Cost/Benefit Analysis in Weather forecasting

The Riposte & Parry is a private forum for serious discussions.
occasional thoughts
Grand Pooh-bah
Posts: 2744
Joined: Sep 6th, 2006, 11:07 pm

Cost/Benefit Analysis in Weather forecasting

Post by occasional thoughts »

My wife and I are heading to the Texas gulf coast in about four weeks for a little spate of birdwatching, beachcombing and general relexation. I like to get a sense of weather expectations before I go somewhere, so I Googled to find out what the local USNWS (National Weather Service) office(s) is (are) for the region. (Corpus Christi and Brownsville, for the record.) The U.S. federal government maintains 122 local weather offices!

Juxtapose this to Canada, where, after federal government downsizing over the last decade, we barely maintain one such per province. The newish Mountain Weather Office in Kelowna, itself built to centralize meteorologists and forecasters in the 1990s (I believe) from such places as Penticton, Prince George and Whitehorse, was closed, falling victim in turn to centralization of weather services for B.C. and Yukon in Vancouver over the last decade.

Two points:

1. Weather forecasting these days and for several years now for the Central Okanagan is a joke. How often right vs. how often wrong? Very heavily weighted to wrong, I believe. I'm sure the Canadian weather service does verification stats on its forecasts; I'd love to know the results for our region. It has to do with knowledge of "local conditions" according to my retired weather forecaster friend. If the Kelowna experience is typical in other provinces, I think the federal government has gone too far in its centralization of weather offices in Canada.

2. On the other hand, I'd say the U.S. weather service at 122 local weather offices nationwide is bloated, a proxy for the troubles of that great but declining nation. The U.S. weather service is a Cadillac.

Nothing much beyond partisan bickering gets done anymore in the United States, because someone's ox is going to get gored, and someone in or seeking public office will take it on, and no one will make a move. The U.S. is going to go broke one day. I'm still in shock from realizing that the big political kerfuffle in the U.S. a couple of months ago was not about getting a grip on its debt and deficit, but on whether the politicians would agree to raise the spending/debt limit for the federal government.
User avatar
Glacier
Admiral HMS Castanet
Posts: 33255
Joined: Jul 6th, 2008, 10:41 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by Glacier »

I apologize in advance for a LONG post, so you'll have to bare with me.

I'm not sure that the number of weather offices necessarily reflects the costs, rather it's likely a reflection of the amount of attention paid to field recording stations. In my view, we should model our weather recording after Australia where they have excellent spatial distribution. I'm not sure how much it costs, but I suspect that weather monitoring and forecasting is a very small fraction of any country's budget. Even within Environment Canada, it's only one of many departments.

Here in Canada we have adequate monitoring in populated areas, but very little in remote locations. As a matter of fact, we have less monitoring in rural and remote areas than we did 40 years ago. Much less. The problem with centralization is that it typically leads to specialization whereby everyone is hired to one thing and one thing only. Before I explain what I mean, I need to provide a little background introduction.

At one time Environment Canada had meteorologists all over the place recording weather conditions. In 1947 when Canada received her coldest cold snap on record, meteorologists were there in Snag, Yukon, to record the event. Over 60 years later as our population (aka tax-base) has tripled, we have no meteorologists in the Territories recording the weather. Instead, we have automated weather stations that cost $100,000 to install and $11,000 per year to maintain doing all the work. In the long run this might save a few bucks, but the integrity of automated data is frequently subject to errors.

Environment Canada performs quality checks on the data, but since the people doing the checking live in large centres far away like Toronto, they don't actually recognize errors when they occur because they don't know what the climate is like in those locals. I can give you dozens of examples of such mistakes that have gone unnoticed, but I'll spare you those details, except to say, I once emailed them about fixing some of brutally obvious ones I discovered, but all they did was thank me for pointing them out without actually correcting anything.

Another important part of Environment Canada's weather record used to be (and still is) the volunteer observation stations. In the 1980s we had 2,500 of these volunteer observation stations across the country. Today there are just over 600. Each one of these stations costs about $2,500 a year. To re-emphesize this point, the number of cheap, cost-effective, and reliable weather stations today is about 25% of what it was 30 years ago.

