OBJECTIVE WEATHER PREDICTION PROJECT (P. Dallavalle)
Short-Range Weather Forecasting (P. Dallavalle):
We have completed an evaluation of the Satellite-derived Cloud Cover Product (SCP) used to complement the ASOS sky cover reports above 12,000 feet. This study was described more completely in last quarter's Progress Report. Essentially, we used either the GOES-8 imager-based or sounder-based SCP to complement the ASOS observations of sky cover valid below 12,000 feet. We found that the imager-complemented observation tended to overestimate the frequency of clear conditions while correctly estimating the frequency of overcast skies. On the other hand, the sounder-complemented observation tended to overestimate the frequency of overcast conditions while correctly estimating the frequency of clear skies. However, both the sounder-based and imager-based technology added information to the ASOS observation. As a caveat, the analysis was based solely on observations recorded during one spring month. Results of this evaluation were presented by K. Hughes at the 15th Conference on Weather Analysis and Forecasting held in Norfolk, Virginia, in August. We also provided an analysis to the Office of Meteorology.
At the Fifth National Heavy Precipitation Workshop held in State College, Pennsylvania, in September, M. Antolik presented a talk entitled: "What are our needs for synoptic-scale interpretive statistical QPF guidance in a modernized National Weather Service? How should TDL help meet these?" This discussion highlighted some of our plans for future development of quantitative precipitation guidance. We are using the Workshop discussions to prepare a proposal describing future development of statistical QPF guidance.
Medium-Range Weather Forecasting (M. Erickson): Efforts to move to the CRAY all the operational software processing required for the Medium-Range forecast products have nearly been completed. MOS forecasts based on either the Aviation (AVN) model or the Medium-Range Forecast (MRF) model are now being produced operationally on the CRAY. Guidance which had previously been provided to NCEP's Hydrometeorological Prediction Center (HPC) via computer printouts is now available through NCEP's N-AWIPS display system. Also, as a consequence of the conversion effort, we are providing a number of new products to HPC, including gridded forecasts of max/min temperatures, PoP, and departures from normal of the max/min temperature and PoP. A set of tables displaying the medium-range guidance for Alaska is also available. At the 15th Conference on Weather Analysis and Forecasting, J. Settelmaier presented a paper entitled "Evaluating medium-range weather forecasts: An analysis of errors." Both the MOS guidance and the HPC forecasts of max/min temperature and PoP were evaluated in this paper.
In our effort to produce statistically-based products from the output of the MRF ensembles, we are now routinely generating MOS forecasts from 11 ensemble runs at 0000 UTC. These forecasts are not being disseminated, but are being collected for evaluation. Efforts to establish an MRF and ensemble archive on the CRAY are continuing. At the 15th Conference on Weather Analysis and Forecasting, M. Erickson presented a paper entitled "Medium-range prediction of PoP and max/min in the era of ensemble model output." In this paper, TDL's plans for future statistical development from the medium-range ensembles were discussed.
National Verification Processing (V. Dagostaro): In support of the AFOS-Era Verification (AEV) program, max/min temperature and PoP verification summaries for April - June 1996 have been distributed to the regional Scientific Services Divisions (SSD's). The summaries contain comparative verification results for local and MOS forecasts for approximately 95 stations in the contiguous U.S. No data were available for Alaska. Two sets of verification results were provided, namely, the official results comparing local forecasts to NGM MOS and a separate verification comparing the local forecasts, NGM MOS, and AVN MOS.
We've also processed and distributed to the SSD's the 1996 warm season verification results for the aviation weather elements (i.e., ceiling height, visibility, and wind). For those elements, local forecasts have not been available since July 1 when a change to the TAF format for the terminal forecasts occurred. Thus, for the aviation weather elements, the 1996 warm season consists of data for April through June 1996. For ceiling height and visibility, comparative verification results for local forecasts and persistence observations were produced for approximately 95 stations in the contiguous U.S. Summaries for wind speed and direction contain local forecast verification results for the same stations. For the aviation weather elements, local and MOS forecasts are not directly comparable because the valid times of the forecasts do not always match. Therefore, we produced separate verification summaries comparing the NGM MOS guidance and persistence observations for ceiling height and visibility and verification summaries for just the MOS wind speed and direction. No data were available for Alaska.
