Weather Data For Building Energy Analysis

Transcript

AUTODESK GREEN BUILDING STUDIO WHITE PAPER Weather Data for Building Energy Analysis By Stuart Malkin, Meteorologist – Software Development Manager – Autodesk, Inc. Overview Autodesk® Green Building Studio® provides a complete year of weather data for design and building energy analysis. The data set is in a user friendly and binary DOE2 ready format and includes meteorological variables of dry bulb temperature, dew point temperature, relative humidity, wind speed and wind direction, direct normal radiation, global and diffuse horizontal radiation, and total sky cover, among others. With 55,000+ locations at a spatial resolution of approximately 20 km, a GBS virtual weather station is no further than 14 km (8.8 miles) from any given project within the contiguous 48 states of the United States. This whitepaper discusses the benefits of using Green Building Studio (GBS) weather data as well as information about the models used to generate the data set, and provides a comparison of the GBS data set to actual observations at several locations within the United States. Finally, the whitepaper provides guidance in choosing a weather station for energy analysis. Benefits Currently, using the Typical Meteorological Year (TMY2) weather data set provides only 231 locations. Almost all TMY2 locations are situated at large airports and may not represent the weather at your building. For example, the nearest TMY2 weather station is 23 miles from the city center of Washington, DC, while a GBS virtual weather station is within 4 miles of the center. The map of California and Nevada below shows the density of the weather data locations. The yellow points represent Climate Server locations (~2000 in view), while the red dots represent TMY2 locations (17 in view). 1 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Copyright © and (p) 1988-2006 Microsoft Corporation and/or its suppliers. All rights reserved. © 2008 Autodesk, Inc. All rights reserved. The Green Building Studio web service automatically generates several weather graphs and reports, saving you time and expense in preparing for the climate study portion of your design process. Most of these graphs and reports are customizable too, providing you with site specific results instantly. The graphs below showing an annual wind rose and ambient temperature are just two examples of the many custom graphs available. 2 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Architects and building engineers need to increasingly pay attention to designing for changing climatic conditions. Using older data does not capture changing conditions and puts designers at risk. The graph below shows how cooling degree days (CDD) have been increasing in Salt Lake City (SLC) over time. The red bars represent anomalies from normal for the years 1948-2005, while the blue line denotes the 10 year running mean anomaly. The 10 year running average has increased approximately 20% above normal. Using older data sets such as TMY2 does not capture changing conditions such as what is occurring in Salt Lake City and throughout the world. Normal CDD (1948-2005): 1084 Data provided by National Weather Forecast Office, Salt Lake City, UT About Autodesk Green Building Studio Weather Data The GBS weather data includes 55,000 + virtual weather station, 231 TMY2 stations, and 16 California Climate Zone (CCZ) stations. GBS virtual station data was derived using two weather models—the Rapid Update Cycle (RUC) and Mesoscale Meteorological Model version 5 (MM5). RUC current conditions (analysis fields) were used for the bulk of hourly data in 2004. Because RUC is an operational model where no effort was made to make up for model failure or to correct errors, MM5 was used to fill in the four gaps that occurred in 2004. Rapid Update Cycle (RUC) The Rapid Update Cycle (RUC) model is a NOAA/NCEP operational weather prediction system comprised primarily of a numerical forecast model and an analysis and assimilation system to initialize the model. The model was developed to serve users needing frequently updated short-range weather forecasts, including those in the US aviation community and US severe weather forecasting community. RUC runs at the highest frequency of any forecast model at NCEP, assimilating recent observation to provide hourly updates of current conditions (analyses) and short-range forecasts. The GBS virtual station data set uses the current condition update at the surface for each hour. The current condition field is generated using an optimal interpolation (OI) analysis to assimilate observations and satellite data. The current condition update is an analysis field using the previous 1 hour forecast and current observations to correct the forecast. This method provides an hourly update of conditions very close to actual observations and 3 AUTODESK