The primary objective of the Techniques Development Program is to develop analysis and forecast techniques which, when implemented, will help improve forecast accuracy and service to the wide range of users of NWS products. These techniques are implemented on NOAA's computer system, when appropriate, and guidance products disseminated via AFOS, facsimile, or other NWS distribution systems. Techniques are produced for basic weather elements used in public and aviation forecasts, such as temperature and visibility. Also, special emphasis is given to marine-related forecasts and to those forecasts especially associated with mesoscale processes. For many synoptic-scale forecasts, the output of operational numerical models is used to produce forecasts of weather elements of interest to users. When dealing with forecasts of shorter time and space scales, more use is made of data sets rich in information on those scales; for example, hourly and automated surface reports and radar, satellite, and profiler data. For marine-related forecasts, numerical and statistical models relate elements of interest, such as storm surge, to atmospheric analyses and forecasts.

Much effort was necessary over the past 18 months to convert software and data to operate on the CRAY. Some software will not be converted unless absolutely necessary; rather, development of a new MOS-2000 system is now progressing rapidly. Expected amount precipitation guidance and MOS forecasts from MRF ensemble runs are now being provided to NCEP. Results of the National Verification Program are being provided to the field electronically via the OSO file server. New aircraft icing forecasts for NCEP's Aviation Weather Center joined the earlier clear air turbulence forecasts in March, and GEMPAC displays of them were improved. Experimental results from the Thunderstorm Product were put on the TDL home page, and arrangements have been made for further testing at the Sterling, Virginia, Forecast Office as part of NSSL's Warning Decision Support System. The ICWF/LAMP combination is nearly ready for AWIPS implementation, culminating efforts by TDL staff and contractors GSC and PRC. IFP risk reduction is planned for four AWIPS sites in addition to the four non-AWIPS sites that now have ICWF software. Many new capabilities have been added both in the interactive and formatting portions, and the merger of ICWF and AFPS is well underway. Encouraging results were received from Penn State which will support gridded initialization of IFP. Experimental forecasts from the LAMP QPF system are available on the OSO server, and access software was provided to the field. Hurricane preparedness was strongly supported through participation in workshops and meetings, and a near-shore wave model is being developed.

OBJECTIVE WEATHER PREDICTION PROJECT (P. Dallavalle)

Short-Range Weather Forecasting Task (P. Dallavalle): As discussed in last quarter's progress report, the Office of Systems Development is funding statistical forecasting studies at the recently-created Cooperative Institute at Pennsylvania State University. Work is continuing on creating a monthly gridded climatology for the maximum/minimum (max/min) temperature, dew point, and wind speed. The climatic data will be interpolated to a 5-km grid that can be used for the Interactive Computer Worded Forecast (ICWF). Although the ICWF is currently using a 20-km grid, we've chosen the finer resolution grid as a possible approach for future work. Efforts continue on choosing appropriate interpolation methods. A non-linear regression method which relates the climatic value to local predictors such as the latitude, longitude, or elevation of a site appeared to perform adequately. In most regions, the error of the estimated max/min temperature normal for stations withheld from the developmental sample was typically 1-2 øF. Several problem regions were noted, however, particularly along the California coast in the summer and in regions of extremely complex terrain in the winter. A method of interpolation called Kriging is now being tested. We are currently providing the Penn State researchers with the complete archive of Model Output Statistics (MOS) forecasts since January 1993.

The software modules generating the operational forecast products for which TDL is responsible have been converted to run on the Cray. Similarly, developmental archives are now being generated on the Cray, including archives of the Eta, Nested Grid (NGM), Trajectory, Aviation (AVN), and Medium-Range Forecast (MRF) models. All of the model data are being stored in the new MOS-2000 format; however, both the AVN and MRF archives are interim datasets until we can increase geographical coverage, type of parameters saved, and the horizontal and temporal resolution of the data. Work on writing software to store the hourly observational data in the new format is continuing. For all of our archives, we are also writing software and procedures to convert the old archives to the new data formats.

Beginning in the fall of 1997, a risk reduction exercise will be held in the Eastern Region to demonstrate the end-to-end forecast process required to produce probabilistic estimates of quantitative precipitation and subsequent river stage information. We are contributing to this effort by performing extensive post-processing of the NGM-based MOS quantitative precipitation forecast (QPF) guidance. We've developed software to fit a Weibull probability density function to the categorical probabilities. From the Weibull distribution, we've been able to generate expected values of the precipitation amount as well as specific fractiles. Preliminary evaluation indicates that the expected values from the Weibull distribution reasonably match those obtained algebraically from the discrete categorical probability values.

