Printer Friendly

An analysis of NASA's MERRA meteorological data to supplement observational data for calculation of climatic design conditions.


"The ASHRAE Technical Committee TC 4.2 (Climatic Information) is concerned with identification, analysis and tabulation of climatic data for use in analysis and design of heating, refrigeration, ventilation and air-conditioning systems. Promotion of effective use of weather information in these applications is also included." (ASHRAE TC 4.2 2009). Accordingly, there is a continuing effort to expand the number of global observational surface sites, to provide methods to fill in missing meteorological data on an hourly, daily, or monthly scale, and to introduce new ways of viewing data or to show year-to-year variability in the current data.

As evidence of this ongoing effort by TC 4.2, Chapter 14 of ASHRAE Handbook--Fundamentals (ASHRAE 2009a), entitled "Climatic Design Information," provides climate design conditions for 5564 locations around the world. The meteorological-based design information is derived from hourly observations of dry-bulb temperature, wet-bulb temperature, dew-point temperature, wind speed and direction, and surface pressure. These data are generally from surface stations having a minimum of 8 years, that need not be continuous, up to a maximum of 25 years of continuous data. The ASHRAE solar-related design conditions are not based on surface measurements, but on satellite model-derived solar radiation. Weather Data Viewer Version 4 (WDV-4) (ASHRAE 2009b) contains the same climate design tables for the 5564 locations in the ASHRAE Handbook--Fundamentals along with background information such as the specific years and months used in the calculation of the climate design conditions.

The 2009 design conditions provided a significant enhancement over the 2005 design conditions with respect to the global coverage--5564 locations in 2009 versus 4422 locations in 2005. Even with the improved geographical coverage in ASHRAE Handbook--Fundamentals, many regions of the world continue to rely on design conditions based upon the closest surface station. As examples of the dearth of hourly surface observations in some regions of the world, only 30 surface sites are listed in ASHRAE Handbook--Fundamentals for Brazil, South America; 2 for Kenya, Africa; and none for Ethiopia, Africa. Moreover, as noted in the ASHRAE Handbook--Fundamentals, observational data from many surface stations are incomplete or missing significant blocks of time within the design period. Having a full 25-years of continuous MERRA meteorological values should help reduce uncertainty in discontinuous surface records by more completely representing the variability over the entire period. These global and time contiguous meteorological and solar data offer significant enhancements to ASHRAE's ability to provide Climate Design Conditions in regions where meteorological data is either unreliable, unavailable, or too far removed from the desired location to be of any real value. NASA's POWER (Prediction of Worldwide Energy Resource) project (NASA 2013a) was initiated to facilitate access to NASA solar and meteorological data related to renewable energy, sustainable buildings, and agroclimatology (Chandler et al. 2011; Eckman and Stackhouse 2012). Both daily averaged solar and meteorological data are available through an online data portal at (NASA 2013a). The solar data is based upon satellite observations. The meteorological data is based upon reanalysis models developed by NASA's Global Modeling and Assimilation Office (GMAO) and can be found at (NASA 2013b). Currently, through the POWER data portal, both solar and meteorological daily averaged data are available globally on a 1[degrees] latitude by 1[degrees] longitude grid. The solar data covers the time period from July 1982 to within one week of the current time and the meteorological data covers the time period from January 1982 to within several days of current time.

The NASA POWER project has recently made available meteorological data based upon an improved reanalysis model, Modern Era Retrospective-Analysis for Research and Applications (MERRA) (Rienecker et al. 2008; Bosilovich 2008). The output from MERRA yields global, hourly surface meteorological parameters for the years 1981 to present with a remapped spatial resolution of 0.5[degrees] latitude by 0.5[degrees] longitude. An initial evaluation of the MERRA temperatures indicate accuracies sufficient to warrant their use when surface observations are unavailable.

In this paper, we present our evaluation of the accuracy of the MERRA temperatures, followed by an assessment of the applicability of the MERRA hourly temperatures in the development of annual dry-bulb climate design criteria.



