The UAO Observation Network — NYnet

NEW: Integrated Terminal Weather System (ITWS) — Notes on Data Assimilation Software

Introduction — Background — Goals — Existing Networks — Rooftop — Canyon — Remote Sensing — Data Assimilation
Operations Schedule
30 July 2003

Mesoscale meteorology is the science of local and regional atmospheric phenomena. These phenomena include individual storms, fronts, convective clouds and boundary-layer circulations, including topographical effects. Mesoscale systems interact strongly with microscale processes at scales of order a kilometre or less and with the large-scale global circulation. Microscale meteorology is the study of air motions on scales less than 1 km. This includes turbulence, wakes, and, in urban situations, street-level canyon circulations. Atmospheric Transport and Dispersion (ATD) covers scales from tens of meters to several kilometers or more and thus it spans from the microscale, through the mesoscale, and into large-scale weather.

Figure 1: The photo at the left shows an extreme example of how a sea breeze front can effect the boundary layer flow. The sea breeze and the sea breeze front are ubiquitous and daily features of the wind (Photo by Ralph Turncote. Simpson(1994)Sea breeze and Local Winds.) Our goal is to provide an accurate, high-resolution, real-time wind field above New York City. In order to properly describe the transport and dispersion of a ground-level atmospheric plume, wind direction and speed uncertainties must be known to better than +/- 5 degrees and +/- 1 m/s.

Real progress in our ability to predict the evolution of a gaseous plume, be it accidental or deliberate, depends on (a) model improvement and (b) observations of the plume and of the atmospheric field. (NRC 2003). Field studies and wind tunnel experiments show that in urban canyons a change in wind direction of 10 degrees can lead to a completely different pattern of transport and dispersion at street level. When the winds above the buildings are known, canyon models will be required to define street-level conditions. Lethal dose statistics, crucial for first responders, are highly dependent on accurate estimates of turbulence on all scales.

By “high-resolution” we mean a spatial scale of ~100 m in the horizontal and ~25 m in the vertical, i.e. on the scale of individual blocks. By “real time” we mean the fields must be available to modelers and first responders <1 minute of measurement. (Note: modeling virtually as they are occurring or even with a small--1 hour--prediction capability. These goals are ambitious but, with technology available today, are achievable in a reasonable time frame.

This paper provides our current plan of how we can move from the current situation, a few unreliable stations, to the goal stated above.

top

The data set that is required to support an emergency hazmat situation is much different than the normal data required to support the usual National Weather Service weather forecasting. Normal forecasting over regions of 4 to 400 km use observations that are widely spaced and sampled at hourly or more rates. Hazmat operations on the other hand need information on the scale of a few hundred meters and with updates in the 1-5 minute time scale. The transport and dispersion processes differ as we descend from above the rooftops down into the canyons, thus real information is needed on the vertical distribution of the flow and turbulence.

This requirement makes most existing meteorological networks of minimal use. The existing networks, described below, can certainly fill in and support an ATD network, but they must be used very carefully.

Figure2: The different scales of flow in the urban situation. We see here the importance of having information on the vertical and horizontal wind fields.

top

The NYC mesonet will be an assimilation of existing meteorological stations supplemented by UAO stations to provide a real-time high resolution coverage in support of the scientific and real-time requirements of the UAO New York City is the perfect, and most obvious place to develop this tool: It is a primary terrorist target, (Population of the city > 8 million, including Northern NJ and Long Island > 22 million) and thus there is an immediate need to have the best emergency tools here. Major public events (e.g. the Republican Convention in the summer of 2004) are held in NYC throughout the year. The five-borough area has minimal topography (compared to West Coast cities) so remote sensors will have the widest possible coverage.

1. Survey existing observation platforms:

Existing meteorological stations are generally spaced and operated to support synoptic meteorology and general weather familiarity.

2. Data Ingest and Archival

When existing stations have been selected as useful for the ATD network, the computer links must be established for collecting the output from the stations, storing the data in an archive, and setting up software where it can be easily used by plume dispersion software such as NARAC.

3. Add rooftop turbulence stations

Standard meteorological stations give information on the wind speed and direction in the vertical only. If they measure temperature and humidity they can be seriously in error, especially in strong sunlight. Predictions of dispersion require information on the vertical winds and the turbulence. This is especially true in the urban situation. The typical rooftop turbulence station is described below.

4. Add Sound Detection and Ranging (SODAR) at rooftops.

The SODAR (described below) will allow us to define the wind field above the rooftop turbulence. As air encounters the face of the building or swerves around obstacles on the rooftop, a local field of turbulence is created. The turbulence field can seriously skew any idea of the overlying wind field.

5. Add canyon turbulence stations

Predictions must be made of the atmospheric transport and dispersion in the canyons, at street level. We anticipate that any of a variety of models will be used to match the above-rooftop winds to street-level winds, but these models will rely heavily on actual observations.