Severe Weather Prediction (R. Reap): Implementation of the operational NGM-based three-dimensional trajectory model on the CRAY mainframe was scheduled by NCEP for late 1996. In support of this implementation, we are currently working on two trajectory model graphics programs that are designed to produce parcel trajectories and automated convective outlooks of thunderstorms and severe local storms for display on AFOS. Technical Procedures Bulletin No. 437 entitled "Probability forecasts of aircraft icing for the contiguous U.S." by R. Reap was written and is currently undergoing review. The bulletin contains details of the development procedure and verification statistics for the new NGM-based MOS system to predict regions favorable to aircraft icing. The probability forecasts will be generated on the CRAY mainframe in the operational jobstream following the NGM-based trajectory model. The forecasts, as described in last quarter's progress report, will be used as guidance by the NCEP Aviation Weather Center's (AWC's) forecasters in issuing operational icing notices for use by the aviation community. The forecasts will be displayed in graphical form on AWC's N-AWIPS system.
LOCAL TECHNIQUES DEVELOPMENT PROJECT (W. Seguin)
0-3 Hour QPF and Severe Weather (D. Kitzmiller):
We also began collecting Stage III precipitation analyses from the Pleasant Hill, Missouri, RFC. These cover much of the northern Plains states, greatly extending the coverage of our high-resolution precipitation observation archive. All Stage III data for May-July 1996 from both the Tulsa and Pleasant Hill RFC's have been merged into a set of common files for future development work.
As another component of the 0-3 h QPF system under development, we continued developing code for preparing 0-3 h forecasts of cloud-top temperature based on GOES 8 IR data. Forecasts produced in this manner are being collated with corresponding Stage III observations over the Plains states.
We began testing code for making 0-3 h extrapolation forecasts of radar reflectivity using the 10-km national RCM mosaic prepared by the Aviation Weather Center (AWC) in Kansas City. We are doing this to obtain a better understanding of the availability and characteristics of these data as they represent potential QPF predictors.
We have created a new, experimental 0-1 h rainfall forecast product, the probability of > 1 inch or more of rain within a 28-km square region in the projected path of a storm cell. This supplements the 4-km gridded rainfall forecasts which specify probabilities for rainfall amounts only up to 0.75". The logic for the 1-inch probability product is similar to that for severe weather probability; the probability is for at least one 4-km square area within the 28-km square region to receive > 1 inch or more though the exact location is not specified. The probability value is based on the rainfall extrapolation forecast and on VIL. This product indicates individual storms which have especially high potential for localized heavy rains, based on the water mass within the storm and its velocity. We plan to incorporate the 1-inch rainfall probability in the output of the thunderstorm identification and tracking algorithm.
We developed and tested a newly-implemented capability for automatically sending real-time radar graphic files from the AWIPS NMT computers to the AWIPS GDP systems and the machines located in TDL. This enables us to generate radar forecast products on systems other than AWIPS computers on the seventh floor of SSMC2.
A briefing was provided for the Hydrologic Information Working Group (HIWG) meeting on the 0-1 h rainfall forecast algorithm in July. A talk and a poster presentation on the 0-1 h rainfall forecasting algorithm was presented at the 5th NWS Heavy Precipitation Workshop in State College, Pennsylvania. We provided assistance to the Office of Meteorology and an outside agency in analyzing and interpreting radar Archive III and Archive II datasets containing observations of debris from the crash of TWA flight 800. Thunderstorm Identification and Forecasting (S. Smith): We continued to carry out a real-time evaluation of the AWIPS Thunderstorm product on the AWIPS computers. A post-convective season evaluation of the Phase-I Thunderstorm product was begun.
Papers entitled "An Overview of the AWIPS Thunderstorm Product" and Forecasting and "An Automated Technique for Determining Satellite Cloud-Top Temperatures for Thunderstorms" were presented at the 15th AMS Conference on Weather Analysis.
A paper entitled "Comments on 'An Interesting Mesoscale Storm-Environment Interaction Observed Just Prior to Changes in Severe Storm Behavior'" was submitted for publication.
We updated the lightning decoder software to handle a change in the encoding instituted by the contractor. We provided the modernization office with input on the comparison of the decoded ASCII lightning data stream from AWC in Kansas City with the encoded lightning data stream from the contractor. We also provided NSSL with a McIDAS-X program written by TDL. The software decomposes a GOES infrared image into discrete local minima in temperature.
Local AWIPS MOS Program (LAMP) (J. Ghirardelli): We are working on a new cloud layer algorithm as part of LAMP to forecast continuous cloud heights and categorical cloud amounts for up to three layers of clouds. The LAMP forecasts are compared to thresholds to determine whether the LAMP forecasted cloud height should be accepted, or persistence should be forecast. Studies were done on the 1100 UTC cool season ceiling forecasts to assess the stability of the algorithm and to evaluate the behavior of the thresholding technique. The percentage of time that the algorithm yielded a forecast of persistence instead of the LAMP forecast was determined, and mean absolute errors by layer were calculated for persistence and for the LAMP forecasts before and after post-processing. The number of times the modified LAMP forecast was superior to persistence was also calculated. Since LAMP's forecast of sky cover plays a key role in the algorithm, the sky forecasts were also investigated to determine the relationship between the forecast of sky cover and the actual ceiling observations. These studies were done on both the dependent sample of cool seasons from 1981 to 1990 and the independent sample of the 1993 to 1994 cool season.