In a populated area like the Okanagan, we are doing okay because we at least still have weather stations, but less populated areas completely non-existent today. This is more than a bit ironic when you consider how important climate change is supposed to be. In the Chilcotin area (also known as the western half of the Cariboo), there are no longer any weather stations recording snowfall. In 1980 there 13. Heck, even as far back as 1894 they were recording snowfall in a place called Big Creek, but as of 2006, that is all history now.

Now that you've heard me through my introduction, let's talk about centralization and why it's not all that it's cracked up to be. I'm sure there are times when centralization is necessary, but overall, I believe it leaves us worse off. For every savings there is a hidden cost.

During the 80s many of the forestry offices around BC kept weather records for Environment Canada, but then the powers that be decided it was time to start closing rural locations (long before the Campbell government decided to centralize further).

Centralization is purported to cut labour costs, but at what cost? In 1980 forester John heading out from his office in the West Chilcotin community of Tatla Lake and do his field work could be at work in a few minutes. Today, he gets paid to make a 400km round-trip drive from his office in Williams Lake conduct the same work. A local workforce provides a better social fabric than a transient one.


Centralization contributes to urbanization of our population while killing the rural populations and school budgets. In the early 90s there were over 70 kids in the local school (K to grade 10) when the NDP government build a brand spanking new school. Today there are 13 with only one classroom operating. The one kid in kindergarten might not have anyone within two years of their age to play with today. Unfortunately, this puts great strain on the school district's budget because the district has no choice to operate this school at almost the same operating cost because the nearest school is way too far away to bus kids.

Once the population decreases, so do the services. A few years ago Global TV did a little piece about how the government (Interior Health) was going to downgrade the local nursing clinic. After a protest, they changed their mind, and thankfully they did because a little boy's life was saved just this past fall because of the medical staff on duty. He had to bet sedated and medevaced to BC Children's Hospital.

Along the same lines there has been a push centralize lighthouses in BC, but what they don't tell you is that many of the manned lighthouses along the coast provide weather monitoring. I've highlighted three sparsely populated areas of BC in the following graphs: the Cassiar Highway area, the Chilcotin, and the Central Coast. Notice how diminished the weather stations have become in the two areas with no lighthouses. (BC weather stations that recorded precipitation for the entire month of January in the years given are shown.)

1981weatherstations.jpg
2006weatherstations.jpg
2010weatherstations.jpg
You do not have the required permissions to view the files attached to this post.
occasional thoughts
Grand Pooh-bah
Posts: 2744
Joined: Sep 6th, 2006, 11:07 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by occasional thoughts »

Glacier: interesting, nay, fascinating. Are you a retired meteorologist or forecaster or maybe a forester? I think we're largely on the same page, but . . .

My post tried to address two issues: government expenditure/downsizing; and weather forecasting. You've brought in the collateral and important issue of weather observing as well.

I'm no right-winger, but I do believe that governments do have to get and keep their expenditures under control. That said, I feel that Canada has gone too far in cutting costs over weather forecasting while the U.S. hasn't even started, and this is a small example of why the U.S. deficit is in the trillions and counting.

Does the U.S. need a full-blown NWS office in Pendleton, Oregon when it has offices in Spokane, Portland and Seattle? It can't hurt, but they're not free. I'd say that Pendleton is probably a luxury. By the same token, I think Canada has gone too far by closing the Mountain Weather Office in Kelowna, and probably the offices in Prince George and Whitehorse as well. (I'm unfamiliar with the situation on the Prairies and in the NWT).

Castanet uses Environment Canada for its weather forecasts. For the most part, they're a joke. Two days ago the high on Tuesday was forecast at +6, which looked bad for x-country skiing. Today it is offered as 0 with a low of -5, which would give potentially great conditions. I'd really like to know the verification results on Central Okanagan forecasts since the Mountain Weather Office was decapitated and the function centralized in Vancouver. I suspect they'd be abysmal, and I think the department and the government should sweat over them and attempt to justify that action.

I could be barking up the wrong tree, however, because my friend(s) in the weather service seem to be thinking that the federal government wants out of weather entirely. Privatize it to, say, the Weather Network. I just don't know where we are going these days with some of the cost-cutting that government is doing.
User avatar
Glacier
Admiral HMS Castanet
Posts: 33255
Joined: Jul 6th, 2008, 10:41 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by Glacier »

I apologize for my little tangent. l guess I forgot to explain the correlation. When you reduce the number of weather monitoring stations to a certain level, how do you know if your forecasting was accurate? I have no weather training, so I can't say for sure that verification is necessary, however, I think it could work like this:

    They forecast 2 cm of snow for region X that has no weather station, but for whatever reason 20cm falls. They expect a slightly larger storm to push through the next day, so they mention the likelihood of up to 5cm of snow since they have no idea that 20cm fell the day before. Everyone is caught unaware when they get hit with 30 cm of snow.

