I thought some might be interested in an old man's recollections of how weather forecasting was done back in the 1970s, before the internet and most of today's vast array of weather modelling.
This would be the North American view, perhaps some details were different in Europe.
I got into forecasting through a back door, after walking out the front door in university. My initial objective was to graduate from a science degree program and seek out a career in either astronomy or meteorology. These had been strong interests of mine since high school days. My parents had supplied me with the instruments required for a home weather station, which I kept from about 1964 to 1968 in a small town west of Toronto (Georgetown), and handed off to some friends with younger siblings who then kept it going to 1972. But I was equally interested in astronomy and made a couple of trips to open house sessions at observatories in the region.
At some point during my second year at university, I decided that I was more interested in climatology than either of the two sciences. Historical weather records fascinated me even half a century ago and I had become a frequent visitor to the weather bureau's library (it was on the campus of the university at that point in time, later on the headquarters moved out into a much larger building in suburban Toronto). They had monthly publications of all Canadian weather data going back as far as 1840 for Toronto and various other 19th century years for other locations.
So I graduated from that climatology program (in 1971) before there was any whisper of "climate science" or much concern about global warming. We actually went through an interval in the 1970s where prominent atmospheric scientists were debating the possibility of a severe cooling trend. That was generally abandoned after the strong El Nino warmings of 1982-83.
After doing some travelling and teaching for a few years, I started looking into career possibilities in the general field of climatology. Back in those days, your two options might be to work in the small but growing climate division of the national agency, or to take on a university teaching job. Climatologists were graduating at the rate of about three or four a year from that and one or two other colleges. It was not considered a science but a sort of book-keeping position to assist the science of meteorology, tell them what the climate statistics looked like and records etc, otherwise no input into anything that forecasters were doing. And I never really thought much about forecasting until I got hired on by a private forecasting company in Toronto to be a "met tech" to support their two-person forecasting office.
They had contracts to supply air quality forecasts for various large industries in Canada with pollution problems. In the early 1970s, air quality in Canada was at its worst point in history. Yellow smog was a frequent occurrence in the Toronto region, and air quality alerts were frequent. But we were working for even worse offenders in smaller places scattered across Canada, where the wrong blend of weather and operations could create very severe health risks. Governments had begun to legislate these operations much more strictly, and they needed expert advice on when to shut down to avoid getting into legal trouble. We understood going in that they were already scrambling to install better technology to clean up their mess, and that our work was a bit of a temporary fix.
So in getting involved in that work, I came to learn a lot about atmospheric stability and weather forecasting operationally. At that time, you did not have access to very much numerical weather prediction. There were two North American models that produced 12, 24, 36 and 48 hour panels of just 500mb and surface maps, then a 72 hour "prog map" would follow. The schedule for these was about what we are used to seeing nowadays although the 72h map (which was regarded as a shot in the dark) came out quite a bit after those main guidance products.
When I say "came out" I am referring to a technology you may never have heard about, fax transmission. Not a phone-based fax, but a separate communications network with a big, chemically rather stinky, machine called an "Alden FAX" which slowly generated your charts on paper that was still a bit moist from the chemicals (the smell was pungent like ammonia). Those had to be hung up on a sort of clothesline in a back room (to keep the smell away from the non-meteorological staff) until they had dried out, then were hung up on walls in places where the forecaster might expect to find them.
Of equal consideration as guidance, we drew up maps on a regular basis and even made our own "prog charts" using the models for some guidance and also a certain amount of subjective estimation. Maps were drawn up from weather data coming in regularly on teletype. The delay factor was minimal, probably less than the 15-20 minutes you have to wait to see hourly observations on the websites. "Special" obs would come in at random intervals as they were observed, but otherwise, you would get the local regional "runs" of data first then clusters from further afield until you had basically all the current obs for North America hanging on clipboards around the forecast office (after being used for plotting maps). I gained a reputation as a good map analyst and the senior forecaster at the company started asking me informally questions about forecasts. So did the staff (mostly weekend issues) and I then began to take on graveyard or weekend shifts that the forecasters didn't want, to continue their work programs. This would have displeased the boffins over at the government agency but then almost anything would.