6. Begin Software Data Assimilation

As the number and diversity of the measurements develop, software tools can be utilized to synthesize the measurements into a real-time gridded field. (A gridded field has estimates of wind speed, direction, turbulence, etc. at a regular spacing. In the urban case that might be 100 m in the horizontal and 25 m in the vertical.) Software assimilation tools are described below. As we progress, any additional stations or other measurements will directly feed the assimilation tools and improve their accuracy. The NOAA software, Local Analysis and Prediction System (LAPS), can be used immediately as part of the NOAA MADIS system (see below). As we progress, the new National Center for Atmospheric Research's Variational Lidar Assimilation Software (VLAS) will be developed to provide accurate high-resolution gridded fields with a one-hour prediction capability.

7. Add radar profilers.

The radar profilers will be used to improve our knowledge of the atmosphere in the vertical. Again, the vertical information will make significant improvement in the assimilation software and allow matching from the mesoscale to the large synoptic-scale flow predictions.

8. Add Doppler-Lidar.

As a final stage to the current plans, we will add doppler-Lidar systems. These new technologies will provide wind vector measurements with a 60-m resolution.

figure 3. Wind vectors defined by the VLAS data assimilation model that assimilated lidar data, and a plume predicted by HPAC-SCIPUFF using VLAS winds as input. The Manhattan street map is used for background as a scale reference. (Courtesy Tom Warner, NCAR) The VLAS has the capability of providing accurate, high-resolution gridded fields with a one-hour look-ahead (prediction) capability.

top

NYCnet will utilize all available meteorological stations operating in the entire city and in surrounding areas. There are many collections of meteorological stations (listed below) but it is felt that before these can truly be included in a quality meteorological grid that they must be evaluated.

CCNY/NASA Weather Network

figure 4. The CCNY network has about 30 stations throughout the NYC region. We have visited two of these stations: one on Medgar Evars College in Brooklyn and one on the CCNY science building. Both of these stations are in terrific locations but need to be improved. It is felt that we can work with CCNY to upgrade selected of these stations.

Automated Surface Observing System (ASOS)

Figure 5. The ASOS stations are basically the weather observing systems at airports. the figure of existing stations lists the wel known stations at JFK, Laguardia, and Newark airports as well as one station in downtown Manhattan (KJRB). Efforts are being made to find and inspect this station. ASOS stations are automatically fed to the NWS and are ingested by NARAC and HPAC. They provide data every hour and are widespread.

Road Weather Information System (RWIS)

Figure 6. The RWIS system is designed to help predict road conditions, especially road icing. The figure here indicates that while many stations are listed, virtually none are giving any data. Further, these stations update on an hourly schedule and thus are of limited use for ATD applications. Stations on the Triburough Bridge and elsewhere may be well exposed and with encouragement put into operation year round.

AWS Convergence Technologies, Inc. (AWS)

Figure 7. AWS is a private company who's aim is to deploy weather stations, generally to schools, and to provide web access and software that allows users to get real-time data. The figure at left is a recent rendering of the wind speed and direction available from AWs stations. We have made contacts with AWS and have plans to use their information. Nevertheless, with one station in Manhattan and only a few in the other boroughs their is usefulness is limited.

Citizen Weather Observing Program (CWOP) ♦

Presently the CWOP program brings in data from about 900 surface weather stations (many located in urban areas), places these data in the NOAA mesonet database, and makes these and other weather data available to NWS forecasters and other users. In the upper right of this page “CWOP info” gives more information on CWOP and gives an extensive list of software and other pertinent information on what a citizen would need to get a station up and working. The author of this link is Dave Helms, who works at National Weather Service Headquarters in Silver Spring, MD and is our liaison to the NWS. Dave designed our logo, and is one of our data contributors, CW0351.

The original purpose of the CWOP network was to acquire surface weather data and our data providers take various meteorological measurements and send digital data. However, with suitable transducers, the equipment could also take chemical, biological, radiological (CBR) measurements and send the digital data over the same communication system. The CWOP system architecture and computer programs could serve as a model for what ever system is developed for CBR monitoring. It has the advantage of a fairly large real-time system that is widely deployed, scalable to any size with commodity computers, and uses standard software. This means the system need not be expensive. In addition, it uses both radio and internet transmission so that the measuring stations can be mobile or fixed and still report data to emergency managers and other users in near real-time.

“APRS-IS info” gives many details on the Automatic Packet Reporting System - Internet Service, which is how our weather data packets (and many other types of data) get transferred around the country and the world. The author is Pete Loveall, who lives in Melissa, TX, and has written and maintains a Java program that is used at many different servers that handle our data. Pete is one of our data contributors, AE5PL.

“findu.com” gets about 3 million hits per month. The author of the findu.com server is Steve Dimse, who also was the original developer of the APRS-IS. In his spare time, Steve is an emergency room physician at a large Miami Hospital. The findu.com server is the public database for CWOP and thousands of amateur radio operators around the world who send information packets to APRS-IS. Steve is one of our data contributors, K4HG.

www-sdd.fsl.noaa.gov/MADIS/ ♦

The NOAA Meteorological Assimilation Data Ingest System (MADIS) is a program in the NOAA Forecast Systems Lab that acquires a wide variety of data sets and puts the data into a common data format (netCDF). The data are then made available to federal agencies, educational institutions, commercial companies and private meteorologists. This is how our CWOP data get to NWS field forecasters in a form compatible with their work stations. All of the MADIS data can be seen on the “NOAA mesonet display”. The manager of the MADIS program is Patty Miller, who is employed by NOAA here in Boulder.