When you talk about the number of weather offices in the states, you have to remember that the number of offices doesn't necessarily correlated with the total cost. How much does the U.S. spend now compared to generations past (per capita)? How about Canada? I do not know, but it would be interesting if someone would dig that up.

Maybe there is good reason to have the Pendleton, OR, office open. The climate on the east side of the mountains is starkly different than that of Portland. Perhaps land is so much cheaper over there that it actually saves them money to have a small office there instead of having a larger Portland office.

I'm no left-winger, however, I believe that there are many non-monetary advantages to decentralization that should always be taken into account before centralizing services.

As for forecasting in Canada, the first two days are looked at by humans, but after that the numbers are computer generated. It appears from looking at the 5 day forecast many times over, that Environment Canada's computer model tends to converge towards the seasonal average at the end while The Weather Network's computer model tends to diverge to one extreme or the other. Mark Madryga mentioned something about this on one of his forecasts (the EC part). He stated that even though the official 5 day forecast calls for more seasonal forecast, he fully expected the heatwave to continue well into the next week.

The say it takes too way too much manpower to display human generated forecasts beyond 2 days in Canada; it would be interesting to know how this compares to the U.S.
occasional thoughts
Grand Pooh-bah
Posts: 2744
Joined: Sep 6th, 2006, 11:07 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by occasional thoughts »

Yesterday the Environment Canada high for Kelowna on Saturday was forecast to be 3, to day it is forecast at -1 . . . that sudden and significant change, and to close to the actual day in question, to me is sloppy bordering on incompetent. Who know whether it will actually be a pleasant above-freezing mid-winter day or a chilly below-freezing day: stay tuned for the verification!
User avatar
GenesisGT
Guru
Posts: 5185
Joined: Jun 19th, 2010, 12:21 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by GenesisGT »

I'm not sure that the number of weather offices necessarily reflects the costs, rather it's likely a reflection of the amount of attention paid to field recording stations. In my view, we should model our weather recording after Australia where they have excellent spatial distribution. I'm not sure how much it costs, but I suspect that weather monitoring and forecasting is a very small fraction of any country's budget. Even within Environment Canada, it's only one of many departments.

Here in Canada we have adequate monitoring in populated areas, but very little in remote locations.

At one time Environment Canada had meteorologists all over the place recording weather conditions.

Environment Canada performs quality checks on the data, but since the people doing the checking live in large centres far away like Toronto, they don't actually recognize errors when they occur because they don't know what the climate is like in those locals.

Now that you've heard me through my introduction, let's talk about centralization and why it's not all that it's cracked up to be. I'm sure there are times when centralization is necessary, but overall, I believe it leaves us worse off. For every savings there is a hidden cost.

In a populated area like the Okanagan, we are doing okay because we at least still have weather stations, but less populated areas completely non-existent today.


I would just like to make a few comments about the number of weather stations and forecasting.

I cannot think of any Environment Canada (EC) manned weather observation stations. The majority of weather reporting stations are located at airports and whether they are manned or automated these weather stations are operated by NAV CANADA. Transport Canada used to provide this service, but in 1996 TC sold off the Aeronautical Navigation Service to NAV CANADA which is a private company. Meaning that the weather observations at Kelowna, Vancouver, Pentiction airports, etc.. across Canada are provided by NAV CANADA. TC had approximately 124 manned sites across Canada, NAV CANADA today has approximately 67. NAV CANADA also has numerous contract weather stations and unmanned automated stations. Even at manned sites the majority of the weather observation elements are automated. I can remember going outside every hour checking thermometers, reading the barometer, visibility, checking the wind speed/direction, looking for lightning, all these elements are automated even at manned facilities. Most manned facilities only do a quick quality check and add any supplemental info as required. QA is mostly done electronically by the weather reporting system. Because NAV CANADA (NC) is focused on providing weather info for aviation, they have little requirement for weather observations at every small community.

