ASHRAE ST-16-019-2016 Mesoscale Climate Modeling Procedure Development and Performance Evaluation.pdf

《ASHRAE ST-16-019-2016 Mesoscale Climate Modeling Procedure Development and Performance Evaluation.pdf》由会员分享，可在线阅读，更多相关《ASHRAE ST-16-019-2016 Mesoscale Climate Modeling Procedure Development and Performance Evaluation.pdf（16页珍藏版）》请在麦多课文档分享上搜索。

1、186 2016 ASHRAEThis paper is based on findings resulting from ASHRAE Research Project RP-1561.ABSTRACTThis paper presents the results of ASHRAE Research Proj-ect RP-1561, Procedures to Adjust Observed Climatic Data forRegional or Mesoscale Variations. This project included aWeather Research Forecast

2、 (WRF) modeling campaigndesigned to cover ten significant climate regions across NorthAmerica. Model results were compared against mesoscalemonitoring data to assess the models performance for a singleyearshourlyweather.Subsequently,along-termclimatemodelevaluationwasperformedbyrunningWRFoverfourreg

3、ionsinNorth America for eight years. Overall, the model performedwell against observed temperature and humidity, reasonablywell against observed wind, and relatively poorly againstobservedsolarradiationandprecipitation.Guidedbythiseval-uation, a complete mesoscale numerical modeling procedurewas dev

4、eloped for coastal regions, mountain valleys, mountainplateaus, and major city centers to provide site-specific climatedata (i.e., a freely available software solution for developinglocalized climate data).INTRODUCTIONCurrent architectural and engineering practice involvescareful consideration of lo

5、cal meteorology as a key factor inmanydesignprojects.Parameterssuchaswindspeed,tempera-ture, humidity, precipitation, and incoming radiation can allinfluence a buildings design and ongoing performance.ASHRAE members typically use climate data derived frommultiyear (minimum 8 years, typical 30 years)

6、 measured data,provided by ASHRAE for locations worldwide (ASHRAE2013).However,thesedataareoftentakenfrommeteorologicalstations, which may be remote from a study site by tens tohundreds of kilometers or perhaps in completely differentterrain(urbanversusrural,mountainversusvalley),whichmaynot be repr

7、esentative. Factors such as urban heat island effects,sea/land breezes, and varying terrain can all significantly altermeteorological conditions between a weather station (often atan airport outside of an urban center) and an actual project site.Alternatively,theremaybemanystationslocatedincloseprox

8、-imity to a design site and the user must select the most appro-priatedatasource.Forinstance,thethreemajorairportsservingNew York City (JFK, LaGuardia, and Newark) are all locatedwithina15km(9.3mi)radiusofthecitycenter(Manhattan)andhave a 2.4C (4.3F) difference in 0.4% cooling dry-bulbtemperature, a

9、 difference of 265 cooling degree days (CDD65)and a difference of 225 heating degree days (HDD65). Whilethis example may account for 10% differences in degree days,considering the number of HVAC systems in the city designedto meet the conditions, the accuracy of climate data becomesvery important fr

10、om an energy conservation and efficiencyperspective. The challenging question is which airport datashouldbetreatedasrepresentativeofasiteonManhattanIsland.There are a number of techniques available to derive themeteorologyofagivenasiteintheabsenceofasuitableobser-vational station. These techniques o

11、ften consist of some formof interpolation or more complicated aerodynamic physicalprocess/modeling to downscale data to a finer resolution. Theinterpolation approach can be as simple as spatial linear inter-polation of neighboring observational stations or morecomplex, such as forming linear regress

12、ion models based onpredictors accounting for elevation or proximity to a coast(e.g., the PRISM model Daly et al. 2008). At the far end ofMesoscale Climate Modeling ProcedureDevelopment and Performance EvaluationXin Qiu, PhD Michael Roth, PhD, PEngMember ASHRAE Member ASHRAEHamish Corbett-Hains, PEng

13、 Fuquan Yang, PhDXinQiuisaprincipal,HamishCorbett-Hainsisanairqualityengineer,andFuquanYangisanairqualityandmeteorologymodeleratNovusEnvironmental Inc., Guelph, Ontario, Canada. Michael Roth is a director at Klimaat Consulting it can be difficult to execute themodel without fully understanding the p

14、hysics schemes anddynamic mechanisms that are being used for various weatherconditions. Additionally, significant computational resourcesare required to execute these models. In an effort to provideengineers and designers with more accurate meteorologicaldata, a methodology has been developed for mo

15、deling site-specific meteorological data with next-generation meteoro-logical models, potentially on desktop computers.ASHRAE RP-1561, Procedure to Adjust ObservedClimatic Data for Regional or Mesoscale Climatic Variations,was created with two main goals in mind:To develop a methodology for ASHRAE m