Hourly dry-bulb temperatures are taken from the output of the NASA's Global Modeling and Assimilation Office (GMAO) MERRA assimilation model (Rienecker et al. 2008; Bosilovich 2008). The MERRA assimilation model is a general circulation model that provides estimates of various parameters via "An atmospheric analysis performed within a data assimilation context [that] seeks to combine in some 'optimal' fashion the information from irregularly distributed atmospheric observations with a model state obtained from a forecast initialized from a previous analysis" (Bloom et al. 2005). The model seeks to assimilate and optimize observational data and model estimates of atmospheric variables. Types of observations used in the analysis include land surface observations of surface pressure; ocean surface observations of sea level pressure and winds; sea level winds inferred from backscatter returns from space-borne radars; conventional upper-air data from rawinsondes (e.g., height, temperature, wind, and moisture); additional sources of upper-air data including drop sondes, pilot balloons, aircraft winds; and remotely sensed information from satellites (e.g. height and moisture profiles, total precipitable water, and single level cloud-motion vector winds obtained from geostationary satellite images).

The initial output of MERRA yields hourly surface meteorological parameters at a global resolution of 0.5[degrees] latitude by 0.67[degrees] longitude for the years 1981 to present. The POWER project uses bilinear interpolation to regrid the MERRA data to a resolution of 0.5[degrees] latitude by 0.5[degrees] longitude in order to be more compatible with the 1[degrees] solar data also available though the POWER data portal.

Surface Station Data

Our initial evaluation of the MERRA temperatures was based on a comparison of the MERRA 2-meter daily average (Tave), daily maximum (Tmax), and daily minimum (Tmin) temperatures with the corresponding 2-meter surface observations obtained from the National Climatic Data Center (NCDC) Global Summary of the Day (GSOD) files (NCDC 2003). The MERRA Tave, Tmax, and Tmin values are based on the MERRA hourly values. The observational data reported in the NCDC GSOD files are hourly observations from globally distributed surface stations with observations typically beginning at 0000 Universal Time Coordinate (UTC). The NCDC daily Tmin, Tmax, and Tave reported in the GSOD were derived from the available hourly observations. For the MERRA versus NCDC comparison reported below, the NCDC daily Tmin, Tmax, and Tave values were filtered by an 85% selection criteria applied to the observations reported for each station. Namely, only values from NCDC stations reporting 85% or greater of the possible 1-hourly observations per day and 85% or greater of the possible days per month were used. This selection criteria results in a balance of having enough global NCDC stations reporting daily data, while also having enough 1-hourly observations to capture the daily temperature cycle (Tave, Tmin, Tmax). It should be noted of the 1116 NCDC stations used in this study, 55 stations were found to be reporting some daily temperatures in Celsius rather than Fahrenheit as per the NCDC GSOD documentation. These values were identified and converted back to Fahrenheit. Additionally, of the 1116 NCDC stations which passed the 85% selection criteria, only 1105 of them were used to calculate the annual dry-bulb temperature critical design conditions because of the ASHRAE requirement for the stations to have 8 years of continuous data.

Design Conditions

An important product of ASHRAE Handbook--Fundamentals is the climate design criteria for 5564 locations in the United States, Canada, and around the world. The ASHRAE design criteria reported in ASHRAE Handbook--Fundamentals and on the WDV-4 are based primarily upon hourly surface-based observations from the NCDC Integrated Surface Dataset (ISD) for the data period 1982 through 2006 and the Environment Canada (ASHRAE 2009a; ASHRAE 2009b) hourly weather records for the period 1982 through 2006. The ISD data are the hourly observations from which the NCDC GSOD values are derived.

