Entity

The calculation requires as many 014's returned as prior iterations outstanding. With hindsight, there were a lot more prior iterations outstanding than 014's returned at that time.

As yet, I haven't received one let alone returned it and now the queue has frozen.

Mike

Entity

Arp is an iterative process. The first number is the 9 sq. Km. patch of Africa and the second number is the 2-day period starting from 1 July 2019. Each 2-day period cannot be released until the same patch has reported its previous period.

However, because some WU's get crunched faster than others, some patches have got further ahead than other patches, especially when re-sends are needed.

Up until now, the spread has been small, only about 2 iterations being issued at the same time, but now it seems to have grown to 3.

I would have thought that they would want to keep the spread down by giving precedence to wu's from earlier iterations, but it seems not.

Because they are crunching a whole year, which happens to be a leap year, there should be 183 iterations (000 - 182).

So the calculation that I did was to divide 14 iterations (000-013) by 183 to get 7.7%.

It has been explained before in this thread, but now that there are over 800 posts in this thread it needs repeating, especially as the spread of iterations is growing. It is a very different system from, say, opn.

Mike

Last time it happened it was also at the weekend although this time it started on Thursday evening. The techs work remotely and have to spot it first and then hit/kick it in the right spot to restart it.

Mike

I don't think that is correct. If so, ARP would run like FAH2. Which it doesn't. One could run any iteration of the period at anytime. All you need is current observations for a period previous to the forecast run. It wouldn't make any sense to use the output from a forecast run as input to the next run. Think about it. The forecasts aren't exact and can be wrong. They are predictions based on data and that data may not be correct or missing. Therefore, if an error is induced by a run it would be propagated to the next run. Current weather forecasts are based off current observational data that goes through a stringent quality analysis, not forecasted output. They already have the observational data from the period (The Weather Channel data) so any iteration could be run today if so desired.

From 4000 * 2 to 6000 * 2 in the wild to 1,849 validated at halftime.
https://www.worldcommunitygrid.org/stat/viewP...&numRecordsPerPage=30

I'm on the last one, no apologies, YCMTSU.