16、embers toharness the power of modern mesoscale modeling tech-niques in order to derive meteorological conditions spe-cific to a study areaTo evaluate this methodology against available meteoro-logical and climatic observational data in a variety ofgeographic categories, including coastal, mountain v

17、al-ley, mountain plateau, and major cityHere, an overview of the modeling methodology isprovided and evaluated based on performance in estimatingboth hourly meteorology (i.e., the weather) and designelementssuchasthe99.6%heatingdry-bulbtemperature(i.e.,theclimate).Adiscussionandevaluationofasimplifi

18、edmeth-odology designed for ASHRAE members with intermediatecomputer skill levels is also provided. This paper provides acondensed version of the project final report (Qiu et al. 2015).MODELING PROCEDUREThere are several numerical meteorological models,commonlyusedwithinthemeteorologicalcommunityfor

19、high-resolution weather forecasting, which are capable of producinghigh-quality gridded hourly climate data. The WRF model wasused in this study due to its ability to generate high-resolutionand reliable climate data at any location within its modelingdomain. The WRF model is a next-generation mesos

20、calenumerical weather prediction system designed to serve bothoperational forecasting and atmospheric research needs. Themodel is suitable for a broad spectrum of applications acrossscales ranging from meters to thousands of kilometers (yards tothousands of miles). The WRF model is developed and mai

21、n-tained as part of a collaborative effort principally among U.S.government agencies, universities, national laboratories, andinternational communities (Skamarock et al. 2008).Domain ConfigurationThemodelingprocedurebeginswiththeestablishmentofthe model domain. Mesoscale modeling is three-dimensiona

22、l;the methodology divides the atmosphere from the groundsurface to the top of the troposphere, around 100 hPa(1.45 psi),into35verticallayersandhorizontallybygridcellsin kilometers (miles) covering an entire domain. Domainsdefined in this paper are very large based on the requirementsof the scope of

23、work in RP-1561. The final recommendeddomain sizes for ASHRAE members can practically be muchsmaller than those in this study.The WRF model takes a nested approach where a largecoarse-resolution domain feeds into a small but fine-resolutiondomain. For example, the parent or first domain spans approx

24、i-mately 1800 1800 km (1120 1120 mi) with 36 36 km(22.5 mi 22.5 mi) cells. Nested in this domain is a medium-resolution child domain, which is 730 730 km (450 450 mi).Nested in this domain is an even finer-resolution child domain,which is 280 280 km (170 170 mi). The final grid resolutionin the smal

25、lest domain is approximately 4 km (2.5 mi).Input DataModel topography was interpolated from the 30 arcsec(about 1 km 0.6 km) USGS geophysical data center globaldata coverage, which are included with the WRF preprocess-ing system. Vegetation type and land use data at 1 km (0.6 mi)horizontal resolutio

26、n were also interpolated in to the WRFgrids.The WRF model must be coupled with the output of alarger three-dimensional reanalysis data set, one with greaterarea coverage:The larger model is interpolated in to the smallerregional model and serves as the initial state of theatmosphere, (i.e., the init

27、ial condition).The larger model also provides the air masses that enterthe domain along the lateral boundaries as the simula-tion evolves (i.e., the boundary conditions).Finally, the large model also constrains the coarsestdomain such that the model simulation does not drift toofar from reality; thi

28、s is termed nudging.Here, the reanalysis data set used in this study was the32 km (22.5 mi) North American Regional Reanalysis(NARR) data set (Mesinger et al. 2006). While NARR wasused throughout this project as a WRF model input, readilyavailable alternatives include Climate Forecast SystemReanalys

29、is (CFSR) (Saha et al. 2010) and Modern Era Retro-spective-Analysis for Research and Applications (MERRA)(Rienecker et al. 2011), two data sets with global reach.Published in ASHRAE Transactions, Volume 122, Part 2 188 ASHRAE TransactionsTheWRFmodelisalsoabletoassimilateavailableupper-airand surface

30、-station observational data to provide increased accu-racy and solution stability. While this observational nudging wasused throughout the evaluation, it is not used in the simplifiedprocedure discussed later in “Simplified Methodology” section.Model PhysicsThe physical modules applied in the model

31、were selectedbased on a series of sensitivity simulations to provide a solu-tion that adequately represents the regions studied:Cumulus parameterization: The Kain-Fritsch scheme(Kain and Fritsch 1993) was applied to coarse WRFdomains (36 and 12 km 20 and 7.5 mi) to parameterizeor represent cloud pro

32、cesses not able to be explicitlyrepresented at those scales. However, this parameteriza-tion was turned off for the fine 4 km (2.5 mi) domainsas, at this scale, clouds are partially resolved explicitly.Planetary boundary layer scheme: Planetary boundarylayer (PBL) physics models estimate low-level w

33、ind,temperature, turbulence, cloud cover, and radiation. TheYonsei University (YSU) scheme (Hong et al. 2006) wasselected.Explicit moisture scheme: The explicit moisture schemeis used to resolve cloud formation and precipitationwithin a grid cell for the fine grids (4 km 2.5 mi).Here, we used the si

34、ngle-moment WSM6 scheme (Hongand Lim 2006).Radiation schemes: For longwave radiation, the rapidradiative transfer model (RRTM) (Mlawer et al. 1997)was used. For shortwave radiation, the Dudhia scheme(Dudhia 1989) was used.Land surface scheme: The commonly used unified Noahland surface model (Chen an

35、d Dudhia 2001)a five-layer thermal diffusion schemewas used to simulatesurface skin temperature.It should be noted that for a given region, it is possible thata different selection would better represent the local climateconditions. For example, the solar radiation schemes recom-mended here do not i

36、nclude the regional impact of aerosol andozone. If solar radiation is of great importance, much morecomplexschemesareavailable.Inaddition,theexplicitmoisturescheme was tested and selected differently in the Orlando, Flor-ida,studyareatoappropriatelyaddressprecipitationsimulation.Model SimulationFirs

37、t, a multiyear run is split into a number of three-dayperiods (i.e., 72-hour chunks). Each period is preceded by a12-hour spin-up period. For example, a single period may beset up to run from January 1 to January 3, 2014. The spin-upadded on the front of this simulation is a 12-hour portion(12:00to2

38、4:00inDecember31)thatallowsthemodeltoreacha balanced state with the boundary conditions. Thus, the totalrunperiodbecomes84hours.Beyondthisspin-uptime,areal-istic flow field develops from which a usable portion of thesimulation is generated. Splitting the simulation into fullycontained, independent p

39、eriods allows minimizing modelcumulative(systematic)errorsandrunningdifferenttimeperi-ods in parallel, if resources are available. That is, the simula-tion does not have to be run in sequence, saving a significantamount of time.PostprocessingThe output of the WRF simulations are a massive set offile

40、s (1 TB/year/study area) consisting of hourly three-dimensionalfields.Astheultimategoalofthemethodologyisto produce a time series of the relevant meteorological quan-tities at the surface, these three-dimensional files need to bemined. Derived products such as typical meteorological year(TMY)filesor

41、climaticdatasheetsfollowing ASHRAE Hand-bookFundamentals criteria (ASHRAE 2013) can be gener-ated from these data sets.METEOROLOGICAL EVALUATIONOne of the main objectives of this project was to deter-mine if the WRF model and proposed methodology suffi-ciently approximate meteorological data to ASHR

42、AEsstandardsfordistribution.Furthermore,thisevaluationsoughtto provide reasonable estimates of the models biases anderrors. To address these objectives, it was necessary tocomplete both an operational evaluation (i.e., quantitative,statistical, and graphical comparisons) as well as a morephenomenolo

43、gical assessment (i.e., qualitative comparisonsof observed features versus their depictions in the model).In this section, annual simulations (2008) from multipledomains were evaluated to determine how well the proposedmethodology correlates against the hourly observed meteoro-logical data (i.e., th

44、e weather). In the “Climate Evaluation”section, we discuss using a series of eight-year simulations toassess how well the methodology correlates against theobserved statistical data (i.e., the climate).Domain ConfigurationThe WRF model was evaluated over eight separatedomainswith4 km(2.5mi)gridresol

45、utionwithinNorthAmer-ica (see Figure 1). The eight domains shared five large 36 km(22.5 mi) grid resolution domains and five middle domainsconsisting of 12 km (7.5 mi) grid resolution (not shown). Thesmallest domains that were used in the WRF model evaluationroughly centered on New York, Orlando, Ne

46、w Orleans, LosAngeles, San Francisco, Denver, and Seattle have approxi-mately150160gridcells(600640km360394mi),eachwith 4 by 4 km (2.5 by 2.5 mi) grid resolution.Performance CriteriaThe WRF evaluation focuses on comparisons of specificmodeled hourly meteorological parameters to observedhourly data.

47、Performance criteria and benchmarks for theWRF meteorological model results were based on the evalu-Published in ASHRAE Transactions, Volume 122, Part 2 ASHRAE Transactions 189ation protocols of Zhang et al. (2006) and Wu et al. (2008),whicharerecommendedfortheevaluationofmesoscalemete-orological mo

48、dels such as WRF. The suggested criteria fortemperature, humidity ratio, wind speed, and wind directionforbothregularandcomplexterrainsaregiveninTable1.Themetrics used in this study are (a) correlation coefficient (Ra measure of the linear correlation between two variables),(b) mean bias (MBthe aver

49、age difference between two datasets), (c) mean gross error (MGE, a measure of the absolutedifference between two data sets), and (d) root-mean-squareerror (RMSE).Observational Mesonet DataThe Meteorological Assimilation Data Ingest System(MADIS) mesonet (NOAA 2014) network of meteorologicalstationswaschosenastheobservationaldatasetgivenitshighspatial resolution and data availability. The MADIS m