Kangaloosh
New Member
Summary: The MetOffice charge extra for their Marinecall service, boasting “Most forecasts are computer generated. Ours are written by Met Office forecasters” and “ . .forecast that is quality controlled by Met Office forecasters. The model is run 4 times a day . .”, yet I found that the extra bit that you pay for is computer generated, can be dangerously inaccurate, can escape quality control for at least four consecutive days and gets updated just once per day.
Okay, it is an inflammatory headline, but before you do as I did and pay your hard earned money for a detailed MetOffice forecast read on to find out how they tried to tempt me out into the Dover Straits with a forecast of 'Force 1, sea state calm', when ships in the Dover Straits were reporting 36 knots of wind.
At the end of October I was taking a yacht up the English Channel from the Solent to the Thames and we were enjoying typically grotty autumn weather. I did not want to wait weeks for a high pressure system to settle the weather; I wanted to make the most of what gaps appeared. I did some web browsing and invested in a MetOffice Marinecall subscription.
On Friday October 20 I downloaded my first forecast and was delighted to see Force 1 sea state calm, with nothing above a F3 for four days. It didn't feel that calm, the rigging was humming and small waves were reaching as far as our berth tucked well into a marina. A quick check online revealed shipping in the Dover Straits reporting 15 m/s (approx 30knots), with the MetOffice's own inshore forecast for North Foreland to Selsey Bill realistically predicting SW5 increasing 6 or 7. I emailed the MetOffice on both their own (24 hour) and their Marinecall (Mon-Fri daytime only) email addresses.
On Saturday, again, I got what I had paid for (!) - Another perfect forecast. Unfortunately the real weather was still awful, and again on Sunday, and again on Monday; two more dream forecasts as we hid down below from the wind and rain lashing the deck.
By Monday I had received a reply by email acknowledging the problem, but for at least four days the MetOffice had been sending out dangerously inaccurate forecasts. There is no way of knowing how many people made decisions to sail based on those forecasts. Like me, they may have been seduced by their attractive claim: "Handmade by humans Most forecasts are computer generated. Ours are written by Met Office forecasters. Trust your human instincts."[sic], quoting http://www.metoffice.gov.uk/leisuremarine/marinecall.html directly (graphic at bottom left of page.).
Yet to be this badly wrong, the forecasts I received must have been produced automatically, and quality controlled, by a computer, no human could be so consistently wrong? However, even the small print, that notorious resting place of get-out clauses confirmed that I could not blame computers for the error: "Met Office Atmospheric Weather Model creates a deterministic forecast that is quality controlled by Met Office forecasters. The model is run 4 times a day to offer the latest in local accuracy.” according to http://www.marinecall.co.uk/services/Dover.pdf
It turns out that the MetOffice use a company called iTouch to send out the forecasts. According to Martin Kidds, Customer Feedback Manager, Met Office. "On receipt of your complaint, we investigated what was causing these inconsistencies and traced the fault to a coding issue with the data after it had left the Met Office. Our partners iTouch have worked to correct this problem as a priority and I am pleased to report that the service should now be back to normal.”
So it was a computer error - so much for the quality control. These forecasts cannot have been checked, or if they were the person responsible should be made to venture out into the Dover Straits in a small yacht in the actual conditions prevailing at the time.
But surely they cannot be this cynical - are they really charging us extra for computer generated output that isn't checked by a human forecaster? It would seem that the only human generated part is the inshore forecast - which is freely available anyway on VHF and the internet. I believe this is confirmed by an email I received from Hayley Turnball, Weather Brand Manager at iTouch: "Sometimes there is a slight variance between the inshore waters forecast and the 6-hourly data and this is because the former is forecaster written and covers a vast area, whereas the latter is computer generated and covers a specific LAT/LONG.". Yes, it seem I am paying for a Marinecall subscription and the only bit that is human-forecast is the bit I get for free anyway.
iTouch immediately offered me a full refund and were very apologetic. The MetOffice were similarly apologetic and offered me a free week's subscription. I got the free week - the forecasts were much better though still only updated once per day, I am still waiting for the refund and my last email chasing the payment was ignored, but it has only been two months, still mustn't grumble - as my partner says, 'You could be dead'.
If it were a genuine error, instantly rectified, I would have forgiven them two months ago. But I was led to believe that I was paying for quality human forecasts, when in fact their mistakes revealed that I was getting computer generated data, which I get for free already from the NOAA (I believe http://www.weatheronline.co.uk uses NOAA data) and Deutscher Wetterdienst (http://www.dwd.de/de/WundK/W_aktuell/Seewetter/NOsee/Seewetter72.htm#EnglE.KAN).
