...From the Director's Office Dr. A.E. MacDonald
New Supercomputer for FSL The Department of Commerce Secretary William M. Daley announced late last year that FSL would receive an HPTi supercomputer, one of the fastest computer systems in the world. The $15 million contract was awarded to High Performance Technologies, Inc. (HPTi) of Reston, Virginia. When the system was installed in December, it began running a third of a trillion arithmetic operations per second, providing a computer system that is 20 times more powerful than FSL's old computer system. By the final upgrade in 2002, it will be processing about four TeraFLOPS of data, or four trillion arithmetic computations per second.
The supercomputer will greatly advance FSL's weather forecasting research, particularly the development of next generation mesoscale models such as the Weather Research and Forecast (WRF) model, a collaborative effort with NOAA's National Centers for Environmental Prediction (NCEP), the National Center for Atmospheric Research, and various universities.
The computer will be used to develop new versions of FSL's RUC and LAPS models, providing substantial improvements to their accuracy. The North American Atmospheric Observing System (NAOS) will also benefit from the HPTi system. The accuracy of current weather forecasts is limited to a great extent by incomplete knowledge of current conditions. The high performance supercomputer will allow scientists to conduct experiments with computer forecast models to optimize the design of future atmospheric observing networks with respect to cost and forecast accuracy. The HPTi system will be used 40% for weather prediction models, 40% for NAOS, and the remaining 20% will be available for other NOAA research laboratories to use for developing ocean models and other modeling efforts.
Photo of installation of FSL's supercomputer in December 1999.
FSL Employees Receive Awards The following FSL employees received awards last year:
Photo of FSL staff receiving the DOC Gold Medal last year.
...From the Forecast Research Division Dr. Thomas W. Schlatter, Chief
NAOS Notes The NAOS program has been testing a hypothesis during the past year or so with three numerical prediction models operational at NCEP: the global spectral model, the Eta model, and the Rapid Update Cycle (RUC) model. The hypothesis states that "it will be possible to reduce the number of rawinsondes in the U.S. network without noticeably reducing forecast accuracy, provided that the sites removed have substitute observing systems already in place." In practice, 14 rawinsonde sites have been identified near busy airports where landing and departing aircraft deliver frequent slant soundings of wind and temperature.
Wintertime tests covering late January through February 1998 are complete. Each model assimilates an operational mix of observations and makes periodic forecasts. There are three different runs with each model: the first, a control run, includes all operational data sources; the second deletes only the 14 selected rawinsondes; the third deletes not only the 14 rawinsondes but also the nearby aircraft soundings. Forecasts from all three models are rather insensitive to the loss of the 14 rawinsondes. A notable exception is that the analyses and short-term forecasts of moisture are adversely affected by the loss of humdity data, but the effect quickly disappears as the forecasts lengthen. Rapid Update Cycle forecasts of wind and temperature were degraded about two-thirds of the time when rawinsondes were withheld, but the effect was quite small.
Though aircraft soundings, which do not yet include measurements of moisture, seem to compensate for the loss of rawinsondes, one should not conclude that moisture observations are useless. Since the moisture field is severely undersampled, the definition of the moisture field is inadequate. In fact, the U.S. Weather Research Program and NAOS both recognize the need for a comprehensive program to better sample the moisture field. No recommendations for closing rawinsonde sites are warranted, at least not until a substitute for moisture soundings can be guaranteed.
Progress in testing Hypothesis 1 has been possible only because NASA has generously donated time on its Cray J-90 computer; even so, results have been slow. Much more rapid progress in testing hypotheses approved by the NAOS Council is now expected. NCEP's acquisition of a powerful new IBM SP computer for operational numerical prediction will accelerate the pace of NAOS testing there. FSL's acquisition of its own new supercomputer is also expected to speed up results.
FSL Researchers Mentor COMET Trainees FSL scientists continue to participate in mentoring programs such as the NOAA-sponsored Mesoscale Analysis and Prediction (MAP) course presented by the Cooperative Program for Operational Meteorology, Education, and Training (COMET). Referred to as COMAP'99, this course supported National Weather Service Science and Operations Officers in training their staff to use advanced datasets and apply cutting edge scientific methods in operational weather forecasting. COMAP was designed to increase graduate students' knowledge of mesoscale meteorology and new observing systems, enhance forecasting skills, and provide mentors for independent research activities.
Last year Drs. Stanley Benjamin, John Brown, Cecilia Girz, and Edward Szoke provided guidance for four students who worked on independent research projects. These and other FSL scientists enjoy the rewards offered through NOAA mentoring programs.