As an initial assessment of the applicability of the MERRA data for developing comparable design conditions for any global location, the MERRA hourly dry-bulb temperatures were used to calculate dry-bulb related design conditions given in Table I of Chapter 14 of ASHRAE Handbook--Fundamentals. In particular, MERRA-based annual heating and cooling degree-days and MERRA dry-bulb temperatures corresponding to the 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% annual cumulative frequency of occurrence were compared to the corresponding values taken from the WDV-4.

We should note that the MERRA air temperatures represent the average value on a 0.5[degrees] x 0.5[degrees] latitude, longitude grid cell at an elevation of 2 m above the average elevation of earth's surface above mean sea level encompassed by the MERRA grid, while the NCDC values are the 2 meter local surface observations at the site's elevation above mean sea level. Further, the NCDC observations are typically from airport sites characterized by relatively flat terrain with low-lying vegetation. Accordingly, the elevation of the MERRA data, particularly in complex terrain, is typically above the elevation of the surface observations. Extensive analysis of temperature values from a previous reanalysis model clearly demonstrate improved agreement with surface observations by applying a lapse-rate correction based on the elevation difference between the model grid and the corresponding surface-site. Results from application of a lapse rate-correction are detailed in a section of the online POWER methodology documentation at Methodology/Building1d0_Methodology_AppendixA.html (NASA 2012). For the analysis reported herein, no lapse-rate corrections were applied to the MERRA values since they have yet to be developed for the MERRA data.


To evaluate the MERRA air temperatures, we compared the MERRA daily Tave, Tmin, and Tmax temperatures with the corresponding values taken from NCDC GSOD files. Comparison metrics include the parameters associated with linear-least squares fit to the MERRA versus NCDC scatter plots (i.e., slope, intercept, and the Pearson correlation coefficient, [R.sup.2]), and the root means square error (RMSE), mean bias error (MBE), and the absolute mean bias error (AMBE) of the MERRA-based values relative to the NCDC values. The MBE, AMBE, and RMSE are given as:

MBE = (1/N)([[SIGMA].sub.j]){[[SIGMA].sub.i][[([T.sub.ij]).sub.MERRA] - [([T.sub.ij]).sub.NCDC]]} (1)

AMBE = (1/N)([[SIGMA].sub.j]){[[SIGMA].sub.i]ABS[[([T.sub.ij]).sub.MERRA] - [([T.sub.ij]).sub.NCDC]]} (2)

RMSE = [([[SIGMA].sub.j]{[[SIGMA].sub.i](1/N)[[[([T.sub.ij]).sub.MERRA] - [([T.sub.ij]).sub.NCDC]].sup.2]}).sup.1/2] (3)

where [[SIGMA].sub.i] is summation over all days meeting the 85% selection criteria described above, [[SIGMA].sub.j] indicates the sum over all stations, [([T.sub.ij]).sub.NCDC] is the temperature on day i for station j, and [([T.sub.ij]).sub.MERRA] is the MERRA temperature for day i and station j, and N is the number of matching pairs of NCDC and MERRA values.

The temperature evaluation was performed by comparing the daily averaged (Tave), and daily maximum (Tmax), and daily minimum (Tmin) temperatures from surface sites within the CONtinental United States (CONUS) study region defined by 23[degrees]N x 125[degrees]W and 50[degrees]N x 65[degrees]W to the corresponding values determined from the MERRA hourly temperatures.

The evaluation of the annual dry-bulb design criteria is based on a comparison of the annual 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb temperatures estimated from the MERRA-based cumulative distribution functions (CDFs) developed from the MERRA hourly temperatures and the corresponding annual 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb temperatures taken from the WDV-4, where the site specific values from the WDV-4 are based upon surface station hourly observations. The methodology for determining the annual CDFs and subsequently the 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb temperatures is described in Thevenard and Humphries (2005) and by Thevenard (2009). Evaluation of the MERRA-based annual design conditions focused on the CONUS study-region, and was accomplished by first acquiring the MERRA hourly temperature for the 6480 0.5[degrees] grid boxes contained within the CONUS study-region. The MERRA hourly data covered the time period January 1, 1982 through December 31, 2006. The design values taken from the WDV-4 were nominally for the same time period, although the hourly data for many of the surface sites were incomplete (i.e., missing days and/or months). Forthe analysis described below there was no attempt to match the temporal coverage of the MERRA data to the record from specific surface sites.