We also completed an extensive verification of the current NGM MOS QPF product as well as an experimental derivation of MOS QPF equations based on a small sample of data. During the period from October 1994 through September 1996, the NGM MOS QPF guidance improved significantly over the precipitation amount obtained directly from the NGM. In addition, the skill of the MOS QPF guidance for the probability of 0.01 inches or more of precipitation was equal to, or exceeded, that of the operational NGM MOS probability of precipitation (pop..) guidance. Given these results, we will likely eliminate the separate development of QPF and Pop. equations in the future. In our small sample tests, we developed cool (October-March) and warm (April-September) season QPF equations from both the NGM and "early" Eta models for approximately 200 stations in the eastern U.S. Verifications on 1 year of independent data showed that probabilistic forecasts from the Eta MOS system outperformed those of the short-sample NGM MOS at all projections, and consistently scored within 2% of the operational NGM MOS system for the climatic skill score.

An analogous, limited experiment was conducted for the temperature guidance. Using two cool seasons of Eta model data (October 1994 through March 1996) from the 0000 UTC cycle, we developed MOS temperature and dew point forecast equations for the 6- through 27-h projections for the same set of stations used in the QPF experiment. In a test on independent data taken from December 1996 and January 1997, we found that the Eta MOS temperature forecasts were more accurate than the operational NGM MOS forecasts at the 15- through 27-h projections. In contrast, the NGM MOS temperature forecasts were more accurate for the 6- through 12-h projections. The NGM MOS dew point forecasts were more accurate at all projections; however, our choice of very simple predictors in the Eta MOS development may have somewhat handicapped the Eta dew point forecast equations.

A large number of queries from Central Region forecasters during the past winter led us to investigate the quality of the NGM and AVN-based MOS temperature guidance during Arctic outbreaks in the Great Plains. As the forecasters indicated, we found a pronounced warm bias in the MOS guidance. One of the major deficiencies of both the NGM- and AVN-based MOS guidance is the inability to predict the intensity of cold air during shallow cold-air events. In the case of the Arctic outbreaks, we concluded that the NGM and AVN models themselves failed to depict the intensity or movement of the Arctic air mass. The problem was exacerbated in the MOS system because the MOS thermal predictors indicated a relatively warm air mass while the real atmosphere was substantially colder. We do not have an easy solution to this problem. However, a map-typing predictor that indicates the presence of an Arctic air mass might help. We also suspect that an Eta-based MOS system would likely give better guidance during these events because of the increased vertical resolution in the Eta model.

Medium-Range Weather Forecasting Task (M. Erickson): Efforts to produce statistically-generated products from the output of the MRF ensembles are continuing. Preliminary graphical displays containing the MOS forecasts from the 11 NCEP MRF ensemble runs and the operational MRF-based MOS forecasts are now available to NCEP's Hydrometeorological Prediction Branch through the N-AWIPS system. Efforts to evaluate the guidance, to establish a new archive of the ensemble model data, and to determine the most effective ways to disseminate the statistical information to interested users are continuing.

The MRF-based MOS max/min temperature forecasts were recently verified against the climatic normals for the cool season (October - March) of 1995-1996, and for October 1996 through January 1997. This information will be used to target areas of improvement in the medium-range temperature forecasts.

National Verification Processing Task (V. Dagostaro): In support of the AFOS-Era Verification (AEV) program, verification results for the 1996 warm season were generated and distributed to the regional Scientific Services Divisions (SSD's). For max/min temperature, Pop., and 42-h significant wind, we compared the local and NGM-based MOS forecasts for approximately 95 stations in the conterminous U.S. (CONUS). For cloud amount, we compared the local and NGM MOS forecasts for 24 stations in the CONUS; stations whose observations were collected by ASOS are no longer included in the cloud amount verification. Note that by the end of 1996 only 9 AEV stations were not commissioned ASOS sites. For max/min temperature and Pop., MOS guidance based on the AVN run of the Global Spectral Model was included in separate verifications comparing the two MOS guidance systems (NGM and AVN) and the local forecasts. For the aviation weather elements (ceiling height, visibility, and wind), verification results were processed several months ago and were discussed in a previous progress report. No data were available for stations in Alaska.

