RAP Achievements ... Inflight Icing Snowfall and Freezing Convective Weather Atmospheric Turbulence Numerical Weather Prediction Water Resources Land Surface Atmospheric Remote Sensing Precipitation Enhancement Ceiling and Visibility Oceanic Weather Surface Transportation Statistics Applications NCAR Iinitiatives

B. Snowfall and Freezing Precipitation

[Background] [Freezing drizzle]
[WSDDM System at DIA]
[Deicing fluid testing] [Short-term forecasting]

1. Background

RAP continues to be involved in the research related to winter storms and in particular, their influence on freezing precipitation. In past years RAP focused on the impacts of winter storms and frozen precipitation within the aviation community. During 2004, RAP expanded this scope to include the needs and problems of other forms of transportation. RAP is involved in the development of new types of frozen precipitation detection, measurement instrumentation, and improved numerical forecasting, as well as the development of new data analysis, product synthesis, and decision support tools. RAP also focused on transferring many of these technologies into the commercial sector. In 2004 Yankee Environmental Systems began manufacturing the RAP/ Desert Research Institute's Hot-Plate precipitation gauge, and final testing and validation of this new instrument is currently underway at NCAR's Marshall field site. Additionally, a commercial WSDDM aircraft de-icing and Airport Operations system is being operated and maintained by RAP at Denver International Airport (DIA) and is being used as a testbed for new products and for gathering user feedback. Technology from this WSDDM system is now being transfered to a newly formed company, WSDDM Technologies Inc., which will build, operate, market and maintain WSDDM systems at aviation facilities throughout the United States.

Highlights of the winter weather research group at RAP include:

• documenting freezing drizzle damage to 737 jet engines during two Halloween winter storms;

• testing and improving the Hot-Plate precipitation gauge;
  

• operating an indoor snowfall laboratory where de-icing fluid testing can be done in a controlled environment;
  
  
• developing an automated detection and reporting mechanism for freezing drizzle using instruments already fielded as part of the Automated Surface Observing System (ASOS);
  
  
• developing a 2-D NEXRAD reflectivity to frozen precipitation rate product;
  
  
• developing a 2-D precipitation type product;
  
  
• developing a visibility nowcast based on a snowfall nowcast; and
  
  
• improving numerical modeling data assimilation techniques for snow prediction using radar data.
  
  

2. Freezing drizzle damage to 737 engines at DIA

On the evening of 31 Oct 2002, twelve United Airlines 737 aircraft incurred jet engine damage after experiencing a winter storm in Denver. The damage was primarily bent fan blade tips, and was consistent with ice being ingested into the engines at normal flight engine speeds. Total damage reported by United Airlines was over $2 million, with one engine requiring replacement. The damage was noted after the aircraft landed at their destination airport. The aircraft that incurred damage departed Denver between 0800-0300 UTC 1 Nov 2002. The weather observer at Denver reported mist and light snow during this time with temperatures ~ -8°C. While in-bound to Denver, these aircraft encountered light to moderate icing aloft as given by pilot reports. After landing, the aircraft were deiced at the gate to remove ice build up that occurred via in-flight ice accretion. This deicing included the removal of any ice from the jet engine fan blades. Once deiced and loaded, the aircraft taxied to deicing pad B (located just west of Concourse B at DIA), and were further examined for ice accumulation due to the reported snow conditions. During this period ground personnel reported freezing drizzle despite the METAR report of mist and light snow. Because Denver is a category A airport, there was an observer present who augmented the METAR to report snow and mist.