A paper entitled "Cloud Layer Forecasting Within the Local AWIPS MOS Program (LAMP)" was submitted to the AMS for the Seventh Conference on Aviation, Range, and Aerospace Meteorology to be held in February 1997.
A technique for producing LAMP forecasts at 0400 UTC from the 0500 UTC regression equations was developed and implemented during this quarter. Necessary changes were made to the 0500 UTC equations packing, display code, scripts, and crons to run and display the 0400 UTC forecasts daily on the GDP equipment. In addition, these forecasts are now being sent to the four WFO sites which use the ICWF and will be used as guidance for the 0600 UTC issuance of the terminal aviation forecast (TAF).
Discussions began with WFO Charleston concerning how best to use the LAMP guidance, details about utilizing the display software, and how to verify the forecasts.
Heavy Precipitation Forecasting (J. Charba): The LAMP operational QPF system is now producing forecasts for all eight start times. Software was added to accomplish several post-processing functions including normalizing the probabilities such that their ranges are forced to be 0 - 100%, smoothing the probability fields to remove some fine scale noise, and adding the best category and expected value as additional products. Development of software that will archive the real time LAMP QPF products is near completion. Both the original and experiment-ally smoothed forms of the forecasts will be archived by this software.
The LAMP QPF products were described in a talk at the Fifth Heavy Precipitation Workshop held in State College. We were approached by representatives from several field offices who expressed an interest in receiving the realtime LAMP QPF products.
MARINE ENVIRONMENTAL PREDICTION PROJECT (W. Shaffer)
Hurricane Storm Surge Forecasting (W. Shaffer):
Support was given to NHC during recent hurricanes Bertha and Fran, both of which made landfall as hurricanes in the Wilmington, North Carolina, area. Surge estimates were made by NHC forecasters using SLOSH simulation output from basins along the southeastern coastline. Evacuation plans, based on the simulation studies, formed the basis for coastal evacuations. Following both hurricanes, the OFCM's Working Group for Post-Storm Data Acquisition coordinated the federal collection of high water marks.
Extratropical Storm Surge Forecasting (W. Shaffer): Although the extratropical (e-t) storm surge model was designed for large-scale extratropical events, it apparently did quite well in capturing hurricane Fran's surges in the Chesapeake Bay. Tide gage measurements at Washington, Baltimore, and the Chesapeake Bay Bridge Tunnel were compared to e-t model forecasts from several AVN cycle runs. Beginning with the 0000 UTC run on September 6, the AVN model forecast the storm's location, size, and timing well. (Prior to that time, the AVN misplaced the storm's center and did not correctly depict its magnitude.) The e-t model seems to react properly to such large-scale flow well away from a decaying tropical system. Had Fran taken a track further east, we expect that the e-t model would not have properly handled the surge. The e-t model should continue to be used only for extratropical storm systems; SLOSH for tropical cyclone events.
Coastal Wave Forecasting (C. S. Wu): A review of the current ocean wave models has led us to two approaches for treating hurricane-generated waves at the coastline. The first approach is to adapt one of the most up-to-date spectral wave models. The U.S. Army Corps of Engineers 3GWAM model; the Office of Naval Research's SWAN model; and the NCEP's regional wave model are all candidate models for forecasting hurricane waves on the shelf.
The second approach is to use a parametric wave model, such as the formulation suggested by Dr. Jelesnianski. Such a parametric wave model could easily be coupled to our SLOSH model. Both approaches will be followed, with initial tests planned on data hurricanes Felix and Opal (1995) and Fran (1996). Wave data for Felix and Fran were collected at NDBC buoys and at the Corps of Engineers' pier at Duck, North Carolina. There was an extensive set of high water marks measured for hurricane Opal in the Pensacola area.
LOCAL TECHNIQUES DEVELOPMENT PROJECT (D. Ruth)
Field Applications Assistance (R. Beasley):
The development system was updated as well in response to the METAR/TAF conversion. Database files were updated as well as to the ASYNC subsystem.
The program CHKSFCOBS was completed and distributed to the field in late July. A version of ASOSCOMP was also completed along with its documentation. This program will be distributed to field sites in early October.