...From the Regional Analysis and Prediction Branch Dr. Stanley G. Benjamin, Chief
Development of an Assimilation Technique for GOES Cloud-Top Pressure for RUC/MAPS Drs. Dongsoo Kim and Stanley Benjamin have developed and tested in real-time a preliminary technique for assimilation of a cloud-top pressure product from NESDIS. This technique has now been tested in real-time parallel cycles over the last 10 months, and its performance has been examined for both warm-season and cold-season environments. This cloud-top assimilation procedure combines the GOES cloud-top data with the previous 1-hour RUC/MAPS three-dimensional forecast of cloud water/ice and includes both cloud-clearing and cloud-building. Water vapor fields are modified, along with hydrometeor mixing ratio fields, depending on the GOES cloud-top data. Verification results in all seasons show a strong positive impact on subsequent cloud-top forecasts for 1-hour and 3-hour predictions, and a weaker positive response out to 12-hour predictions. The effect of cloud-top assimilation for forecasts of relative humidity with the current technique has been shown to be seasonally dependent. While slight positive effect on 3-hour relative humidity forecasts was demonstrated in AugustSeptember 1999, the effect in February 2000 has been a slight degradation on short-range relative humidity forecasts. Precipitation forecasts, especially at 3-hour duration, show improvement with the cloud-top assimilation during the winter season.
The nature of cloud forecast errors in the control and parallel cycles was further examined with histograms of cloud-top pressure differences, contingency tables for MAPS/GOES and clear/cloudy conditions, and spatial maps from an automatic classification technique for determining dominant modes in situations for which MAPS failed to predict observed clouds. Examples of these techniques were shown for a specific case, revealing remaining problems with the proper use of GOES cloud fraction and the prevention of overclearing from GOES.
In the case examined from August 1999, the number of cloudy grid points (7114) was somewhat larger than the number of clear grid points (4514). This proportion, even from only one case, indicates that cloud clearing alone significantly underutilizes the GOES cloud product for numerical weather prediction initialization. Therefore, it is desirable to unify the treatment of both cloudy and clear columns in a single initialization procedure. Our results and experience also confirm the need to understand errors or ambiguities in the GOES cloud product that sometimes occur with low-level or semitransparent clouds.
Despite the generally positive results up to this point from testing of this preliminary cloud-top assimilation, considerable additional development is planned, including incorporation into a one-dimensional/three-dimensional variational analysis, inclusion of radar reflectivity, surface cloud, and lightning observations. This work is being sponsored in part by the FAA Aviation Weather Research Program and by NOAA's Office of Global Programs. The Cooperative Institute for Meteorological Satellite Studies (CIMSS) at the University of Wisconsin has provided valuable collaboration in providing cloud-top data and in analyzing test results.
...From the Meteorological Applications Branch Dr. Cecilia M.I.R. Girz, Chief
Global Air-ocean IN-situ System (GAINS) GAINS tests were performed at the Tillamook, Oregon, Hangar from 1122 October 1999. Cecilia Girz and Robert Anderson worked with personnel from GSSL, Inc. conducting hangar tests and a variety of checkouts on the GAINS systems in preparation for a flight test of GAINS radios. During the first week the secondary terminate device was tested. FAA Aviation regulations require that balloons have two independent terminate mechanisms. FSL and GSSL tested the helium vent mechanism on a previous trip, and the rip button mechanism was successfully tested in the hangar with the 16-feet prototype II balloon on this trip.
During the second week a sequence of critical preflight activities was accomplished. The FSL radio command terminate package was incorporated with a backup GSSL timer for the top instrument package. Next, the FSL instrument package (providing data transmission and secondary terminate functions) was incorporated with the GSSL radio command terminate package for the bottom package. Top and bottom instruments were then integrated with the 16-ft balloon, and an indoor inflation was performed. FSL groundstations were also installed in three operations vehicles.
For flight operations, the base station van was located at about 3,140 feet near the top of Mt. Hebo, the highest local point on the Coast Range in the vicinity of Tillamook. Mt. Hebo allowed line-of-sight communications from launch at the Tillamook hangar into the Willamette Valley, on the east side of the Coast Range. The van collected line-of-sight data from the FSL telemetry radios and an onboard real-time video (RTV). FSL groundstations were also installed in the four-wheel drive recovery vehicle and a chase aircraft; both vehicles are equipped to record telemetry and RTV and have a terminate capability. Crucial functions were thoroughly checked: the uplink radios for termination, downlink radios for data transmission, and voice communication channels for coordination among the ground vehicles and the aircraft. Successful completion of these activities finalized the preflight preparations, and by the end of Thursday, 21 October, the way was clear for a flight.
Unfortunately, the winds were no longer cooperating. To ensure recovery of the balloon and its payload, winds from the northwest were needed to advect the balloon over the Coast Range so that it could be dropped into the flat agricultural area of the lower Willamette Valley. On our final day in Tillamook, FSL's balloon trajectory code (written by Randy Collander) indicated a north- northeasterly track for the balloon. GSSL's local terrain expert, Jim Hurliman, believed that there was less than a 50% chance of recovering the balloon because of the steep terrain, tall trees, and few roads for all points along the forecast track. Consequently, the flight had to be postponed, and tentative plans were made to return later in the year.