One year later, on 31 Oct. 2003, Denver again experience freezing drizzle during which additional jet engines were damaged. Another similar case of engine damage occurred at Oslo Gardemoen airport in Norway with Braathens Airlines in February 2003. Roy Rasmussen and Chuck Wade examined the weather conditions associated with these cases, and came to the conclusion that the actual weather condition during the time period when the engine damage occurred, in both cases, was most likely heavy freezing drizzle rather than light snow and mist as reported in the 31 Oct 2002 Denver case, or light freezing drizzle as reported during the 31 Oct 2003 Denver and February 2002 Oslo cases. Had the flight crew been aware of the heavy freezing drizzle conditions, standard engine run-up procedures would have been implemented to shed any potential ice accumulation. Since the official weather observations did not indicate heavy freezing drizzle, this procedure was not implemented, and the ice that accumulated from taxiing from the gate to to takeoff was likely shed during the takeoff rotation, causing the damage reported to the aircraft.

R. Rasmussen conducted a training class on this hazard to United 737 pilots on 15 April 2004, and a journal paper is in preparation. [Top]

3. WSDDM system at Denver International Airport

In 2004, NCAR/RAP continued to operate a commercial prototype of the WSDDM system, first installed in April 2003, at DIA. This system is currently in full operational use on a continuous basis at the DIA Operations Command centers and in their Ramp Control Tower. A prototype WSDDM Check-time system was also installed and operated at United Airlines aircraft de-icing facilities and was used as an operational test bed and basis for gathering user feedback. Enhancements to the WSDDM system developed during the year have been integrated into the standard WSDDM system and are scheduled for transfer to a commercial WSDDM vendor in fiscal year 2005. These enhancements include an automated mechanism for the detection and reporting of freezing drizzle, and gridded precipitation rate type products.

Figure B-1. Example display of current and forecast weather to various operational facilities at DIA.

[Top]

3.1 Development of precipitation type and rate grids.

Interpreting NEXRAD radar data presents several challenges to non-meterologists. While the location and intensity of storms are seemingly easy to discern, establishing a storm's intensity is complicated by the fact that the strength of the returned signal is affected by the size of precipitation particles (i.e., snow, rain, drizzle) in the air, rather than their mass. During winter storms the intensity of the signal from the radar is an especially poor indicator of the water content in falling snow. This is problematic for aircraft de-icing and airport decision makers, as the water content of the snow directly affects how long anti-icing treatments remain effective on both aircraft, runways and taxiways. Providing estimates of the snow storm's intensity in terms of water content over the a wide area would assist decision makers in their planning efforts, by helping them to deploy the correct amount of de-icing fluid on aircraft or runways to mitigate storm effects. The WSDDM system is well-suited to provide a solution to this problem by providing a mechanism for calculating the liquid equivalent precipitation rate from the radar's signal. WSDDM systems have high-resolution snow gauges on the airport grounds and in the area around the airport. Precipitation data are gathered from these gauges and compared with the radar data above the gauge to derive best fit Z/S coefficients for each gauge location. NCAR developed software to apply the Z/S function to each data point in a radar image. The algorithm NCAR developed uses the coefficients from the nearest gauge to compute a liquid equivalent precipitation rate at each radar gate (grid cell). If a radar cell is located more than a specified distance away from a snow gauge, climatological averages for Z/S coefficients are used for that grid cell. The result is a radar based image, where the color coding indicates a quantitative precipitation rate, measured in mm/hr. See Figure B-2.

Figure B-2. An example image of gridded precipitation rate from the DIA WSDDM system.

NCAR used similar nearest-neighbor methods to derive a gridded product that depicts the precipitation type over a larger regional area. In this case, METAR data are used to set each radar image cell color. The latest METAR data are used in conjunction with the latest NEXRAD data to produce a 2-D precipitation type product. See Figure B-3.

[Top]

Figure B-3. This image depicts both precipitation type and intensity from a mixed, rain and snow event which occurred in the Denver area on 21 April 2004. In this case, there was a band of cold air that had pooled over central Colorado, producing snow with warmer air to the north and south which produced rain. To the west of the airport was a METAR station indicating the presence of freezing rain or drizzle. A METAR station to the east indicated unknown precipitation (UP).