GREEN BUILDING STUDIO WEATHER DATA provides a numerical stable field for the model to provide rapidly updated forecasts. Observations come from a variety of sources. These sources of observations and data include surface reporting stations and buoys, commercial aircraft, wind profilers, rawinsondes, dropwinsondes, Radio Acoustic Sounding System (RASS), Velocity-Azimuth Display (VAD) winds from Doppler radar, GOES, GPS, and SSM/I satellite data. OI spatially interpolates meteorological observations to generate an analysis field. OI includes a quality control check on the residuals between observations and the analysis field. At a given observation point, if the estimated and measured observation differ by more than a prescribed amount, further checks determine whether the observation or one if its neighbors is erroneous. If the observation is determined to be erroneous, it is dropped from the analysis (Benjamin et al, 2004). For more information on RUC data assimilation, see http://wwwfrd.fsl.noaa.gov/pub/papers/Benjamin2004c/j.pdf. For more information on RUC visit the RUC home page at http://ruc.fsl.noaa.gov/ Mesoscale Model version 5 (MM5) In 2004, there were four periods of missing RUC data. Eight hours were missing on February 28, four hours were missing on March 8, nine hours were missing on July 3, and 5 hours were missing on July 14. MM5 was used to fill these gaps. MM5 is a limited-area, nonhydrostatic, terrain-following sigma-coordinate model designed to simulate or predict mesoscale atmospheric circulation. It has been developed at Penn State and NCAR as a community mesoscale model and is continuously being improved by contributions from users at several universities and government laboratories. A nested domain with a grid spacing of 20 km provided a similar domain as the RUC model. Unlike RUC analysis, MM5 simulation output provides a true forecast. In order to match the output variables from MM5 as close as possible to those from RUC, a Land Surface Model (LSM) soil model had to be used. Please see http://www.mmm.ucar.edu/mm5/mm5-home.html for more information about MM5. For a detailed description of the MM5 model, see http://www.mmm.ucar.edu/mm5/documents/mm5-desc-doc.html (Grell et al, 1994). Comparison to Observations Hourly comparisons of Dry Bulb Temperature between GBS virtual stations and observations for various locations within the United States are shown below. Observation stations were chosen based on their availability of data, data completeness, and geographic location compared to GBS virtual stations. Figure 1 shows the observation locations chosen for this discussion mapped with the RUC domain and terrain elevation. Approximate location of observation stations in discussion are shown as the violet dots. 4 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Figure 1—Locations of observations in discussion plotted with the RUC Domain and terrain elevation of the 20-km Contour elevation is 200 m (adapted from Benjamin et al, 2004). Sacramento Executive Airport, CA (WBAN 23232) and Stillwater, Oklahoma (WBAN 53926 and 53927) were chosen as inland observation stations away from the influence of large water bodies. The Sacramento site at an elevation of 17 feet is approximately 2.3 km (1.4 miles) from the nearest GBS virtual station which has a similar elevation of 39 feet. In comparison, the Stillwater sites were 6.8 miles (10.9 km) or more from the nearest GBS virtual station as shown in figure 2. The distance between the Stillwater observations and the nearest virtual GBS station represents a ‘worst case’ scenario in terms of distance between a project location and a GBS virtual station. A GBS virtual station will be no further than approximately 14 km from any project location in the contiguous United States. The elevations of the Stillwater stations are similar to the nearest GBS virtual stations (890 feet for Stillwater and 919, 978, 932, and 1001 feet respectively GBS virtual stations labeled GBS_04R20_156091, GBS_04R20_155091, GBS_04R20_156092, GBS_04R20_155092 in Figure 2). 5 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Figure 2—Shows location of two colocated surface observations near Stillwater, OK (marked as the green building) and the location of four GBS virtual stations (GBS_04R20_15609, GBS_04R20_155091, GBS_04R20_155092, GBS_04R20_156092) Sacramento dry bulb hourly temperature for 2004 shows nearly linear fit match in figure 3 (slope of 1.03 and an intercept of -2.51 with a total of 8544 paired samples). The R-squared value of .96 indicates that the GBS virtual station dry bulb data can explain changes in the observations 96% of the time. Missing hours not used in the regression were due to missing observational data (216 hours). 