During that time I also worked up some computer models to predict lake effect snowfall in an effort to give the company some tools to use in seeking out contracts. It so happened that the years I was working there had heavy lake effect snowfall. A computer back in those days was a fairly large unit although not quite as monstrous as the ones I encountered a few years earlier in my college days, those filled entire rooms and were fed by punch-cards. This newer variety of computer was like what you may have seen at NASA during the Apollo missions. We designed some programs to print out snowfall forecasts based on various parameters such as wind direction, temperature at start and finish of lake crossing, dew point, stability parameters, etc. And it seemed to work fairly well. The company showed this off to some visitors from the government agency and they basically stole it from the company around the time I was leaving to go somewhere else. (Somebody took notes on how it was done and passed it off as his own work over at the headquarters). Plus ca change ...
Another detail I could reveal is how we got radar information back then. We did have weather satellite imagery, it was pretty fuzzy and once again came in over the good old Alden FAX. But radar information was only available in digital form. Basically you had a read-out of distance from site, degrees of vector to return, and an estimate of height of the cell producing the return. And that was all you had. Our severe weather forecasting had to be based more on synoptic meteorology and clues from satellites, the main use of the radar was to get a fix on where the more obvious cells were located, and rates of movement, but all of today's technology was in the developmental stage. Maybe some agencies had a form of it available, but we didn't.
So how was the forecasting? Actually, I think it was as good as or better than today for the first 24 and possibly even 48 hours. Forecasters had more hands-on experience with the weather and you couldn't model hug because the models could go tits up within 36 hours. I can remember one storm in April of 1979 that went off at about a 45 degree angle to where it was supposed to be going. But the models were fairly reliable for most weather patterns within the first two days. The third day was regarded as a bit of a crap shoot and a 96h experimental product got a very brief once-over about like you might assess the current 10 to 16 day portion of the GFS.
With that better feel for actual weather, there tended to be a fairly high success rate in nailing down details. I remember one validation study for a forest fire support program where we were over 90% within 2 deg on temperatures even for the second day, and we got very positive feedback about our wind and stability profile forecasts. So it wasn't hit or miss as you sometimes hear about forecasting. What was different was that nobody attempted to go much past three or four days and "next week" was regarded as wait and see.
Long range forecasting existed but was not practised very widely or by most larger companies or government agencies. I knew nothing about it in the 1970s and what I did learn came from reading Hubert Lamb's ground-breaking three volume opus published around 1977-78. People knew the Farmers' Almanac existed but studies had already been published in weather magazines to show that its method was an analogue from a past winter and the results validated were not significantly better than random, so in the profession that was ignored while in the general public, you would occasionally hear people saying something about the predictions. Anyway, for some outfit that liked to point to sensational winter scenarios, the mid-1970s turned out to be glory years for them.
I was arriving at the forecast office on the morning of the famous January 26 1978 superstorm or Great Lakes Blizzard. We knew a big storm was coming, the models had shown a low deepening to 960 mbs over Lake Ontario and so of course I was excited to show up for work that morning. About the same time as I drove into the parking lot, an air conditioning unit landed in the next parking space and I could hardly walk from the car to the front door of the office with the wind howling. I noticed it was southwest but cold, which told me the low must have tracked further west. As soon as I got in and drew up a map, I could see it was a 955 mb low over Lake Huron. The duty forecaster showed up soon after that and looked at my map, and I saw quite an expression on his face. "Are you sure about that 955 reading at Sarnia?" was the first question. We saw some obs that nobody had seen in meteorological history, at least in our hemisphere, screaming south winds with lake effect snow squalls and rapidly falling temperatures (London, ON around 9 a.m.). Now that was wrap-around. We got that blast shortly afterwards and the temperature was falling visibly for several hours. Anyway, we were able to warn one of our clients in northern Ontario that a bad storm was coming in and they should expect to be snowed in at their offices. In Toronto there wasn't a lot of snow from that, maybe four or five inches, but the storm did tremendous damage to the hydro main lines in southwestern Ontario, blowing down some of the pylons near Lake Erie.
That's part one, I will post part two and tell you a bit about my brief forecasting experiences at Accu-weather and then how I got into research and long-range forecasting after that. Part three you already know about. (MTC 2009-20 plus a bit of this and that).