4. De-icing fluid testing with NCAR snow machine

The NCAR Snow Machine was developed to perform de-icing fluid endurance time testing in a controlled environment year-round instead of the seasonal outdoor testing that is more commonly conducted. The NCAR Snow Machine was tested in 2004 as part of a cooperative project between NCAR, the Anti-icing Material International Laboratory (AMIL), and APS Aviation Inc. (APS). Two NCAR Snow Machines, (one used at NCAR and one used at APS,) along with one AMIL Snow Machine were used in a series of tests to determine if the machines could be used to establish endurance times for anti/de-icing fluids in the aviation industry. Outdoor fluid tests were chosen by APS from previous testing. The testing procedure required that the test plate be maintained at a constant temperature throughout the individual experiments.

A comparison of the fluid endurance times from the NCAR Snow Machine and the outdoor testing is shown in Figure B-4. The figure shows that almost the entire set of fluid endurance times from the NCAR snow machine fell within ±25% of the outdoor endurance times. The value of ±25% is considered reasonable based on the results considering typical errors inherent in the outdoor and indoor experiments. Figure B-4 also shows there is more scatter in experiments with longer endurance times and lower snowfall rates. This is believed to be caused by larger variations in snowfall rate and temperature over longer outdoor tests.

A comparison of the results between the NCAR Snow Machine operated at NCAR, the NCAR Snow Machine operated by APS, and the Snow Machine developed and run by AMIL for the tests show that all the machines were in good agreement with one another.

Overall, results from the testing showed good correlation between the two Snow Machines and with the outdoor data (over 80 points tested). This improved result is attributed primarily to the use of a constant plate temperature slightly below the outdoor ambient temperature. By using a constant plate temperature, many of the temperature dependent characteristics of the various de-icing fluids (viscosity, surface tension, etc.) are controlled to values close to what the fluids experienced outdoors, allowing for an improved estimate of endurance times by the machines. The results also showed that the plate temperature needs to account for snowfall rate and wind speed for the machines to show good correlation. Other factors such as horizontal snow distribution and the crystal shape seem to be of secondary importance. The current results suggest that the NCAR snow machine is adequate for reproducing outdoor snow testing of de-icing fluids.

Figure B-4. Comparison of Outdoor and NCAR Snow Machine Endurance Times for all fluids used in the Round Robin testing. The black line shows where the endurance times would be equal and the blue lines show a margin of error of 25%. [Top]

4.1 Estimating snowfall rates and accumulations using an optical sensor

During FY04 C. Wade developed an algorithm that derives snowfall rates and accumulations remotely from an optical sensor located at National Weather Service Automated Surface Observing System (ASOS) stations. The optical sensor, called LEDWI (Light-Emitting Diode Weather Identifier), is used to primarily determine of precipitation type. However, LEDWI is also capable of measuring precipitation rate. When rain is occurring, LEDWI’s high- channel data are proportional to the rain rate, and when it is snowing LEDWI’s low-channel data are proportional to the rate of snowfall. LEDWI reports new low- and high-channel values each minute, so the precipitation rates can be derived each minute. Summing these rates provides accumulation information. An example of a snowfall accumulation trace computed from the LEDWI sensor located at Boston’s Logan Airport is shown in Figure B-5. Snow depth observations from the weather observer located at the airport are shown for comparison.

Figure B-5. Snowfall accumulations derived from LEDWI for a snow event that occurred at Boston, MA on 16-18 March 2004. Observed snow depths from the station observer are plotted on the diagram for comparison.

5. Short-term forecasting of snow in the northeast corridor

Mei Xu, Andrew Crook and R. Rasmussen are working on a data ingest and numerical forecasting system suitable for 1-12 h forecasts in the terminal area during winter storms. These scientists adopted the real-time, mesoscale data assimilation and short-term forecasting system that NCAR developed for the Army test ranges (RTFDDA) as the modeling framework. The system is based on a high-resolution MM5 and an observational nudging (Newtonian relaxation) scheme, and it assimilates observations from various sources, including NEXRAD Level II data from multiple radars. During FY 2004, numerical experiments were conducted to test various radar data assimilation techniques and a wind analysis algorithm based on the volume-velocity processing (VVP) method was developed and tested. Case studies were conducted for snowstorm events that occurred in the northeastern U.S. during winter 2002-2003. A highlight of the FY04 work was a seven-week real-time operational demonstration of the RTFDDA system with radar data assimilation during the winter of 2003-2004. Statistical verifications, as well as analysis of individual winter storms, were also performed.

