) The urban heat island in Oklahoma City on 23 July 2009 at 1245 UTC (0745 CDT). The observations include 9 m air temperature at OKCNET traffic signal stations and Oklahoma Mesonet stations along with interpolated (a) air temperature (black contours every 1 °C) and (b) wind vectors. Shading has also been added to enhance the visibility of the air temperature gradient. This figure is available in colour online at wileyonlinelibrary.com/journal/met
The urban heat island in Oklahoma City on 5 August 2009 at 0900 UTC (0400 CDT). The observations include 9 m air temperature at OKCNET traffic signal stations and Oklahoma Mesonet stations along with interpolated (a) air temperature (black contours every 1 °C) and (b) wind vectors. Shading has also been added to enhance the visibility of the air temperature gradient. This figure is available in colour online at wileyonlinelibrary.com/journal/met
A second area of atmospheric science that will benefit from OKCNET observations involves the occurrence of severe weather. Because of the dense population and urban sprawl of cities, there is growing concern that future severe thunderstorms (i.e. those with damaging winds, large hail and tornadoes) will lead to significant loss of life and localized catastrophic damage in urban areas (Wurman et al., 2007). To date, few studies have quantified the impacts of the urban landscape on severe weather development. However, OKC is located within a region prone to severe weather (Brooks et al., 2003; Doswell et al., 2005; Hocker and Basara, 2008a, 2008b) and during the decade prior to the deployment of OKCNET, OKC experienced violent tornadoes on 3 May 1999 (Speheger et al., 2002) and 8–9 May 2003 (Hu and Xue, 2007) as well as numerous other severe weather events.
The addition of OKCNET already has demonstrated how the rapid collection of in situ observations, combined with other operational (e.g. the Oklahoma Mesonet; WSR-88D radars, Crum et al., 1993) and experimental (e.g. the Phased Array Radar; Heinselman et al., 2008) platforms, can enhance the analysis of severe weather events. For example, on 14 May 2009, a thunderstorm developed over the northern periphery of the OKC metropolitan area and quickly evolved into a supercell thunderstorm as it moved southward. While radar data revealed an intense supercell thunderstorm moving into the CBD at 0314 UTC, the OKCNET data provided near-surface temperature, humidity and wind characteristics in the forward- and rear-flank downdrafts and the inflow notch (associated with the updraft of the storm).
Along with severe thunderstorms, heavy rainfall and frozen precipitation can have adverse consequences throughout OKC. To prepare better for these conditions, the City of Oklahoma City uses OKCNET data to allocate personnel and resources to locations that need critical attention. For example, during localized heavy precipitation, OKC Public Works crews are dispatched to locations that need on-site monitoring to mitigate storm water flow. Similarly, during frozen precipitation events, street and maintenance crews are deployed to locations across OKC with temperatures below freezing to clear and maintain roadway conditions.
OKCNET observations have also captured details of atmospheric processes including frontal passages, thunderstorm outflow boundaries, dry line passages, gravity waves and heatbursts associated with decaying thunderstorms. Further, the voluminous data collected by OKCNET will help validate or will be assimilated into atmospheric dispersion models. Future scientific research using and applications of OKCNET data appear to be extensive.
6. Conclusions
The design, deployment and operation of OKCNET represents a significant innovation applied to the collection of meteorological observations within an urban setting using state-of-the-art technology. Each station was designed to be robust, low power and aesthetically inconspicuous while at a reasonable cost. Because of the design of each traffic signal station, installation procedures were straightforward and low cost. Careful consideration was made to select each site in the network to be representative of the local atmospheric conditions while monitoring the larger urban environment of OKC.
With an average station spacing of approximately 3 km, OKCNET observes atmospheric conditions at a fine spatial resolution. Additionally, a key component of OKCNET is rapid collection and dissemination of high-quality observations. Every minute, atmospheric conditions at each traffic signal station are measured, transmitted to a central facility and made available to users via the Internet (http://okc.mesonet.org). A similar procedure occurs every 5 min for observations at the Oklahoma Mesonet stations in OKC. As a result, approximately 640 000 unique observations are collected daily across OKC.
Overall, OKCNET provides a solid foundation for long-term monitoring, research and applications focused on urban-atmosphere interactions that benefit a range of end users in and around OKC. The success of OKCNET is contingent upon the quality of the observations collected, continued funding and an increasing number and diversity of end users. In addition, the Oklahoma Mesonet infrastructure was a key component to the development of OKCNET and will be critical to its future. Further, it is hoped that the lessons learned during the establishment of OKCNET will spur and enhance the development and installation of future observing networks in other urban areas.
Acknowledgements
Funding for the Oklahoma City Micronet and support for this study was provided by the Oklahoma State Regents for Higher Education (equipment), the Office of the Vice President for Research at the University of Oklahoma (personnel and supplies), the Oklahoma Mesonet (personnel), the Oklahoma Climatological Survey (personnel and in-kind support), and the City of Oklahoma City (in-kind support). Oklahoma's taxpayers fund the Oklahoma Mesonet through the Oklahoma State Regents for Higher Education and the Oklahoma Department of Public Safety. The authors thank the City of Oklahoma City for their extraordinary support of OKCNET and, specifically, for the vision and assistance of J. C. Reiss, Mike Degiacomo, Mark Meier, and Steve Eaton. The authors also thank the many staff and students at the Oklahoma Climatological Survey, Dr. T. H. Lee Williams, the Oklahoma Mesonet Steering Committee, Dr. Kelvin Droegemeier, Dr. Petra Klein and Dr. May Yuan for their guidance and support. Lastly, the authors thank the anonymous reviewers who provided constructive criticism that improved the quality of this article.