The MERRA CDF was developed and the corresponding 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% annual dry-bulb temperatures for each grid box were determined. The MERRA-based design conditions were compared to design conditions taken from the ASHRAE WDV-4 for any surface station or stations contained within the particular MERRA grid box. Comparison metrics for the annual design conditions are similar to those described above for the MERRA temperatures.

The MERRA-based heating degree-days (HDD) and cooling degree-days (CDD) were also evaluated by comparing to values taken from the ASHRAE WDV-4. For the current evaluation only the annual HDD and CDD averaged over the study years (i.e., 1982-2006) were considered. The HDD and CDD definition is given in Chapter 14 of ASHRAE Handbook--Fundamentals as:

HDD = [summation][([T.sub.base] - <[T.sub.i]>).sup.+] (4)

CDD = [sumamtion][(<[T.sub.i]> - [T.sub.base]).sup.+] (5)

where [T.sub.base]([degrees]C [[degrees]F]) is the reference temperature to which the degree-days are calculated, and <[T.sub.i]> is the mean daily temperature calculated by adding Tmax and Tmin for the day and dividing by two. In this paper, Tbase for both HDD and CDD will be the same and will be 10[degrees]C (50[degrees]F) though 18[degrees]C (-65[degrees]F) are also used. The + superscript in Equations 4 and 5 indicates that only the positive values of the bracketed quantity are taken into account in the sum. The sum is taken from i = 1 to N, where N is number of days in the individual years 1982-2006. The annual averaged HDD or CDD is then the average of the annual values for the 1982-2006 period.


Figure 1 shows the scatter plots of daily Tave, Tmax, and Tmin based upon MERRA hourly temperatures versus corresponding values from 1116 surface sites within the CONUS study-region reporting observations meeting our 85% selection criteria. In the upper left corner of each figure are the parameters for the linear-least squares regression fit to theses data, along with the MBE and RMSE between the MERRA and surface-site observations, and the gray scale bar gives the number/percentage of observations within each shade of gray. Overall the MERRA data exhibits a very high degree of correlation with the surface observations ([R.sup.2] > 0.9), and perhaps equally significant is the fact that well over 80% of the MERRA versus surface data is along a central core (i.e., nondark gray values). The MERRA Tave bias relative to the surface observations is near zero, while the Tmax and Tmin biases are -1.22[degrees]C (-2.2[degrees]F) and 0.71[degrees]C (1.3[degrees]F), respectively. Since MERRA Tave bias relative to the surface observations is near zero, this indicates the MERRA data generally reproduces the integral of the daily temperature cycle at the surface sites. However, the MERRA data does tend to underestimate the range of the diurnal cycle as shown by the biases in Tmax and Tmin.


Part of the difference in the MERRA and NCDC Tmax and Tmin values could be explained by the method used to determine the maximum and minimum temperatures. The MERRA Tmin and Tmax are derived from the lowest (Tmin) or highest (Tmax) hourly temperature data during the local midnight-to-midnight period. The NCDC Tmin and Tmax are derived from either the lowest (Tmin) or highest (Tmax) available hourly temperature observations, or more importantly from the explicit minimum (Tmin) or maximum (Tmax) temperature recorded at any time during the day and not from the available hourly temperature observations during the local midnight-to-midnight period. At the surface sites, the explicit minimum (Tmin) or maximum (Tmax) temperatures recorded at any time during the day are generally lower or higher, respectively, than the hourly temperature observations. Another contributing cause is the uncertainty of the model that produces an area estimate of these quantities to properly represent the daytime surface heating and nighttime cooling relevant to a point measurement represented by the surface station observation. Regardless, the mean biases near [+ or -]1[degrees]C ([+ or -]1.8[degrees]F) represent a useful estimate of the Tmax and Tmin temperatures in the CONUS by MERRA.