The system uses the output from the 40-km Eta model or the 110-km AVN model as initial and boundary conditions. The data assimilation engine of the RTFDDA system is based on observational nudging. Each observation is ingested into the model at the observed time and location, with proper space and time weights. The system runs in a three-hour cycling mode and is cold started once a week. The RTFDDA system is suitable for both case studies and real-time operations for short-term snowstorm forecasting. The numerical core can be easily reconfigured and the data ingest modules highly independent of each other, providing the flexibility required for case studies and system upgrades. An average configuration can run on a 16-node Linux cluster and it usually takes about one hour to receive, decode and assimilate the observations.

In recent years, real-time Level II radar observations have become available via the Collaborative Radar Acquisition Field Test (CRAFT) network. The Corridor Integrated Weather System (CIWS) northeast domain has more than 20 NEXRAD radars, and covers an area of approximately 1500 km x 850 km. 3-D mosaic reflectivity for the region, as well as radial velocity from the individual radars, were accessible to RAP in real-time (see Figure B-6). Doppler radars provide wind observations at very high spatial resolution, however, with single Doppler coverage, only winds in radar radial directions are observed. To obtain estimates of the full atmospheric wind, the VVP method is one of the analysis techniques used to derive full wind vectors at a reduced spatial resolution. The technique divides the radar sweeps into many small volumes for which the wind parameters are estimated by multivariate regression analysis. Provided that the 3-D wind field is spatially linear within each volume, wind parameters can be obtained by solving a set of linear equations using radial velocity observations within the volume. An example of wind analysis and VVP processing in the CIWS northeast domain at 12 December 2002 is shown in Figures B-7 and B-8.

Figure B-6. NexRad coverage of continental U.S. at 3 km ASL (From Maddox, R., J. Zhang, J. Gourley, K. Howard, 2002). The orange box indicates the CIWS northeastern domain. [Top]

Figure B-7. The low-level wind fields at 12 December 2002. Plotted are RTFDDA analysis (upper panel) and VVP analysis using the CIWS radial velocity data (lower panel).

Figure B-8. The low-level wind fields at 2002121200. Plotted are RTFDDA analysis (upper panel) and VVP analysis using the CIWS radial velocity data (lower panel). [Top]

The case studies show that the RTFDDA system was skillful in predicting a storm's occurrence, though not very accurate at predicting the individual precipitation bands. The 7-week real-time run in FY04 demonstrated the usefulness of the radar data nudging scheme in blending the cloud mixing ration observations into the model analyses, and in improving the skills for 0-3 h precipitation forecasts. Even though the present RTFDDA system assimilates wind observations from surface and upper air stations, profilers, etc., the data are insufficient for accurately specifying the wind fields at scales relevant to individual snowbands. However, radar data are shown to have a positive impact on temperature forecasts in the 3-9 h range, suggesting a possibility of improving the RTFDDA forecasts by the radar data assimilation in this range.

Figure B-9. Precipitation rates at BWI (a, b) and LGA (c, d) during the March 16, 2004 snowstorm event. Pink: radar observation. Green: RTFDDA analysis without radar data. Blue: RTFDDA analysis with radar data. Black: 1-3 h forecast from RTFDDA with radar data. Grey: 4-6 h forecast from RTFDDA with radar data.

[Top]

RAP Achievements ... Inflight Icing Snowfall and Freezing Convective Weather Atmospheric Turbulence Numerical Weather Prediction Water Resources Land Surface Atmospheric Remote Sensing Precipitation Enhancement Ceiling and Visibility Oceanic Weather Surface Transportation Statistics Applications NCAR Iinitiatives