I am sure the forecasts produced deep inside the bowels of the MetOffice are checked by humans. I naively assumed they meant that the forecasts I was _buying_ were both forecast and quality controlled by humans.
Okay, it is an inflammatory headline, but before you do as I did and pay your hard earned money for a detailed MetOffice forecast read on to find out how they tried to tempt me out into the Dover Straits with a forecast of 'Force 1, sea state calm', when ships in the Dover Straits were reporting 36 knots of wind.
At the end of October I was taking a yacht up the English Channel from the Solent to the Thames and we were enjoying typically grotty autumn weather. I did not want to wait weeks for a high pressure system to settle the weather; I wanted to make the most of what gaps appeared. I did some web browsing and invested in a MetOffice Marinecall subscription.
On Friday October 20 I downloaded my first forecast and was delighted to see Force 1 sea state calm, with nothing above a F3 for four days. It didn't feel that calm, the rigging was humming and small waves were reaching as far as our berth tucked well into a marina. A quick check online revealed shipping in the Dover Straits reporting 15 m/s (approx 30knots), with the MetOffice's own inshore forecast for North Foreland to Selsey Bill realistically predicting SW5 increasing 6 or 7. I emailed the MetOffice on both their own (24 hour) and their Marinecall (Mon-Fri daytime only) email addresses.
On Saturday, again, I got what I had paid for (!) - Another perfect forecast. Unfortunately the real weather was still awful, and again on Sunday, and again on Monday; two more dream forecasts as we hid down below from the wind and rain lashing the deck.
By Monday I had received a reply by email acknowledging the problem, but for at least four days the MetOffice had been sending out dangerously inaccurate forecasts. There is no way of knowing how many people made decisions to sail based on those forecasts. Like me, they may have been seduced by their attractive claim: "Handmade by humans Most forecasts are computer generated. Ours are written by Met Office forecasters. Trust your human instincts."[sic], quoting http://www.metoffice.gov.uk/leisuremarine/marinecall.html directly (graphic at bottom left of page.).
Yet to be this badly wrong, the forecasts I received must have been produced automatically, and quality controlled, by a computer, no human could be so consistently wrong? However, even the small print, that notorious resting place of get-out clauses confirmed that I could not blame computers for the error: "Met Office Atmospheric Weather Model creates a deterministic forecast that is quality controlled by Met Office forecasters. The model is run 4 times a day to offer the latest in local accuracy.” according to http://www.marinecall.co.uk/services/Dover.pdf
It turns out that the MetOffice use a company called iTouch to send out the forecasts. According to Martin Kidds, Customer Feedback Manager, Met Office. "On receipt of your complaint, we investigated what was causing these inconsistencies and traced the fault to a coding issue with the data after it had left the Met Office. Our partners iTouch have worked to correct this problem as a priority and I am pleased to report that the service should now be back to normal.”
So it was a computer error - so much for the quality control. These forecasts cannot have been checked, or if they were the person responsible should be made to venture out into the Dover Straits in a small yacht in the actual conditions prevailing at the time.
But surely they cannot be this cynical - are they really charging us extra for computer generated output that isn't checked by a human forecaster? It would seem that the only human generated part is the inshore forecast - which is freely available anyway on VHF and the internet. I believe this is confirmed by an email I received from Hayley Turnball, Weather Brand Manager at iTouch: "Sometimes there is a slight variance between the inshore waters forecast and the 6-hourly data and this is because the former is forecaster written and covers a vast area, whereas the latter is computer generated and covers a specific LAT/LONG.". Yes, it seem I am paying for a Marinecall subscription and the only bit that is human-forecast is the bit I get for free anyway.
iTouch immediately offered me a full refund and were very apologetic. The MetOffice were similarly apologetic and offered me a free week's subscription. I got the free week - the forecasts were much better though still only updated once per day, I am still waiting for the refund and my last email chasing the payment was ignored, but it has only been two months, still mustn't grumble - as my partner says, 'You could be dead'.
If it were a genuine error, instantly rectified, I would have forgiven them two months ago. But I was led to believe that I was paying for quality human forecasts, when in fact their mistakes revealed that I was getting computer generated data, which I get for free already from the NOAA (I believe http://www.weatheronline.co.uk uses NOAA data) and Deutscher Wetterdienst (http://www.dwd.de/de/WundK/W_aktuell/Seewetter/NOsee/Seewetter72.htm#EnglE.KAN).
I am sure the forecasts produced deep inside the bowels of the MetOffice are checked by humans. I naively assumed they meant that the forecasts I was _buying_ were both forecast and quality controlled by humans.