The RMSE associated with the direct comparison of the MERRA temperatures

with observational values (Figure 1) suggests an overall accuracy on the order of 2.41[degrees]C (4.34[degrees]F) (i.e., RMSE) for the daily averaged temperatures, and considerably better for over 80% of the comparison (i.e., nondark gray data pairs in Figure 1). Thus, despite the relatively coarse resolution of the MERRA data set, the MERRA data appear to adequately represent the ensemble variability of the world's surface temperature records that approach the accuracy of the measurements themselves. A caveat on this statement is that the location of a particular site relative to the grid box can produce larger differences due to small-scale weather effects, and so accurate, reliable local measurements are still preferred. However, the comparisons here show that on the whole MERRA is useful for supplementing temperature records.


Figures 2a and 2b show a comparison of the MERRA-based CDFs relative to the CDFs taken from the WDV-4 for Atlanta, GA and Toronto, Ontario. Both surface stations provided near continuous hourly data for the period 1982-2006. The tables inserted into each plot for the two stations give the MERRA and ASHRAE annual 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb temperatures estimated based upon the respective CDFs, as well as the difference (ADB). Also, the MERRA and ASHRAE annual 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb temperatures are indicated by a circle (ASHRAE) or diamond (MERRA) on the CDF curves. The average absolute [DELTA]DB and RMSE between the MERRA and ASHRAE values for Atlanta, GA USA are 0.3[degrees]C (0.6[degrees]F) and 0.4[degrees]C (0.7[degrees]F), respectively, and for Toronto, Ontario Canada, the average absolute [DELTA]DB and RMSE are 1.1[degrees]C (2.0[degrees]F) and 1.5[degrees]C (2.6[degrees]F), respectively.

The MERRA temperatures provide contiguous spatial coverage, albeit with a spatial resolution of 0.5[degrees]. Accordingly, spatial maps of design conditions can be constructed without the need for employing an averaging methodology, often used to fill in values between localized surface measurements. The top panel of Figures 3-7 shows the annual design criteria map constructed for the MERRA-based 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb temperatures, respectively; the bottom panel of each figure shows the scatter plot of MERRA values versus ASHRAE values taken from the WDV-4 for 1105 surface stations within the CONUS study-region. The maps are constructed using the dry-bulb design values based on the MERRA hourly temperature for the 6480 half-degree grid boxes contained within the CONUS study-region. Table 1 summarizes the statistical metrics associated with each design criteria. The MERRA values are based on hourly temperatures for the years matching and are used to calculate the ASHRAE values. Note that the MERRA values for the heating conditions (i.e., 99.6% and 99.0%) tend to agree ([R.sup.2] ~ 0.94) with the ASHRAE values better than the criteria for the cooling conditions (2.0%, 1.0%, and 0.4%) ([R.sup.2] ~ 0.75). This may be due in part to the larger dynamic range, or the spread in the plotted points on the scatter plots associated with the heating conditions as compared to that for the cooling conditions. The 99.6% and 99.0% values range from -35[degrees]C to 15[degrees]C (-31[degrees]F to 59[degrees]F) for a dynamic range of 50[degrees]C (90[degrees]F), while the 2.0%, 1.0%, and 0.4% values range from 15[degrees]C to 45[degrees]C (59[degrees]F to 113[degrees]F) for a dynamic range of 30[degrees]C (54[degrees]F). The 99.6% and 99.0% values are more evenly distributed on the scatter plots.