During the week of 20 December 1999, the coastal Oregon skies cleared after weeks of fog, rain, and overcast. Taking advantage of several days of blue skies and cooperating winds that had not been seen there since mid-October, GSSL conducted a test with the PII-LF (15-feet) prototype on 23 December. The objective of the one-hour flight was to test the termination and descent performance of the balloon. The balloon was launched from the Tillamook airport at about 8:30 AM, Pacific Standard Time, terminated approximately 30 minutes later at an altitude of ~28,000 feet, and descended to land approximately 2.5 miles from the airport. Chased by an aircraft and a ground crew, the envelope and equipment were located on the ground in good condition.
Testing New Techniques for Improving Heavy Rainfall Warnings in Hungary Cecilia Girz, Edward Tollerud, Agnes Takacs, and Sandor Kertesz are collaborating on a research project toward improving heavy rainfall warnings in Hungary.
Severe weather causes substantial loss of life and material damage in Hungary. Extreme rainfall can produce flash floods or produce storm sewer overflows in urban areas. These heavy precipitation events are often mesoscale in nature, presenting important challenges to forecasters. Methods are being investigated to identify and monitor the evolution and precipitation of potentially dangerous systems as early as possible. Satellite rain estimates can play a key role by providing timely updates to these forecasts.
The goals of the research, funded by the United States-Hungarian Science and Technology Joint Fund, are to improve forecasting of extreme rainfall events and to increase the timeliness of heavy rainfall warnings at the Hungarian Meteorological Service.
We have implemented and tested an automated satellite rain estimation technique using METEOSAT observations daily to compute precipitation fields over Hungary and neighboring countries. Another potential of these estimates is providing first-guess fields and updates for heavy rainfall prediction techniques (especially the Probable Maximum Precipitation, or PMP methods) that rely on soundings made at 12-hour synoptic intervals.
In preliminary studies using 1998 cases, the Possible Maximum Precipitation (PoMP) method, which uses the Griffith-Woodley satellite rain estimation technique and a modification of the PMP method, indicated success in delineating regions of heavy precipitation. The PoMP method provides a first-guess estimate for the amount of rainfall during episodes of heavy precipitation, and the satellite estimates provide updates. The 1998 results encouraged continuation of analyses on the 1999 cases. PoMP estimates are likely to be more realistic than output from NWP models, which tend to underestimate rainfall in these situations. Operational runs of the satellite rain estimation technique timed to update the PoMP fields during potentially serious rainfall episodes could add critical lead times to heavy rainfall warnings. This is particularly true when soundings from NWP models like Aladin are used to produce the necessary model adjustment factors at times that are several hours removed from synoptic radiosonde observations. Ultimately, the hope is to provide warnings that are more timely and accurate, thus more likely to save lives and property.
...From the Local Analysis and Prediction Branch Dr. John A. McGinley, Chief
LAPS Part of Lockheed Martin Missiles and Space (LMMS) Research Project FSL is collaborating with LMMS researchers and the San Jose State University in an investigation of the retrieval of advanced data products derived from polar- orbiting meteorological satellites for enhanced mesoscale cloud and moisture analyses and subsequent numerical weather prediction forecasts.
The Local Analysis and Prediction System (LAPS) is being configured in the LMMS Center for Remote Environmental Sensing Technology to use polar-orbiting satellite data to augment the data stream while providing valuable cloud cover and vertical structure information to improve subsequent analyses. The utility of high-resolution satellite-derived moisture data on mesoscale data is being evaluated. Moisture fields retrieved from the Advanced Very High Resolution Radiometer (AVHRR) and microwave moisture sounders were integrated into LAPS analysis and forecast fields generated by LAPS and the MM5 model.
The architecture needs further development for implementing more complex cloud properties from polar-orbiting satellites into LAPS. Nonetheless, preliminary results are encouraging regarding the ability to contribute useful cloud and moisture information from polar-orbiting satellites into LAPS and any mesoscale model, according to an LMMS report.
...From the Demonstration Division Margot H. Ackley, Chief
"Safety First" Emphasis at National Weather Service Training in Boulder Last September FSL hosted a two-day Fall Protection Course sponsored by the National Weather Service (NWS) for technicians, managers, and regional safety supervisors. Along with the modernization of the NWS came increased numbers of meteorological sensors and antennas, some of which are placed on towers up to 100 feet high. The pilot stage of the course addressed safety issues relating to climbing high towers housing weather equipment. Class participants learned about fall hazard situations, means of protection, OSHA (Occupational Safety and Health Administration) standards and NWS procedures, and how to use the fall protection systems. The classroom was staged with a heavy-duty tripod for testing harnesses, lanyards, vertical lifelines, and positioning devices during the first day of training. The second day was spent in the field at the Platteville, Colorado, tower. (All towers have been equipped with fall protection fixtures to complement new safety equipment required to be worn by installation and maintenance workers.) After each class, the attendees are certified and receive a fall protection harness, provided by NWS. Two FSL Profiler Program staff, Seth Gutman and Brian Phillips, received certification at the Boulder pilot class.