6 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Figure 3—Linear Regression comparing dry bulb temperature observations at Sacramento Executive Airport to that of the closest GBS virtual station (GBS_04R20_049116). The regression result at Stillwater, Oklahoma was similar to the Sacramento case. Three analyses were performed to investigate if there was a large influence on the distance of a GBS virtual station tracking the observational data. The first analysis used the average between the two co-located observation stations in Stillwater and compared that to the closest GBS virtual station (labeled GBS_04R20_156091 in figure 2). The second analysis compared the average of the two co-located observation stations and the four GBS virtual stations using an inverse distance weighting scheme. Finally, a comparison was made of the two co-located observation stations and the average of all four GBS virtual stations, giving them equal weight. Because all three analyses produced almost identical results to each other, only the regression of the observations against the closest station is shown below in Figure 4. With an R-squared value of .97, the GBS virtual station dry bulb data explains changes in the observations 97% of the time for the entire year (no missing observations). 7 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Figure 4—Linear regression comparing dry bulb temperature observations at Stillwater, OK (two co-located stations) to that of the closest GBS virtual station (GBS_04R20_156091). As an example of how the GBS virtual station tracks dry bulb observation at Stillwater, which is 6.9 miles (10.9 km) away, figure 5 shows an hourly time series of both trends for a ten day period (April 8-18, 2004). This period during the spring was chosen because of great variability in the temperature (temperature ranged from approximately 32 °F – °86 F). Large temperature swings are typical of a large diurnal cycle and frontal passages during this time of year in the area. Qualitatively, the GBS virtual station data and the Stillwater data match well on an hour-by-hour basis. There were two short periods where there was a noticeable difference (between hour 6 and 12 on April 11 and hours 8 through 6 on April 13-14). However, the two sets of data follow the same trend and do not differ by more than 7 °F at any given hour during these brief events. 8 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Figure 5—Time series plot of dry bulb temperature for Stillwater, OK (purple diamonds) and the closest GBS virtual station (green triangle) for April 8-10, 2004. Oakland Metropolitan Airport (WBAN 23230), and JFK, International Airport (WBAN 94789) were chosen because they represent sites whose climate is moderated by the Pacific and Atlantic, respectively. The GBS virtual station for Oakland is approximately 2 miles from the observation at the airport. While Oakland airport is near sea level, the GBS virtual station is at an average elevation of 502 feet. Because RUC uses a slope envelope topography that has a resolution of approximately 20 km, the elevation at a particular point represents an average topographical height and slope within a 20 km x 20 km box (Benjamin et al, 2004). Thus the elevation of 502 feet is an averaged elevation within the grid. The GBS virtual station for JFK International is approximately 4 miles northwest of the airport observation station. The elevations of JFK International and the closest GBS virtual station are 16 feet and 39 feet, respectively. The regression of observations compared to GBS virtual stations are shown in Figure 6 for both Oakland, Metropolitan and JFK International. 9 AUTODESK GREEN BUILDING STUDIO WEATHER DATA Figure 6— (a) Linear regression comparing dry bulb temperature observations at Oakland Metropolitan Airport, CA to that of the closest GBS virtual station (GBS_04R20_045112). The data from Oakland Metropolitan were reported to the nearest degree Celsius, which gives the data a columnar appearance. (b) Linear regression comparing dry bulb temperature observations at JFK International Airport, NY, to that of the closest GBS virtual station (GBS_04R20_256124). The regression at Oakland (figure 6(a)) has an R-squared value of .90 indicating the GBS virtual station dry bulb data explains changes in the observations 90% of the time for the entire year. While this value is not as strong as the nearly perfectly linear fit of Sacramento, Stillwater, and JFK International, it is nonetheless a good linear fit of the data to actual observations. Possible reasons for a less than linear fit may be related to the terrain and land use being resolved to 20 km, the density of observations affecting the RUC analyses fields, and the reporting of dry bulb temperature to the nearest degree Celsius. Oakland Metropolitan Airport sits on the eastern edge of the San Francisco Bay and has complex terrain. The elevation ranges from sea level to over 1400 ft within 5 miles to the east. Thus the elevation, shape of the slopes, and