For forecasting again NAV CANADA is probably the biggest paying customer to EC for forecasts. NC requires forecasts for individual airports (Terminal Airport Forecasts) and large area geographical forecasts (GFAs). NC centralized its aviation weather briefing locations across Canada from 39 to 6, plus Whitehorse in 2004. At this time NC approached EC with the same concept, "Since we pay the bill for aviation forecasts and we centralized our briefing sites, EC we want you to centralize your forecast offices." This was the end of forecast offices such as the one that was in Kelowna, and centralized aviation and public forecasting to, I believe four offices across Canada. As with every story there is lots of behind the scenes activity. By the way every time you fly you might have noticed "NAV CANADA Fees" on your ticket, this is how the company collects money to pay for weather observations, and a lot of other things. (Air traffic control services, navigational aids, etc)

A lot of effort was put into the centralization by NC and EC. For the aviation side NC contracted EC to have forecasters travel across Canada and document the weather patterns and hazards to aviation. If you are interested the link to these manuals is

[url]http://www.navcanada.ca/NavCanada.asp?Language=en&Content=ContentDefinitionFiles\Publications\LAK\default.xml[/url]

You may also be interested in the following comment from the EC website.
As of Dec. 15, the OTTAWA MACDONALD-CARTIER INT’L A and HAMILTON A airport stations will switch over to new partner maintained and operated stations. Due to a new format, hourly data from these stations will not be available on our website. We ask for your patience while we are working to upgrade our system to display these data. If you are interested in daily data for these stations, they can now be found under their new partnered station names, OTTAWA INT’L and HAMILTON.

http://www.climate.weatheroffice.gc.ca/Welcome_e.html

This is because of NC changing the weather observing format to meet ICAO Aviation requirements. As NC owns and operates weather stations across Canada and their priority is aviation weather reporting, EC will be in a game of catch up as NC has to meet ICAO requirements.

A quick note on the operations in the USA. I contracted to work on the their aviation requirements. And my quick comment is that their system is driven by special interest and it is a wonder that they don't have a weather/forecasting office in every town with a population of ten or more.

Hope you found this interesting, and questions either PM or post them, and I will try my best to answer them.
Ladies and gentlemen, the future lies in the future!
occasional thoughts
Grand Pooh-bah
Posts: 2744
Joined: Sep 6th, 2006, 11:07 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by occasional thoughts »

I admit, the premise of my original post was kind of convoluted. Basically, I acknowledged that benefit/cost analysis had to come to weather observing/forecasting because unlimited ongoing public expenditure on everything is unsustainable, but I wondered if it had gone too far in Canada. And I wondered why it did not appear to have even had a start in the U.S., which is bleeding deficit and piling on debt and seemingly incapable of doing anything about it. Your comment about every U.S. village, town and city with a population of 10 or more having its own weather office, while cryptic of course, hit the nail on the head.

In light of what people have written here, I guess the question for Canadians is, who should be responsible for the public forecasts, and would informing the public of these issues make them (us) better advocates of their own interest. To go to the bottom line, will the Weather Network ultimately become our public forecaster? Is that necessarily a bad thing?

As I understand things, the Weather Network can do everything except issue weather warnings; these seem to remain in Environment Canada's purview. I'm also lead to believe by a retired forecaster friend of mine that the Weather Network is very good in Central Canada and esp. Greater Toronto, but very weak in the country's extremities (B.C.). That is not beyond correcting.
User avatar
Glacier
Admiral HMS Castanet
Posts: 33255
Joined: Jul 6th, 2008, 10:41 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by Glacier »

That is very interesting information, GenesisGT! I'm curious as to why so many stations record temperature but not precipitation. Is it because NAV Canada is more concerned with the temperature, or is an EC thing? How about snowfall. It seems that very few weather stations measure daily snowfall amounts these days.

O.T., I'm curious as to why you think weather forecasting is inadequate in Canada (or if you think so)? I mean, the weatherman was able to tell us well over a week ago that we were going to warm up significantly on the weekend. The lay person like me would not have been able to predict what was coming.

Also, in another thread, someone posted an article about how the size of the Canadian government has increased by 50% in less than 10 years (or something along those lines). Has the weather depart at EC increased over the past decade as well?
occasional thoughts
Grand Pooh-bah
Posts: 2744
Joined: Sep 6th, 2006, 11:07 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by occasional thoughts »

Glacier (ah, that tribute to ice ages), I think that public weather forecasting for the Okanagan is dubious. Over time I've seen many sudden reverses by Environment Canada in our public forecasts, correcting to weather and temperatures that were predicted days earlier by the meteorologists of Spokane NWS.

