Existing Climate Data Sources and Their Use in Heat Island Research

Unfortunately, high-resolution data collection efforts in urban areas are not centrally documented in the national or regional climate centers. Thus far, we have found several potentially useful sources of data.

Cooperative network stations

The cooperative network consists of a large number of stations, as discussed previously. As shown in Table 1, large urban areas contain many of such stations. While these stations usually do not offer the spatial resolution required for correlation with urban surface characteristics, they may be adequate for analysis that is more modest. Unfortunately, many of the characteristics of the heat island signal may be lost due to the low time resolution of the cooperative network temperature data, which includes only local daily extrema.

First-order stations and military installations

Higher time resolution is offered by first-order stations and military weather stations, which record data hourly. In addition, since these stations are typically located at airports (i.e. in areas without dense canopies) microclimate effects are very weak, and those that do exist are similar at all stations. Thus, data at different stations can be directly compared and temperatures are representative of large areas. However, the spatial density of such stations is very limited.

Agricultural Networks

To obtain higher spatial densities of weather stations, one must turn to regional or city-scale weather networks operated by various agencies. Due to the need for accurate local climate data for farming, agricultural weather networks have been assembled in a number of states to provide on-line access to hourly weather data.

One such network is the Oklahoma Mesonet, a network of 156 automated observation stations that are operated and maintained by Oklahoma State University (OSU) and Oklahoma University (OU). The stations are located in fenced 10-meter square compounds and are separated by an average spacing of 30 km. This provides a partly uniform microclimate for comparison between stations. Yet, again, the spatial resolution provided by the data is low (Crawford 1993). Each records rainfall, barometric station pressure, solar radiation, air temperature and relative humidity at 1.5 meters, wind speed and direction at 10 meters, and soil temperature under both bare soil and a natural grass cover at 10 cm depth. At approximately half of these stations, several supplemental parameters are also recorded, including air temperature at 9 meters, wind speed at 2 meters, leaf wetness, and soil temperatures at additional depths. Measurements are made every five minutes, and fifteen-minute averages are sent to OU and OSU where they are checked and archived.

A similar network is operated by the High Plains Climate Center. The station density is much lower, as over one hundred automated weather stations cover four states (Kansas, Nebraska, North Dakota, and South Dakota) and parts of five others (Colorado, Iowa, Minnesota, Missouri, and Wyoming). Urban centers, naturally, receive denser coverage than rural areas.

Utility-operated networks

Utility forecasting and demand management rely heavily on accurate weather information. Given the heterogeneity of urban climates, some utilities have established urban weather station networks to improve their characterization of weather conditions. These can add significantly to the station density in urban climate research.

One such network is operated by the Southern California Edison Company (SCE) serving the Los Angeles area, excluding the city of Los Angeles itself and a few neighboring cities. The stations record drybulb and wetbulb air temperature, insolation, wind speed and direction, and atmospheric pressure at hourly intervals. The stations also provide averages and extrema of several recorded parameters at daily intervals. Data have been recorded continually since May 1987 and can be obtained through SCE. Once again, as discussed in Kurn et al. (1994), these stations by themselves cannot provide adequate coverage for interpolation between stations. Adequate coverage may be obtained by supplementing the measurements with other data sources. Unfortunately, the SCE stations are not examined and calibrated on a regular basis. Rather, inspections and adjustments are performed when SCE is notified of probable errors by researchers using the data. Thus, previously recorded data from these stations should probably not be used in urban climate studies, where the resolution of temperature data need to be on the order of 0.5K. However, if a period of data collection is accompanied with regular calibrations, the network can be used fruitfully.

The densest existing station network we have encountered is the Phoenix Real-time Instrumentation for Surface Meteorological Studies (PRISMS) network in Phoenix, Arizona. The station network is co-operated by the Salt River Project (utility serving the greater Phoenix area), the National Severe Storms Laboratory, and Arizona State University (ASU). The network consists of 16 stations located in open areas within utility substation compounds, at least 10 feet away from the border of the compound marked by a fence or wall (see Table 7). Weather parameters, recorded every five minutes at each of the sites, include air temperature, humidity, air pressure, wind speed and direction, and rainfall. Placement of sensors conforms to NWS standards, except that wind is monitored at 6.1 meters from the surface and precipitation is recorded at the surface. Data are collected at ASUÕs Office of Climatology on a real-time basis for archiving, and quality control programs are applied to the data. Data errors, which result mostly from communication errors, are flagged in the archives. Calibration errors are assumed negligible because SRP personnel inspect the sensors constantly. The spacing between stations within the dense central urban area (within a triangle with vertices at Arcadia, Falcon, and Corbell stations) is about 8 km.

Air quality monitoring stations

Finally, a large amount of data may be obtained from air quality monitoring stations that are abundant in large urban areas with air pollution problems. While these stations primarily monitor the levels of several types of air constituents, many stations also record weather conditions and insolation levels. For example, in the Los Angeles area, air temperature, humidity, wind speed and direction, and insolation are monitored hourly at more than thirty locations, providing the most complete picture of the Los Angeles urban climate.

Air quality monitoring stations are operated in the United States by local, state, and federal organizations. Fortunately, the large body of data collected at these stations is centrally compiled and made available to the public through the Aerometric Information Retrieval Service (AIRS) operated by the U.S. Environmental Protection Agency. Through this service, one may perform searches through station inventories; obtain detailed station descriptions that include a description of sensors, their placement, and a history of changes at the station; and obtain archived climate data. AIRS can be accessed either by requesting queries and sets of data or by obtaining remote computer access through the National Computing Center (see appendices).

Obtaining maximal data coverage of specific urban regions

Even when no single station provides sufficient coverage for an analysis, the various sources of climate data in combination may suffice. For example, in the Los Angeles region (as far east as Palm Springs), SCE, SCAQMD, first-order NWS, and military weather stations combine to over ninety stations recording hourly weather conditions. This would enable a detailed study of climate variations in different parts of the city and thus the potential for studying the effects of various surface conditions on the scale of a city neighborhood. However, since the climate variation between nearby neighborhoods is slight (perhaps 0.5 - 1K in air temperature, 0.5 g/kg in absolute humidity), the records must be very accurate. Thus, it is doubtful whether the data previously recorded by different weather networks could be directly combined into a single analysis. For example, systematic absolute humidity differences of as much as 6 g/kg were observed between several pairs of SCAQMD and SCE stations, each separated by no more than 24 km. Data from the various stations should, thus, probably be combined only if their collection is accompanied by a rigorous ongoing calibration effort. Continue to: Chapter IIIC. Analysis of Short-term Data of the Urban Climate of Phoenix