Queens College/GLOBE NY Metro

The Queens college

♦ Notes by Russ Chadwick, NOAA FSL.

top

The NOAA-made rooftop turbulence system will play a major role in the NYCmesonet. The rooftop stations are used in DCnet where over 18 systems have been deployed.

Figure 8: The NOAA-made UAO meteorological station measures wind speed and direction, air temperature, humidity, and pressure to scientific levels of accuracy. In addition the three-dimensional turbulence of the wind is measured using sonic anemometers. Turbulence is an important parameter for dispersion plume models. Fifteen-minute averages are sent to the data assimilation center (DAC) by cell phone internet connections. Satellite communication is optional. The network of UAO stations plus any other meteorological data from the EPA, Weather Service, etc. are collected to the DAC, quality assured, and forwarded to be included in models of the dispersion and transport of pollutants. When combined with radiological sensors the wind measurements provide data for reverse transport calculations to identify the source of the pollutant.

top

Figure 9: The BNL canyon meteorological station. The canyon instrumentation is designed to provide measurements of the flow along the side of a building from the top of the canyon down to the street level. Complex flows in a canyon are dependent on wind direction (as measured from rooftop.) and solar heating.

top

Sound Detection and Ranging (SODAR)

Figure 10. The Sound Detection and Ranging (SODAR). Figure 11. Typical profiles from the SODAR taken in NYC. Figure 12. Example of the flow distortion around a building.

This example is from a wind tunnel evaluation of an offshore structure so the structure sees horizontally homogenious flow. A rooftop tower that is 3-5 m high is shown by the black vertical line. Wind velocities vary from 15% overspeeding to 26% underspeeding in this example. Turbulence levels and direction will be similiarly distorted. The Sound Detection and Ranging (SODAR) transmits pulses of high-frequency sound (4000 Hz approx.) in three different directions. The sound waves are scattered by atmospheric turbulence and the SODAR receives the echo. The slight change in the frequency in the returned sound is measured and used to estimate the wind speed and direction for different altitudes above the instrument.

Coherent Lidar

Figure 13. Coverage for the Coherent Lidar.	Figure 14. The Lidar in its container.

The Infrared Doppler Radar detects and tracks aerosol levels above normal conditions. For high concentrations, detection range is in excess of 15 km and detection sensitivity of a few hundred to a few thousand particles per liter is achievable. Wind data is simultaneously and directly monitored using Doppler techniques and early detection of wind shifts results in improved downwind prediction. Integration with plume dispersion models provides accurate forecasting of the plume track. Early threat warning and exposure assessment reduces contamination of critical resources. Update time is five minutes. (Courtesy of Coherent Technologies Inc.)

top

The LAPS integrates data from virtually every meteorological observation system into a very high-resolution gridded framework centered on a forecast office's domain of responsibility. Thus, the data from local mesonetworks of surface observing systems, Doppler radars, satellites, wind and temperature (RASS) profilers (404 and boundary-layer 915 MHz), as well as aircraft are incorporated every hour into a three-dimensional grid covering a 1040km by 1240km area. LAPS has analysis and prediction components. The prediction component is being configured using the RAMS, MM5, WRF, and ETA models. Any or all of these models, usually being initialized with LAPS analyses, are run to provide short-term forecasts. NOAA produces ensemble forecasts using multiple models and initialization methods, with verification.

Variational Lidar Assimilation System (VLAS)

Lidars have long been used as research tools in atmospheric science. However, recently developed model-based data-assimilation technology now allows lidar data to be used operationally to define high-resolution wind fields for use in transport and dispersion models. The modeling system, called the Variational Lidar Assimilation System (VLAS), synthesizes all available data, including lidar data, to produce three-dimensional gridded analyses and short forecasts of all meteorological variables in the boundary layer. These fields, defined on a regular grid with a spacing of 60-70 m, are then used as input to urban plume models of which there are many. Because the VLAS model has simple physics, it can produce meteorological analyses and half-hour forecasts in a couple of minutes on an inexpensive dual-processor PC. Most standard dispersion models can provide usable results in about a minute, and a major effort is underway to provide urban models, that incorporate real urban geometries, in a comparable time. Thus, the combination of lidar data, VLAS, and the urban dispersion model constitutes a valuable operational tool for forecasting the transport and dispersion of hazardous material on the urban scale. Because of the speed of the system, the meteorological variables and plume tracking capability would be made available in real-time to scientists using the Urban Atmospheric Observatory and to NYC emergency managers.

top

Note: scheduled times are approximate.

Contact R. Michael Reynolds: reynolds@bnl.gov