I like to follow the Spokane NWS forecast discussions, which offer intelligent discourse on emerging weather patterns for their forecast area which includes the U.S. Okanogan from Oroville down to Omak and then some and back west to the Cascades Crest. Environment Canada used to publish forecast discussions, but they are now suppressed; the public is not allowed to view them.

I've learned from following weather for many years that lazy or hesitant meteorologists like to fall back on "climo" or climatology (the norms or long-term averages) when they're forecasting 3, 4, 5+ days out, and they don't get serious until the weather forecast is for, say, 2 days out; thus, the sharp reverses of too-high frequency. I suspect that the forecasters in Vancouver are content to practice "climo" until the forecast is for the next day or the day after, or when overwhelming conditions force them to take greater care. I would be embarrassed to be the author of many of Vancouver's forecasts for us (and probably elsewhere in the B.C. interior and the Yukon).
User avatar
GenesisGT
Guru
Posts: 5185
Joined: Jun 19th, 2010, 12:21 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by GenesisGT »

Glacier, I am glad you found the information interesting. As for recording temperature only, it is definitely less expensive to install the sensors for temp/wind/pressure then to have the add-ons for cloud height, visibility and precip. I think you will find NC normally installs a full package and has added cameras for pilots to get a better picture of the wx conditions not reported by the sensors. In the good old days with manned stations we had rain gauges to measure the amount of rain, and the handy ruler to measure the depth of new snow, and a snow gauge to measure the water content. Which was very accurate, we had automated equipment for rainfall, but human intervention was required at times. You had commented on the difference between 3mm of snow and 3cm of snow before. You will only get snow measurements at manned stations. And there is two measurements the depth of new snow fall and the depth of snow measurement taken at manned stns every six hours. The depth of snow is not an accumulation of new snow, as snow compacts, melts, etc. As you noted NC is more concerned with aviation weather requirements, but EC contracts to NC for additional climate information. But as NC implements more automated stations additional climate data is lost. Automated stations do not provide information such as actual snow depth.

Don't know if you have seen the maps for snow depth on this page

http://www.weatheroffice.gc.ca/analysis/index_e.html

Here is the link for the Kelowna airport NC automated station with cameras.

http://www.metcam.navcanada.ca/hb/player.jsp?id=12&cam=26&lang=e

Is weather forecasting inadequate in Canada. Good question. I am not a meteorologist. But if I want to know what the weather will be, I will check the aviation weather sites and do my own forecast. There are lots of historical weather radar, last 24hours, and forecasts for system movements. Especially if it is for something important like determining a golf day or which day to travel.

The only comment I could make on staffing, is that the aviation forecasting staffing has decreased.

Looks like things will get better
http://www.thestar.com/article/1118901--environment-canada-given-78-7-million-boost-to-improve-weather-warning-systems

occasional thoughts, I do agree with you that looking at the american weather forecasts is a good way to forecast ours. The aviation wx briefers were given access to ADDS, NWS and NOAA wx information and forecasts and it was one of the best things that was done for them.
Ladies and gentlemen, the future lies in the future!
occasional thoughts
Grand Pooh-bah
Posts: 2744
Joined: Sep 6th, 2006, 11:07 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by occasional thoughts »

It's February 29 and snow is falling, in West Kelowna at least . . . a leap year present.

I was amused to see pop up on my Facebook ads a couple of weeks ago the beginnings of a campaign to "Save the USNWS" from President Obama's cost cutting (apparently) plans. As I've said before, I suspect there's way too many full forecast offices in the U.S., and I know there's too few in Canada. More later on this topic.
User avatar
Glacier
Admiral HMS Castanet
Posts: 33255
Joined: Jul 6th, 2008, 10:41 pm

Re: Cost/Benefit Analysis in Weather forecasting

Post by Glacier »

I think I may have stumbled across the reason why the number of weather stations in Canada has been greatly reduced over the past decade:

    The Auditor General reported that between 1998 and 2005, Environment Canada (EC) spent $6.4 billion on climate change. The money came from reducing data collection and other services. Weather stations were closed, some replaced with Automatic Weather Observing Stations (AWOS), to the detriment of the record and safety concerns.


related Castanet topic: viewtopic.php?f=27&t=33069#p1021988

Return to “Riposte & Parry”