September 2016 v4 Development Summary

The following message is the September 2016 v4 Development summary of Taiganet's WS4000 Simulator project.

In September, v4 development covered three main areas:

Generalizing the radar map atlasing code so it could be used on segments that utilize the regional map.

Getting the Regional Observations segments up and running with auto city placement on the map based on the city's longitude and latitude coordinates.

Starting work on the architecture that will map the NWS's National Digital Forecast Database output data to the 4000's forecast categories set.

The lion's share of dev time in September went towards the third area, implementing the NDFD architecture. This is something that has needed to be done for many years now, in order to replace a very aging architecture that attempts to parse zone forecast text products, and then map its interpretation of the wording to the 4000's forecast categories set. This architecture has been largely unchanged since version 2 was released back in May of 2005.

The v2/v3 zone forecast parser has by its nature never been as reliable as desired and also was not able to map to some 4000 forecast categories. A better approach using a digital data format has always been needed, and for many years alternatives like Area Forecast Matrices and Point Forecast Matrices were investigated, but ultimately had too many drawbacks, the main being that they were for fixed locations and not able to cover smaller cities and towns.

The code that maps a set of NDFD precipitation elements to a 4000 forecast category is mostly complete now, and was quite an involved process, as the input data can be in a "set of sets" format, with each set element representing a 3 hour time period of a given day. This has to be mapped to a single element in the 4000's forecast categories that best represents the precipitation summary for the 12 hour period that the whole superset encompasses.

The availability from the NDFD output of discrete digital data for things like high temperatures, low temperatures, precipitation type, cloud coverage, and the Storm Prediction Center's severe thunderstorm probability for a given location has finally made it possible to include more of the 4000's forecast categories into the simulator, and v4 will be able to display things that never showed up in v2 and v3, such as:

"AM Snow"
"PM Rain"
"Wintry mix to Rain"
"Variably Cloudy"
"Strong T'Storms"

on the extended forecast now, and finally eliminate bugs that would happen on occasion from missing temperature information or wrongly interpreting things like elevation/wind speeds from the zone forecasts as high or low temperatures. In addition, it completely sidesteps the issue of having to map the names of days of the weeks into a year/month/day representation which could get tripped up by text like "THROUGH THE REST OF THIS AFTERNOON" and some various permutations of holiday names.

Work in October will continue to focus on completing the NDFD architecture, running tests on it, and then utilizing it to begin implementing some of the forecast segments on the 4000 (Regional Forecast, Extended Forecast, and Travel Cities Forecast).


Thank you again for being a supporter of this project.