WFO Application Development (D. Ruth): Design, Development, and Testing teams continue to develop and test hydrometeorological (HM) applications for the early releases of AWIPS. This work is being carried out in part by TDL and its application support contractor, GSC, and in part by the AWIPS Prime Contractor, PRC. This quarter, significant effort was placed on testing Build 1 applications. HM Build 1 software includes Grid Processing, Contour Gridded Data, NEXRAD Image and Graphic Overlay Displays, Grid Parameter Derivation, and a stand-alone portion of the WFO Hydrologic Forecast System (WHFS). TDL and PRC are working closely together on responding to Deficiency Reports (DRs) identified by PRC's System Integration and Testing, HM Informal/Formal application testing, and at the recent Operation Test and Evaluation (OT&E) at Kansas City and Salt Lake City. Draft Test Reports (TRs) have been completed for the Grid Processing, NEXRAD Image and Graphic, and Contouring applications. The TRs will be finalized and delivered to PRC in October.
AWIPS Application Requirements and Design Documents (RDDs) and the associated manpages for "public" and "private" APIs for Grid Processing, Contour Gridded Data, NEXRAD Image and Graphic Overlay Displays, and Grid Parameter Derivation were delivered to PRC in August. Also, TDL prepared and delivered in August a Version Description Document (VDD) which describes the HM Ver- sion 1.0 application software and the associated documentation. TDL, OH, and GSC personnel continue to work on implementing configuration management software (PCMS) purchased for the headquarters systems to support HM applications development for AWIPS.
Work continued on more than two dozen applications originally scheduled for implementation in Build 3 and later. Many of these applications are now near completion. Most recently, we are assisting in planning for the direct insertion of WFO Advanced Technology into AWIPS.
Product Generation (M. Peroutka): Build 9 of the Interactive Computer Worded Forecast (ICWF) was delivered to the NWS Forecast Offices in Charleston, West Virginia; Norman, Oklahoma; Peachtree City, Georgia; and Boise, Idaho. Development was completed for the program which uses forecasts from station-based digital forecast matrices (DFMs) to generate gridded forecasts. Several maps for this and other programs were developed for various ICWF sites to support this new functionality. The temperature phrase algorithms were enhanced to support a wider variety of range phrases.
TDL continued support of the Olympic Games in July and the Paraolympics through August. Work continued to provide the AWIPS Forecast Preparation System (AFPS) with a source of MOS initialization. A full suite of ICWF applications was installed at the University of Virginia to support the development of techniques to prepare probabilistic QPFs.
Several enhancements were made to the statement generator for the ICWF Watch/Warning/Advisory (WWA) system. Sites can now issue WWA's for severe weather (short-fused) events. Each site can specify a set of "canned" basis statements which can streamline the WWA preparation process. The WWA system was also enhanced to support certain non-WWA products such as Nowcasts.
An enhanced format for the Coded Cities Forecast (CCF) product was developed in concert with the Norman and Charleston Forecast Offices. This format will provide coded forecast information for the later forecast periods in addition to those provided now.
Five manuscripts were prepared for the American Meteorological Society's 77th Annual Meeting in Long Beach, California. Four papers will be presented at the Thirteenth International Conference on Interactive and Information Processing Systems (IIPS) and one at the Seventh Conference on Aviation, Range, and Aerospace Meteorology.
Interactive Techniques Development (D. Ruth): An early version of the ICWF aviation forecast editor was provided in ICWF Build 9. Menu's are now being adjusted in response to forecaster feedback. A Watch/Warning/Advisory interface capable of supporting tornado, severe thunderstorm, and flash flood products was also released in this build. This interface now supports the issuance of products for both short-fused and long-fused events.
The capability to interpolate zone forecasts in space was added to the ICWF matrix editor. Forecasters can choose to interpolate forecast elements by one of three methods: by value, by delta from original guidance, and by nearest neighbor. This technique is expected to be especially useful in the extended forecast periods.
The ICWF slider bar interface was enhanced to allow forecasters to adjust thresholds at any forecast projection. Adjusted threshold values are then linearly interpolated and applied to intermediate hours. An interactive capability to reposition forecasts in time and space has been developed. Slider bars allow the forecaster to "speed up" or "slow down" model guidance and to shift guidance north, south, east, or west. These features will be part of ICWF Build 10.
Work was completed on an interface which allows local offices to create and modify station-to-grid maps with selected weights and offsets. Station-to-grid maps are used to initialize gridfields from sources of station guidance and to map forecaster edits to station forecasts back to the original grid. This, as well as other interfaces currently being developed, will facilitate ICWF set-up at new forecast offices in AWIPS.