With respect to the applicability of the MERRA hourly temperatures to provide annual dry-bulb design temperatures, one metric that can be used is the agreement between design conditions derived from closely spaced surface stations. To address this question we compared the 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% design conditions for surface station pairs contained within the 0.5[degrees] MERRA grid boxes in the CONUS study-region. Figures 8a and 8b show plots of the absolute and the numerical differences between the design conditions based on observations from surface stations with elevation differences of less than 100 meters versus the separation distance between 212 station pairs. The design conditions used in Figures 8a and 8b were taken from the ASHRAE WDV-4. Note that for the stations within the various grid boxes, there is no apparent dependence on the separation distance between stations. Table 2 summarizes the statistics associated with Figures 8a and 8b: the mean of the absolute difference (<[absolute value of Diff]>, Figure 8a) between the surface-site design conditions is given in row one; the averaged (<Diff>) and corresponding standard deviation (<STD>) associated with the numerical difference between the design condition for the pairs of station is given in rows two and three; and the root means square difference (RMSDiff) between the design condition for station pairs is given in row four. The statistics in rows three and four are associated with Figure 8b. The uncertainty attached to climate design calculations using a minimum of eight years of data from ground site observations is given as [+ or -]1[degrees]C ([+ or -]1.8[degrees]F) in the ASHRAE Handbook--Fundamentals. For longer periods of observation the uncertainties can be expected to be smaller.




Comparison of the <[absolute value of Diff]> given in Table 2 to the AMBE given in Table 1 offers a measure of the accuracy of the MERRA-based dry-bulb design conditions relative to values obtained from surface observations. Comparison of the RMS differences shown in Tables 1 and 2 as a function of the dry-bulb percentage levels does show more uncertainty in the warmer categories (2.0%, 1.0%, and 0.4%) than in the colder extremes. While these two tables cannot provide a definitive measure of the performance of the MERRA temperatures at a particular location, they do suggest that in regions without observational data available, design criteria based upon MERRA temperatures can provide reasonable values, but with a larger uncertainty as shown.



Figures 9 and 10 show a comparison of the annual averaged heating and cooling degree-days based on MERRA hourly temperatures and values extracted from the ASHRAE WDV-4 for 1105 surface stations in the CONUS study-region. The top panel in each figure shows the HDD and CDD map constructed using the MERRA hourly temperature for the 6480 0.5[degrees] grid boxes contained within the CONUS study-region. The bottom panel in each figure shows the scatter plot of the MERRA versus ASHRAE values. Both the HDD and CDD use Tbase = 10[degrees]C (50.0[degrees]F). For both the annual HDD and CDD, the MERRA-based values are greater than the ASHRAE values based on surface-site observations by 85 HDD (171 HDD [Figure 9]) and 95 CDD (171 CDD [Figure 10]), respectively. However, both of these biases are well within the value of their respective standard deviations ([+ or -]268 HDD [[+ or -]482 HDD]; [+ or -]281 CDD [[+ or -]506 CDD]) shown in the figures. The sign of the bias in HDD and CDD is due to seasonal dependence of the MERRA daily average temperatures that are derived from the Tmin and Tmax. During the heating season, we find that the average daily temperatures are slightly underestimated, thus overestimating the HDD and vice versa in the cooling season (not shown). Since annual HDD and CDD do not include negative values, compensating seasonal errors that result in the statistics shown in Figure 1 are not represented as well in the HDD and CDD. Despite this, the agreement within 100 degree-days over the course of an entire year shows average daily errors <0.33 degree-days per day.




Recently, the NASA POWER project has remapped hourly MERRA atmospheric model assimilation meteorological products to 0.5[degrees] x 0.5[degrees] spatial resolution for the entire globe. To evaluate the MERRA data as a supplemental data set for the Climate Design Conditions tables offered in ASHRAE Handbook--Fundamentals, we performed an analysis of the 2-meter dry-bulb temperatures through the calculation of the CDF and approximations of the 99.6%, 99.0%, 2.0%, 1.0%, and 0.4% dry-bulb climate design temperatures, and calculated heating and cooling degree-day values based upon the MERRA temperatures. In general, better agreement was obtained in colder extremes than in the warmer extremes when considering the dry-bulb design conditions, but the annual HDD and CDD agreed well within the variability shown by the standard deviation. Thus, the MERRA-based values compare favorably to those derived from surface observation, but we still recommend the use of surface station data whenever possible, and if not available or if of questionable quality, then the MERRA data would make a suitable alternative.