the boundary between the bay and land may have effect the RUC analysis field. In addition, the network of observations on the west coast near the Pacific is generally poorer than inland areas. While RUC ingests observations coming from ships, airplanes, and other observations as the present themselves in real-time, the number of surface stations is limited. Thus the RUC analysis field may also be affected by less dense set of observations near the coast. Finally, the fit may be affected the by the fact that Oakland Metropolitan dry bulb temperature is reported to the nearest degree Celsius, which is apparent by the columnar appearance of the data in figure 6(a). The regression at JFK International (figure 6(b)) has a slightly better linear fit than the other inland stations mentioned above (R-squared value of .98). Thus the GBS virtual station represents actual dry bulb temperature observations very well over the entire year. Which Weather Station Do I Choose? How does one decide on which weather station to choose for energy analysis? While there is no ‘one-size fits all’ answer for every situation, there are several guidelines one should consider when making this decision. The most important factor to consider is distance. The closer a weather station is to a project location, the more representative the data will be. This is especially important in an area that 10 AUTODESK GREEN BUILDING STUDIO WEATHER DATA may have complex terrain or microclimates. The GBS weather data set offers a complete year of weather data for energy analysis at an approximate 20 km resolution. Representative data will be within 14 km or less of a project location. As shown in figure 7, the GBS web service maps and reports the distance of closest set of GBS virtual stations, TMY2 Stations, and California Climate Zone stations to help you decide. Another factor to consider is the elevation of your project. The project elevation and the weather station elevation should be similar. It would inappropriate to choose a weather station that sits on top of a mountain peak or even on the other side of a mountain if the project location sits in a valley. In addition, to mapping and reporting distances, the GBS Web Service enables you to view the locations of available weather stations in a terrain, satellite, or hybrid view through the Google™ map interface. This feature will show topographical and land use features that you can use to help you decide on an appropriate weather station. Land use is also another important factor to consider. Is your project next to a large body of water, which may moderate the local climate? In this case, an inland weather station would probably not represent your local project’s weather. The mapping program in the GBS web service also helps you choose the closest weather station that may represent your location based on visible land and water features. Finally, the GBS web service mapping program automatically calculates degree days and design conditions for each weather station. Figure 7 also shows an example of the Washington, DC area with Cooling Design Conditions for various design thresholds. This information can be useful in determining if the station represents the weather you expect in the location. Or if you want to choose a station that has either hotter or colder extremes than you expect, you can determine this by looking at the design conditions and degree days. Figure 7—The GBS Web Service includes a map view that lets you see locations of weather stations, distances, land and water features, and automatically calculates design conditions and degree days. Because no weather data set is perfect, the Green Building Studio web service gives you flexibility in choosing the appropriate weather. While we have confidence that the GBS 11 AUTODESK GREEN BUILDING STUDIO WEATHER DATA weather data set emulates nearby observations, there may be areas near coastlines and complex topography that may not be as representative. Sources Benjamin, S. G, Ve’nyi, D. D., Weygandt, S. S., Brundage, K. J., Brown, J. M., Grell, G. A., Kim, D., Schwartz, B. E., Smirnova, T. G., Smith, T. L., Manikin, G. S, 2004. An Hourly Assimilation–Forecast Cycle: The RUC, Monthly Weather Review 132, 495 – 518. Grell, G. A., Dudhia, J., Stauffer, D. R., 1994. A Description of the Fifth Generation Penn State/NCAR Mesoscale Model (MM5), NCAR Technical Note NCAR/TN-398 + STR, 138 pp. Autodesk and Autodesk Green Building Studio are either registered trademarks or trademarks of Autodesk, Inc., in the USA and/or other countries. All other brand names, product names, or trademarks belong to their respective holders. Autodesk reserves the right to alter product offerings and specifications at any time without notice, and is not responsible for typographical or graphical errors that may appear in this document. © 2008 Autodesk, Inc. All rights reserved. 12