According to Michael Jacob of the NWS Office of Systems Operations, refinement of the Fall Protection Course is the finishing touch toward a complete set of NWS safety procedures developed jointly with OSHA. Analyzing feedback from the class participants is the next step, then a schedule will be developed, and formal training will begin in October and continue throughout the fiscal year for about 460 workers. Completion of the training sessions will culminate a well-planned program designed to bring all NWS regions into conformity and provide safety procedures that should result in a reduction of accidents, 79% of which have been attributed to falls, according to the Bureau of Labor Statistics.
Photo of participants at the NWS Fall Protection Course held in Boulder practicing use of the safety harness in the field.
Profiler Network and RUC-2 Data Helpful During Oklahoma/Southern Kansas Tornado Outbreaks The performance of profilers and the Rapid Update Cycle is discussed in a service assessment report prepared by the National Weather Service on one of the largest tornado outbreaks in history, which struck west-central Oklahoma and southern Kansas on 3 May 1999.
The report stated that the Severe Weather Outlook was upgraded to a moderate risk for severe thunderstorms in the Oklahoma-Kansas outbreak area based on 7:00 a.m. soundings, midday profiler observations, and RUC-2 forecasts of instability and shear.
NWS reported that from midmorning to midafternoon, the 1-hour, 250-meter interval vertical wind profiles from the profiler at Tucumcari, New Mexico, showed a descending and strengthening jet approaching Oklahoma. The jet was deeper, stronger, and lower in the atmosphere than forecasts from numerical models and favored development of supercells.
It was profiler data that led forecasters at the Storm Prediction Center to upgrade the Severe Weather Outlook from moderate to high risk for severe weather in the outbreak area and caused F2 or stronger tornadoes to be highlighted in the Experimental Probabilistic Outlook. The report stated that, in the opinion of the Service Assessment Team, without profiler data forecasters would not have upgraded from moderate to high risk. Also, the state of readiness of NWS offices, emergency managers, and the media in the severe weather outbreak area would not have been as high.
It was recommended in the report that NWS make a decision on how to support the existing profiler network so that the current data suite becomes a reliable, operational data source.
...From the Aviation Division Michael J. Kraus, Chief
Real-Time Verification System Exercises Last year was a busy year for the Real-Time Verification System (RTVS), which provides a mechanism for managers and forecasters to evaluate the quality of the current products. Several major verification intercomparison exercises were conducted using RTVS. The first was a turbulence exercise was conducted from 21 December 199831 March 1999. Fourteen clear-air turbulence forecasting algorithms based on the RUC-2 numerical weather predication model were verified in real time with statistical results provided on the Web.
Check out the Aviation Digital Data Service (ADDS). ADDS provides the aviation community with digital and graphical analyses, forecasts, and observations of key meteorological variables. Website users can acquire route- specific graphics of aviation impact variables (AIVs) such as icing, turbulence, and convection; meteorological observations, as well as grids of AIVs appropriate for ingest by commercial flight planning systems and FAA automation systems. Developers can acquire feedback regarding the operational utility of new products. Users have provided many suggestions for improving ADDS, such as additional products and expanded domain. In addition to making these enhancements, the ADDS development team will continually strive to enhance the performance and utility of ADDS by incorporating advances in interactive software.
Examples of feedback on general use of the Web pages include comments from pilots: "This page is the best I've seen for aviators. I wish our product at work was as intuitive and useful." "ADDS is superb. I'm concerned we won't need humans in our weather departments anymore"; from an airline dispatcher: "You took a good site and made it fantastic!"
When designing ADDS, our goals were to make it easy to access and easy to use, to provide information in a variety of formats, and to enable users with a reasonably wide range of computing technology and savvy to benefit from ADDS. Since it is available via the Internet, it is easy to access. The design of ADDS conforms to guidelines listed in the Information Architecture for the World Wide Web by O'Reilly and Associates, including suggestions to: organize by topics and maintain consistent page layouts, optimize for 800 x 600 resolution so that most users can view the pages efficiently, require no more than two "clicks" to access desired information, use only a few colors to ensure that they are supported by all browsers and keep the colors consistent across all pages, provide users easy feedback capability, make graphics as small as possible in order to maximize performance without compromising content, and, finally, to prominantly display the "Weather Checklist."
ADDS is being developed by FSL, NCAR, and the NWS Aviation Weather Center, with funds provided by the FAA Aviation Weather Research Program.