The design conditions in ASHRAE Handbook--Fundamentals are commonly used throughout the building industry for many purposes--design, size, distribution, installation, and marketing of heating, ventilating, air-conditioning, and dehumidification equipment, as well as for other energy-related processes. They also form the reference climatic data used in building design standards. TC 4.2 is constantly asked to provide climatic data for more locations globally, usually in locations with few or no surface observations. The additional resource of the MERRA data can fill those critical gaps on a global scale.

Ongoing analysis of the MERRA meteorological values will be focused on completing evaluation of additional climate design parameters. In particular, moisture related parameters along with the addition of the monthly dry-bulb and wet-bulb related values. Additionally, the analysis will be expanded to include comparisons of the MERRA-based parameters to those from global surface sites. Lastly, NASA is supporting the improvement of data sets like MERRA and thus the results here for MERRA are expected to be improved for future versions.



Support for this work was provided through the NASA Earth Applied Science program under the direction of Lawrence Friedl and through a grant under a NASA program to contribute to the National Climate Assessment under the direction of Dr. Jack Kaye and Lawrence Friedl. The authors would like to thank Ms. Susan Sorlie for her help formatting the final document, Ms. Colleen Mikovitz for her help with the Grid Analysis and Display System (GrADS), and the members of ASHRAE TC 4.2 for their useful comments and suggestions in preparation for this paper.




ASHRAE. 2009a. ASHRAE Handbook--Fundamentals. Atlanta: ASHRAE.

ASHRAE. 2009b. Weather Data Viewer, Version 4.0. ASHRAE, Atlanta.

ASHRAE Technical Committee 4.2, Climatic Information. 2009.

Bloom, S., A. da Silva, D. Dee, M. Bosilovich, J-D. Chern, S. Pawson, S. Schubert, M. Sienkiewicz, I. Stajner, WW. Tan, and M-L. Wu. 2005. Documentation and validation of the Goddard earth observing system (GEOS) data assimilation system--Version 4. Technical Report Series on Global Modeling and Data Assimilation, Volume 27, M.J. Suarez, Ed. NASA/TM-2005-104606, Washington, D.C.: NASA.

Bosilovich, M. 2008. NASA's modern era retrospective-analysis for research and applications: Integrating earth observations. Earthzine, Sept. 28, 2008, 2008/09/26/nasas-modern-era-retrospective-analysis/.

Chandler, W.S., J.M. Hoell, D. Westberg, C.H. Whitlock, T. Zhang, and P.W. Stackhouse Jr. 2011. NASA's Prediction of Worldwide Energy Resource (POWER) web services. American Meteorological Society 91st Annual Meeting, January 23-27. Seattle, Washington.

Eckman, Richard S. and P.W. Stackhouse, Jr. 2012. CEOS contributions to informing energy management and policy decision making using space-based earth observations. Applied Energy 90(1):206-210.

NASA. 2012. Sustainable buildings methodology Appendix A, downscaling assimilation model temperatures.

NASA. 2013a. Prediction of Worldwide Energy Resource.

NASA. 2013b. Global modeling and assimilation office.

NCDC. 2006. Global Summary of the Day Data, Version 7. Asheville, NC: National Climatic Data Center.

Rienecker, M.M., M.J. Suarez, R. Todling, J. Bacmeister, L. Takacs, H.-C. Liu, W. Gu, M. Sienkiewicz, R.D. Koster, R. Gelaro, I. Stajner, and J.E. Nielsen. 2008. The GEOS-5 Data Assimilation System--Documentation of Versions 5.0.1, 5.1.0, and 5.2.0. Technical Report Series on Global Modeling and Data Assimilation, Volume 27, M.J. Suarez, Ed. NASA/TM-2008-104606, Washington, D.C.:NASA.

