The Urban Atmosphere Research Program

Focus To take our forecasts and analyses to where people live and work.

Issues --

For forecasting: DispersionFloodsFrontsHeat wavesCold spellsIcing FiresAir Quality

For data acquisition and analysis: ClimateEcosystem loading

Data requirements are different for the various issues. A central requirement is for an improved capability to predict the surface boundary layer accurately, and for data to support it.

Forecasting extreme weather the probability of experiencing hazardous heat waves, extreme cold spells, icing and frontal passage (a linkage with the USWRP).

For flooding taking the spatial variability of precipitation into proper account.

For air quality the probability of exceeding potentially dangerous levels, e.g. of ozone and particulate matter (c.f. EPA/NOAA air quality forecasting).

For dispersion the probability that people will be harmed by releases directly into the local atmosphere.

For urban fires local wind behavior, with an emphasis on gustiness (and precipitation, of course)

We need to learn how to forecast, using local data as guidance. We need to be providing information that can be used. We need to avoid raising false alarms.

A sobering realization more consideration of the surface is required than for weather forecasting, and there are relatively few field studies.

At the Multi-Agency level -- the evolving Urban Test Bed Research Program(building upon the OFCM Challenges in Urban Meteorology conference, September 2004.)

The scientists already talk; the agencies are starting DHSDOEDTRA ArmyEPADOI

Main areas receiving attention Washington, DC New York (Las Vegas?)Oklahoma CitySalt Lake City

Principal groups involved National Laboratories (ANL, PNL, LLBL, LLNL, BNL, ORNL, Sandia)NOAA (NWS, NOS, ARL, ESRL NSSL); EPA (NERL)

Academic linkages So far, through individual university contacts, the AMS and the International Association for Urban Climate

Street canyons present great difficulty. What should we predict -- [C] or P([C] > [C0])?

Skimming flow is driven by the meteorology aloft. The in-canopy environment is controlled by the configuration of streets and buildings, traffic patterns, etc.Washington, DC, and New York illustrate two extremes Washington has broad streets and low buildings.

New York has deep and sometimes narrow street canyons.Physical models and computational fluid dynamics models are being used, inparallel.

Avg. bldg heightRooftop Towers (1c)TTUT Surface Towers (1a, 1b)Tier 1 -- Surface towers1a Conventional meteorological systems1b Enhanced systems1c Specialized systems

Tier 2 Remote sensing (RADAR, etc.)RADAR LIDAR (2)The proposed UrbaNet configurationTUrbaNet is DCNet in other areas (Tier 1c).A vision for measurements --http://dcnet.atdd.noaa.gov/

Some challenges for NOAA

1. Accept the dominant mission given to us to provide forecasts to protect life, property and the environment, related to the atmosphere and the oceans. This includes forecasts of dispersion.

2. Take forecasts to where the predictions affect people, where they live and work. Go probabilistic?

3. Emphasize areas where people and the environment are at risk cities, industrial areas, ports, entertainment areas, etc.

Hard reality check #1

Meteorological models are constructed from the understanding of processes, each of which is represented as an average behavior.

If the models are essentially built from understanding of averages, they should not be expected to apply except on the average.

We are no longer interested solely in the average. We need to address specific instances. Hence, the requirement for more data is extreme.

Hard reality check #2

Meteorological models are becoming increasingly refined. However in daytime the convective process in the boundary layer is largely stochastic and hence deterministic models should not be expected to reproduce behaviors on the scale of convective updrafts, except on the average.

Hence, conventional mesoscale models should not agree well with observations taken over scales that are less than several kilometers, unless these observations are ensemble-averaged.

Hard reality check #3

In theory and in practice, micrometeorological descriptions of the surface apply above about ten times the roughness length (several m) above the zero plane (~ 80% of the average structure height).

Typically RADAR cannot look low enough, and towers are not high enough.

One Modelers response: Model output often agree well with observations. This is because models have been calibrated to the observations. However, Large ensemble effort underway to quantify uncertainties from the schochastic and other poorly represented processes.

NOAA Modeling ApplicationsProvide representation of the skimming flow Incorporate urban scale parameterizations in mesoscale modelsFocused evaluation of NWS non-hydrostatic model forecasts (~4 km) with urban datasetsProvide high resolution meteorological uncertainties using ensemble techniques Explore mesoscale data assimilation techniques with urban datasets

The NMM-WRF Modeling SystemRegional-Scale, Moving Multiple Nest, Atmospheric Modeling System.

Non-Hydrostatic system of equations formulated on a rotated latitude-longitude, Arakawa E-grid and a vertical, pressure hybrid (sigma_p-P) coordinate.

Designed for 1-10 km horizontal resolution scales

Runs operationally at NCEP for:Daily High Resolution Window cyclesSpecial Fire Weather nestsHomeland Security Nests2005 inauguration, 2006 Torino Olympics

The NMM-WRF Modeling System

Metro-Watch(Gopalakrishnam)

Salient Features: Telescopic E-Grid

Large Scale portion of the flow may be easily separated from the small scale structure which may be advantageous for Hurricane analysis.

However, as pointed out by Zhang et al.(1986; MWR), for the sake of smooth solutions across the interfaces it may be necessary to sacrifice mass and energy conservation across the interface in this approach. Nevertheless, for short-term numerical forecasts in which the use of appropriate model physics and the patterns to be forecast may be important than exact mass and energy conservation, as long as the mass (or energy) discrepancy at the interface is small.Courtesy of Gopalakrishnam, EMC

Chart1

00.0002105321

00.0021053208

10.0248427852

40.1941105757

171.0480286829

444.0801116662

7911.8468505454

9426.3603003872

9446.3913749218

9467.4351087503

9483.8060831139

9493.0694945335

9497.0394979231

9497.2561354311

9496.9367582692

9496.7815961279

9496.6693825305

9396.5952752391

9396.5685376652

9396.5999069448

9396.8142285999

9396.4363235204

9392.6541147442

9384.3669405689

9369.82338464

9349.7908363808

9328.8521580591

8512.5883445231

643.6657845373

270.76402091

10.2787444709

10.1364247864

10.0570541931

10.0195794832

00.0061054303

nest maximum

nest mass

Time (h)

Scalar (%)

conserve

0010014749870.00021053211000.0002105321100

1097104741880.00210886821000.002105320899.8317848699

21961184734200.02492501371000.024842785299.6700962342

34969224726840.19505631671000.194110575799.515144625

4179649784719951.05467218931001.048028682999.3700880235

54495193804713014.11202182891004.080111666299.2239787615

679955627147064411.956170693810011.846850545499.0856591865

7949412520847004626.63739293610026.360300387298.959761004

8949422035346946846.936745422510046.391374921898.8380734631

9949432030846892068.307600443610067.435108750398.7227018845

10949339806846839484.985717152710083.806083113998.6119620116

11949344206846790794.477748783410093.069494533598.5094328897

12949346092546744098.606238233810097.039497923198.4111144094

13949246195446701498.916520703910097.256135431198.3214277443

14949246043746661898.675361859210096.936758269298.2380570416

15949245970046626298.592636757910096.781596127998.1631076219

16949145916746592798.549128940810096.669382530598.0925793759

17939145881546560898.541047404710096.595275239198.0254196431

18939145868846533498.571778550510096.568537665297.9677338538

19939045883746508498.65680178210096.599906944897.9151008343

20939045985546487298.920778192710096.814228599997.8704680339

21939045806046466598.578545834110096.436323520497.8268878938

22939044009546452394.741272229810092.654114744297.7969923387

23938940073246440186.290081201410084.366940568997.7713074253

24938933165246428971.432232941110069.8233846497.7477278326

25938923650046422350.945343078610049.790836380897.7338327154

26938813704446418529.523573575210028.852158059197.7258324965

2785885979346417112.881675072310012.588344523197.7228850474

286488174124642033.75094516841003.665784537397.7296220739

29278836294642540.78168416431000.7640209197.7403592098

3018713244643220.28514694541000.278744470997.7546753911

311876484644480.1395204631000.136424786497.7812024329

321872714645610.05833464281000.057054193197.8049925577

33187934647330.02001149051000.019579483297.8412040751

34086294649390.00623737741000.006105430397.8845736831

conserve

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

nest maximum

nest mass

Time (h)

Scalar (%)

Real-Time Mesoscale Analysis(Geoff DiMego, EMC)The RTMA is a fast-track, proof-of-concept effort intended to:leverage and enhance existing analysis capabilities in order to generate experimental CONUS-scale hourly NDFD-matching analyses OPERATIONAL at NCEP and on AWIPS with OB7 by end of FY2006establish a real-time process that delivers a sub-set of fields to allow preliminary comparisons to NDFD forecast gridsHourly temp, wind, moisture plus precip & sky on 5 km NDFD gridalso provide estimates of analysis uncertaintyestablish benchmark for future AOR effortsbuild constituency for subsequent AOR development activities

RTMA ProcedureTemperature & dew point at 2 m & wind at 10 mRUC forecast/analysis (13 km) is downscaled by FSL to 5 km NDFD gridDownscaled RUC used as first-guess in NCEPs 2DVar analysis of ALL surface observationsEstimate of analysis error/uncertaintyPrecipitation NCEP Stage II analysisSky cover NESDIS GOES sounder effective cloud amount

RTMA LogisticsHourly within ~30 minutes 5 km NDFD grid in GRIB2Operational at NCEP Q3 FY2006Distribution of analyses and estimate of analysis error/uncertainty via AWIPS SBN as part of OB7 upgrade end of FY2006Archived at NCDC

Real Time Mesoscale Analysis

SummaryUrban mesonets can be useful now for:Development and evaluation of urban parameterizations in mesoscale modelsAssimilation into real-time mesoscale analyses and into high resolution nested modelingQuantify uncertainties from ensemble predictions systems (eg: JEFS, SREF, other HR ensemble)

BACKUPS

Hard reality check #4

People live within the surface canopy of trees and buildings, where the mesoscale models will not apply except on the average, and where local data are highly influenced by the local distribution of roughness elements and/or obstacles.

We need to learn how to forecast for the canopy boundary layer, below the height at which the mesoscale models apply.

Hard reality check #5

There are plenty of data available, from instruments erected within the canopy by transportation departments, environmental agencies, private companies, for example. These are usually situated within the canopy layer. Their data are not likely to be of use in the context of conventional mesoscale models, but it is not these conventional models that are appropriate for the mission at hand. Hence, we must learn how to extract relevant data from the large number of observations now available.

Hard reality check #6

Date from within the area of interest are not sufficient, by themselves. There needs to be supporting information from the surrounding area, mostly upwind.

Hard reality check #7

We must re-think the issue of data acceptability. Synopticians emphasize representativeness, so that their models and the data have a chance of converging. They insist (correctly) on data well above the canopy, and not influenced by local obstacles.

The present application emphasizes the need for data indicative of conditions within the canopy. We need data that are accurate representations of conditions where the instrumentation is situated.

We need to think in terms of micrometeorological data quality control, not data quality assurance as designed for weather forecasting purposes.

Q1: Is the instrumentation reporting what it experiences?

Q2: Is the data stream useful?

Rethink tower data acceptability:

There are two main questions. Are the data correct, and are they useful? Correctness is essentially a Point of Origin question.

Conduct Point of Origin data assurance,

-- Perform scheduled calibrations.

-- Search for data drop-outs.

-- Inspect the continuous data record and search for periods of constant output (indicative of sensor failure and/or intermittency)

-- Report at every reporting interval the statistics generated from the raw data:

Averages and standard deviationsProportion of time with unchanging output from sensor

Data Usefulness is a Point of Application question.

2. Conduct Point of Application data utility tests,

-- Apply such data quality criteria as may be relevant for the specific application.Scrupulously avoid rejecting data because they fail to meet the requirements of a different application.

-- Perform tests as needed to demonstrate forecasting improvements if the data are used.

-- Interface with the data network authorities to optimize the data collection/transfer/utilization processes.

-- Subject unwelcome critics to such torture as might be intellectually satisfying.

Some relevant heresies:

Instead of nudging forecasts with a few observations, why not rely on nowcasts and adjust these according to predicted changes?

I.e. Start with the best depiction of current reality, and construct expectations using the current situation as the foundation.

The vision synopticians will forecast skimming flow, local experts will refine forecasts to fit local situations.

Remember Meteorology is still somewhat an art. Like all artists, meteorologists tend to fall in love with their models.