In addition to the warm season results, max/min temperature and Pop. verifications for October-December 1996 were generated and distributed to the SSD's. As in the warm season, local, NGM MOS, and AVN MOS forecasts were verified for approximately 95 CONUS stations. Beginning with the October-December 1996 period, we've discontinued the distribution of verification scores on floppy disks and are disseminating the results electronically via an Office of Systems Operations (OSO) file server.

Prior to generating the Pop. verification results described above, we enhanced the usual quality control process by checking all ASOS precipitation amount observations of 0.01 inches reported at AEV sites. In an effort to eliminate spurious observations of measurable precipitation caused by dew, we devised an algorithm to check the validity of the precipitation report. We compared the ASOS observations transmitted by the AEV sites with observations at manual sites (where available) or ASOS observations at nearby sites. The comparative observations were taken from TDL's hourly data archive. Once suspicious AEV reports were identified, we checked the weather and dew point depression reports at those sites. Depending on the results of various data checks, we set the suspect observation of 0.01 inches either to 0.0 when we were confident that precipitation had not occurred or to missing when we were unable to make a determination.

Severe Weather Prediction Task (R. Reap): Routine operational transmission of the MOS aircraft icing forecasts was initiated in early March. The guidance is sent to NCEP's Aviation Weather Center in Kansas City via the N-AWIPS communications system. The forecasts are valid for the 2-8, 8-14, 14-20, and 20-26 h projections after both 0000 and 1200 UTC and provide an index for the risk of icing in both the high band (above 15,000 ft) and low band (below 15,000 ft) regions. The valid periods and vertical separations are identical to those used in the clear-air turbulence forecasts first transmitted to AWC during the past year. To complete the current set of N-AWIPS products available to AWC and NCEP's Storm Prediction Center (SPC), coverage thresholds for determining the risk of severe local storms for individual areas in the automated convective outlook were adjusted and tested. The individual outlook areas indicate whether conditions favor approaching, slight, moderate, or high risk. The thresholds need adjustment to accommodate the small 47-km grid blocks for which the forecasts are valid. Previous risk thresholds were based on MOS probability forecasts for grid blocks that were four times as large. Work is currently underway to adjust and test the thresholds for risk determination during the warm season.

We've also improved the quality of the N-AWIPS graphical displays that show the clear-air turbulence, icing, thunderstorm, and severe local storm guidance described above. Refinements in the generating software and optimum scaling of the forecasts has led to final forecast products that are visually appealing and that exhibit consistency from one projection to the next for individual products as well as between the various products. R. Reap presented a discussion of the icing forecasts entitled "Probability Forecasts of Aircraft Icing for the Contiguous U.S." at the American Meteorological Society's Seventh Conference on Aviation, Range and Aerospace Meteorology held in Long Beach, California, in February.

LOCAL TECHNIQUES DEVELOPMENT PROJECT (R. Reap)

0-3 Hour QPF and Severe Weather Task (D. Kitzmiller): We are testing two national-scale 0-3 h QPF algorithms, one based on radar reflectivity from the 10-km RCM mosaic and NGM output, the other based on GOES infrared observations and NGM output. The displacement velocity of the reflectivity or cloud-top temperature patterns is estimated either by lag-correlation pattern matching or is assumed to track the 700-mb wind velocity. Probability equations for the algorithms are presently based on data from the 1996 warm season. Despite this, they have produced reasonable results during the past cool season based on comparisons with WSR-88D Stage 4 radar/gauge precipitation analyses produced at NCEP. Eventually these two algorithms will be logically merged, with the final output depending upon radar data where it is available, and upon satellite temperatures over areas where radar data are missing or where precipitation is only imminent and not yet occurring.

We are currently archiving the Stage 4 precipitation analyses mentioned above for days with extensive and/or heavy rainfall. The analyses are produced hourly and consist of WSR-88D Hourly Digital Precipitation products referenced to a 4-km national grid and modified by remote-reporting raingauges where and when they are available. We also archive the original radar-only analyses and the hourly gage amounts. All of these products are available through the NIC server.

We have begun the final stages of porting the 0-1 h QPF algorithm, the radar-environment severe weather and large hail probability algorithms, and the AWIPS Thunderstorm Product for execution within NSSL's Warning Decision Support System (WDSS). The WDSS is slated for implementation at WFO Sterling, Virginia, this coming summer.

Thunderstorm Identification and Forecasting Task (S. Smith): Our post-convective season evaluation of the AWIPS Phase-I Thunderstorm product was expanded to four convective days in 1996, including an analysis of the 24 June 1996 severe squall event that affected the District of Columbia metropolitan area. The thunderstorm product's overall probability of detection was 85%, which is an improvement over the 75% POD obtained by the previous algorithm used at the Sterling office. Sensitivity studies show that the new algorithm performs best when radar and lightning inputs are used in combination rather than separately. As a result of the evaluation, program code for the thunderstorm product is being prepared for implementation within the D2D display of AWIPS Build 3.x.

Experimental real-time text output from the AWIPS Thunderstorm Product was also made available to Internet users via the TDL homepage. The output format is a text file containing the current thunderstorm threat decision for approximately 40 sites within the WFO Sterling, Virginia, radar umbrella.

A proposal to support continued development of the AWIPS Thunderstorm Product was completed and submitted to OM for the Severe Weather Program Initiative (SWPI) part of the USWRP funding initiative. The proposal was briefed to Dr. Zevin prior to submitting the final version.

Work was begun to implement the Phase-II Thunderstorm Product on the GDP system for the 1997 convective season. In this phase, GOES infrared satellite imagery is incorporated to estimate the minimum cloud-top temperature of radar-detected storms and to improve our ground clutter/AP detection algorithm.

A 2-day meeting with representatives from OSD, OM, OSF, OH, NSSL, and NCAR was held in Silver Spring, Maryland, to plan for the 1997 field test of the System for Convection Analysis and Now- casting (SCAN). This system will integrate existing software packages in order to provide automated, objective warning guidance for severe weather and flash floods. As a prototype, the AWIPS Thunderstorm Product and NCAR's Thunderstorm Auto-nowcaster will be integrated within the framework of NSSL's Warning Decision Support System (WDSS) at WFO Sterling, Virginia.

A paper entitled "Comments on 'An Interesting Mesoscale Storm-Environment Interaction Observed Just Prior to Changes in Severe Storm Behavior'" was accepted for publication in Weather and Forecasting.

Local AWIPS MOS Program (LAMP) Task (J. Ghirardelli): During this quarter we continued our development of the cloud layer thresholds, endeavoring to develop as many as possible before the removal of the HDS mainframe. After we started development of the layer separation thresholds, which is the last set of thresholds to be developed, it became apparent that there was a problem with the forecasts for the highest cloud layer. This problem was investigated and a solution found. We then re-derived the equations, scaling thresholds, forecasts and persistence thresholds for the highest cloud layer element to correct this problem. As a result, we were able to continue with the layer separation threshold development, which we completed in early March.

In anticipation of the removal of the HDS mainframe, a great deal of effort was expended to ensure that all necessary data and program code was saved on the CRAY mainframe and TDL's GDP's. We previously submitted to NCEP a listing of more than 2,000 mainframe tapes to be transferred to the CRAY system. During the past quarter, we verified that these tapes had been transferred to the CRAY, and investigated discrepancies between our requested list and those that were actually transferred. We also examined mainframe tapes created since our initial request, submitted a request to transfer additional tapes to the CRAY system, and deleted unneeded mainframe tapes. Lastly we backed up all mainframe code that had been changed or created since the last backup.

Changes to station call letter identifiers were investigated this quarter. LAMP's station list had not been updated for changed identifiers since its creation. The required changes were made to LAMP's implementation datasets.

In February, J. Ghirardelli gave a lecture and LAMP demonstration to National Weather Service interns at the NWS Training Center in Kansas City, Missouri. LAMP development and implementation were described, as well as LAMP QPF, how to use statistical guidance, and the differences between LAMP and MOS. She later presented a paper entitled "Cloud Layer Forecasting Within the Local AWIPS MOS Program (LAMP)" at the Seventh Conference on Aviation, Range & Aerospace Meteorology at the AMS annual meeting at Long Beach, California.

LAMP source code was moved to the NHDW system, recompiled, and tested for a test case against our current GDP test system. Site specific files needed for implementation of LAMP at Taunton, Massachusetts; Pleasant Hill, Missouri; Tulsa, Oklahoma; and Wichita, Kansas, were created. Work is continuing with PRC to provide and test the datasets needed for LAMP under AWIPS.

Heavy Precipitation Forecasting Task (J. Charba): A manuscript entitled "Gridded Climatic Monthly Frequencies of Precipitation Amount for 1-, 3-, and 6-h Periods over the Conterminous United States" was submitted for publication in Weather and Forecasting.

In response to the anticipated removal of the HDS 9000 mainframe at NCEP, various data sets and software were transferred to the CRAY mainframe and GDP computer workstations. Several data access and graphical display programs were converted or rewritten so that they can be used on the CRAY. For example, software is now available on the CRAY to access and display archived hourly precipitation data, archived LAMP QPF forecasts, and assorted archived predictor data used for LAMP QPF development and testing. Conversion of CRAY software that processes hourly precipitation data obtained from the National Climatic Data Center is currently underway.

Finally, software that displays the experimental real-time LAMP QPFs at local NWS offices has been distributed to all four NWS regions within the conterminous United States. All local offices should, therefore, have the capability to display the national QPF graphical products that are produced eight times daily on TDL's GDP's.

MARINE ENVIRONMENTAL PREDICTION PROJECT (W. Shaffer)

Hurricane Storm Surge Forecasting Task (W. Shaffer): We have temporarily suspended development of the new basin for the Ocean Springs, Mississippi area. Our contract employee working on this basin is on leave and is scheduled to return in October.

Dr. Shaffer aided N.C. in presenting the NWS/FEMA courses, "Introduction to Hurricane Preparedness," held at N.C.. Three courses were given: one for emergency managers along the Gulf Coast, one for the southeastern Atlantic states, and one for the northeastern Atlantic states. In addition, a separate course was offered to the Organization of American States (OAS) countries. These included the Caribbean island nations as well as Central American countries. The OAS course was presented to emergency managers and national forecasters, fostering closer cooperation between these two groups.

Extratropical Storm Surge Forecasting Task (W. Shaffer): The operational version of the extratropical model was moved completely to NOAA's CRAY computers, with model forecasts going directly from the CRAY to the AFOS gateway. We continue to archive the model's output for verification purposes.

With the help of the Taunton, Massachusetts, office, we have been examining the model's performance on a case in early January 1997. A mesoscale atmospheric feature generated high water in the Boston area. We are trying to determine whether the event was represented adequately in any of the NWS's forecast models. The extratropical model must rely on the atmospheric model and cannot correct other forecast problems.

We also found that we need to clarify to our forecasters the current mean water level vs. the 19-year average Mean Sea Level datum, as referenced by NOS. Large seasonal fluctuations exit at the gage point by examining the NOS's monthly average data. For coastal flooding levels, we need to bring up our model's output to match gage observations on the particular month. We currently do not have an easy solution applicable to real time forecasting.

Coastal Wave Forecasting Task (C.S. Wu): A one-dimensional model was developed for nearshore waves at the coastline. The initial tests will use data from hurricane Opal (1995) on the Gulf of Mexico. For wave calculations, hydrographic data for the Panama City area were extracted from the NOS's CD-ROM of coastal bathymetry. We explored several interpolation schemes for the bathymetric data in the Panama City area. This area has a bar about 200 meters offshore. Our modeling results indicate that the submerged bar is extremely important in determining the residual wave energy in the surf zone.

Observed winds for the hurricane Fran (1996) in North Carolina were analyzed and compared favorably to the SLOSH winds at the buoy locations. Further comparisons of wind waves modeled by the parametric model using SLOSH winds with buoy observations are in progress. The output of parametric wave model will be used as an input for the shallow water wave transformation.

We documented the experimental planning and support for the SandyDuck experiment to be carried out at the U.S. Army Corps of Engineers facility in Duck, North Carolina. NOAA's Sea Grant Office funded the University of Delaware to carry out the proposed work following a presentation of the proposal by Dr. Wu. This project will provide valuable field data that will lead to improved storm surge predictions. The sensor will be deployed in June.

LOCAL PRODUCTS DEVELOPMENT PROJECT (D. Ruth)

WFO Application Development Task (D. Ruth): Design, Development, and Testing teams continued to respond to Deficiency Reports (DRs) against hydrometeorological (HM) applications. AWIPS Release 1.2 distributed in January 1997 included contour setup data fixes, improved functionality of the NEXRAD image and graphic application, and improved performance of the grid processing applications. AWIPS Release 1.3 released in March included the addition of boundary layers for some derived parameters and the modifications necessary to accommodate changes to the NCEP data stream. Final user guides were completed for the grid processing; NEXRAD image; and graphic, contouring, and parameter derivation applications.

Development activities are now centered around the integration of the fog product and ICWF/LAMP into AWIPS Release 2.1. These applications are being prepared for delivery to PRC in April and field implementation in June.

Other areas of work include the preparation of an AWIPS Application Integration Framework Manual which will support local application development for Build 3 and beyond, preparation of a thunderstorm product for Release 3.1 and forecast verification applications for Build 4. TDL and GSC personnel continued to work on implementing configuration management software.

Product Generation Task (M. Peroutka): Build 11.0 of the ICWF was delivered to forecast offices in Boise, Idaho; Charleston, West Virginia; and Norman, Oklahoma. The features in this build will be the same as those fielded with AWIPS Release 2.1 for a Risk Reduction Exercise this Summer. The AWIPS sites which will use ICWF include Pleasant Hill, Missouri; Tulsa, Oklahoma; Wichita, Kansas; and Taunton, Massachusetts. This version of ICWF provides initial support for local effect phrases, Watch/Warning/Advisory enhancements to support watch decentralization, improvements to wind and temperature phrases, and better handling of fog and other obstructions to vision.

ICWF developers visited the Marine Prediction Branch at NCEP to discuss the suite of guidance products which they produce and explore ways that these products might be used to initialize the ICWF for marine product generation. The ICWF will support the Marine Risk Reduction at the Taunton forecast office later this year.

Members of the ICWF team attended the 77th Annual Meeting of the American Meteorological Society in Long Beach, California. They presented five papers to the 13th International Conference on Interactive and Information Processing Systems and provided demonstrations of ICWF and LAMP software at the NWS booth.

Much effort was focused on porting the ICWF to the environment of AWIPS Release 2.1. All software modules were recompiled using HP-UX 10.20 and Informix 7.20. Geographic data were obtained from Modernization Systems Management, and the first group of techniques were developed which generate site customization data. The next set of user documentation was developed in concert with the NWS Training Center. ICWF developers also met with staff from the Office of Systems Operations to plan training sessions for personnel at the National Control Facility.

ICWF developers met with FSL several times to coordinate the consolidation of the ICWF and AFPS projects into a single Interactive Forecast Preparation System. Discussions have initially concentrated on the creation a common data server for forecast grids and point data.

Interactive Techniques Development Task (D. Ruth): The ICWF Matrix Editor was enhanced in ICWF Build 11.0 to handle forecasts for WFO-defined local effect areas (e.g., by the lakeshore). Special matrices are displayed for forecast elements and periods where the local effect forecast differs from base conditions. This version of the Matrix Editor also provides the capability to prepare forecasts for Day 3 through Day 6 at 6-hour resolution.

The capability to employ forecast elements as weights was provided in the ICWF grid editor and model interpretation techniques. For example, a forecaster can raise thunderstorm probabilities in areas with high dewpoints or can base snow amounts on previously-entered quantitative precipitation forecasts. The capability to inclemently pull forecasts toward alternative sources of forecast guidance (e.g., MOS temperatures toward Eta) was also provided.

Several improvements were made to the TAF editor, including the ability to view LAMP guidance. New time tools were added to the Watch/Warning/Advisory interface. Each WFO may now establish default start time round-offs for each of their products.

David Ruth provided demonstrations and answered questions at the NWS Training Center's IFP managers' course in early March. In January, he traveled to Kansas City to participate in planning for this course and the IFP focal point course scheduled for June.

AFOS Field Applications Assistance Task (R. Beasley): TDL continued to address software problems associated with decoding TAFs in the new METAR format. Decoded TAFs are needed in the aviation monitoring program (MONITR) and the aviation portion of forecast verification (VERIFY).

The Coded Cities Forecast (CCF) decoder was rewritten to enable the decoding of extended forecasts and many more stations. The CCF decoder is used in VERIFY. The original decoder was limited to a maximum of 12 stations out to 3 days.

A problem decoding METAR centigrade temperatures was corrected in the surface observations decoder (SAODECII). Also, special modifications were made to process Caribbean METAR collectives.

TDL continued to support the ASOS program by providing reports of missing observations, observations with missing elements, incorrectly formatted observations, and garbled observations. This information was helpful in determining the cause of garbled observations from the Midland PACE system.