Thevenard, D.J., and R.G. Humphries. 2005. The calculation of climatic design conditions in the 2005 ASHRAE Handbook--Fundamentals. ASHRAE Transactions 111(1):457-66.

Thevenard, D.J. 2009. Updating the ASHRAE climatic data for design and standards. RP 1453, ASHRAE, Atlanta.

David J. Westberg

Associate Member ASHRAE

Paul W. Stackhouse, Jr., PhD

Drury B. Crawley, PhD


James M. Hoell

William S. Chandler

Taiping Zhang, PhD

David J. Westberg is a staff research scientist, James M. Hoell is a senior research scientist, William S. Chandler is a senior computer scientist, and Taiping Zhang is a senior research scientist with Science Systems and Applications, Inc.; Paul W. Stackhouse, Jr. is a senior research scientist with NASA Langley Research Center, Hampton, VA; and Drury B. Crawley is director of building performance with Bentley Systems, Inc., Washington, DC.
Table 1. Statistical Metrics Comparing the MERRA Versus ASHRAE
Annual Dry-Bulb Design Criteria with Matched Time Periods.

Metric                      99.6%           99.0%           2.0%

RMSE, [degrees]C         2.70 (4.85)     2.49 (4.48)     3.56 (6.41)
MBE, [degrees]C         -0.54 (-0.97)   -0.41 (-0.75)   -2.91 (-5.24)
AMBE, [degrees]C         1.99 (3.58)     1.82 (3.27)     3.13 (5.63)
Slope                        1.0            1.02            0.92
Intercept, [degrees]C   -0.55 (-0.97)   -0.19 (-0.95)   -0.40 (1.73)
[R.sup.2]                   0.94            0.94            0.75

Metric                     1.0%           0.4%

RMSE, [degrees]C        2.10 (3.78)   3.95 (7.12)
MBE, [degrees]C         0.12 (0.22)   3.22 (5.80)
AMBE, [degrees]C        1.57 (2.83)   3.35 (6.04)
Slope                      0.93           0.93
Intercept, [degrees]C   2.35 (6.50)   5.35 (11.92)
[R.sup.2]                  0.75           0.73

Table 2. Statistical Agreement Between Design Criteria Based
on Surface Station Observations of 212 Pairs of Stations
(see Figure 8)

Design Criteria            99.6%           99.0%          2.0%

<[absolute value of     1.48 (2.67)     1.28 (2.31)    1.02 (1.84)
  Diff]> ([degrees]C
<Diff> ([degrees]C     -0.38 (-0.68)   -0.29 (-0.53)   0.17 (0.31)
<STD> ([degrees]C       1.89 (3.40)     1.71 (3.08)    1.74 (3.14)
RMSDiff ([degrees]C     2.08 (3.75)     1.73 (3.12)    1.75 (3.16)

Design Criteria           1.0%          0.4%

<[absolute value of    1.10 (1.97)   1.04 (1.87)
  Diff]> ([degrees]C
<Diff> ([degrees]C     0.14 (0.25)   0.14 (0.26)
<STD> ([degrees]C      1.65 (2.96)   1.57 (2.83)
RMSDiff ([degrees]C    1.65 (2.98)   1.58 (2.84)

<[absolute value of Diff]> = Mean absolute difference,
<Diff> = Mean difference, <STD> = Standard deviation of the mean,
RMSDiff = Root Mean Square Difference between design criteria
from station pairs
COPYRIGHT 2013 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Westberg, David J.; Stackhouse, Paul W., Jr.; Crawley, Drury B.; Hoell, James M.; Chandler, William
Publication:ASHRAE Transactions
Article Type:Report
Geographic Code:7IRAN
Date:Jul 1, 2013
Previous Article:Revising ASHRAE climatic data for design and standards--Part 2: clear-sky solar radiation model.
Next Article:Principles of energy efficient ammonia refrigeration systems.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters