Forecasting then, and now -- you might be surprised

This is going to be a rambling discourse about forecasting the weather back in the day, as compared to nowadays. It was quite a different world as recently as the mid to late 1980s.

My own experience in forecasting goes back to foundations that are probably similar to many of you on this weather forum. As a high school student, I was marginally interested in the weather but more into astronomy. Despite that, my parents decided to provide me with weather instruments (the kind you have to read manually) at Christmas of 1963 and so I started a daily record of observations that New Years which I kept up for five or six years. When I went off to university and wasn't around the weather station much, a younger friend took on the job and I continued to observe on weekends and in the summer holidays. This was all in a town just west of Toronto where as you can imagine the weather is quite variable and occasionally rather dramatic. Then when I moved north after graduating and lived in what Canadians call "central Ontario" (not that central) I was in a lake effect snow belt and kept up the observations through several winters of rather extreme weather in the 1970s.

However, I never intended to make a career in this field. My education was both math-science and geography oriented and naturally this gave me some courses in climatology, whether or not I graduated as a climate scientist is a matter of opinion rather than fact, but I didn't bother with graduate school, thinking instead that I might want to teach high school (math and or geography). That career was cut short by the untimely death of my mother, herself a teacher, during the final stages of the training year, which basically gave me somewhat more flexibility than I was perhaps expecting, and a chance to travel more (and work less) for a year or two. Eventually, I landed on my feet and determined that maybe I should pursue my interest in the weather. You might have pictured that I was like the most avid 20-something weather geek squared, but to be honest I had my mind more on other things such as playing soccer and well let's not go any further down that road.

In the period 1976 to 1980 then I worked for various private weather companies, first in Toronto and then at Accu-weather where I was hired to help them develop Canadian clients. During that stint, I met Joe Bastardi who is pretty close to my age, and a few other people you might have seen mentioned in their press releases. What I thought might be interesting is how weather forecasting was done in the 1970s and early 1980s, as well as some brief reflections on what it was like even earlier and also since then.

It might be worth exploring very briefly why I was forecasting at all, as I was hired on initially to do technician duties. The Toronto company had a rapid increase in business but couldn't find enough capable forecasters to fill out a 24/7 roster. They had one "senior man" and a couple of part timers on loan from the nearby government offices, but needed more than that. Since I was interested in forecasting (although not really thinking much about being a forecaster) I got sucked in and the rest is history.

In the mid-1970s there were basically two weather models available for forecasting. The better of them was known as the LFM or "Limited Fine Mesh" and it cranked out forecasts for the first 48 hours as well as some experimental maps for 72 and I think eventually 96 hours. Nobody took those overly seriously. All of the weather guidance used to arrive in a weather office on a system called "Alden FAX" which involved some rather stinky chemicals, a continuous feed of paper soaked in those chemicals and newly printed maps that you had to hang up to dry on a sort of clothesline before it was safe to read them. The schedule was pretty much what we're used to, but there were no 06z and 18z forecast maps, just the 00z and 12z runs. There was no European model that we knew about anyway.

Our data arrived on teletype, and on the same schedule that you can see current data nowadays on your computer. There was some limited computer to computer transmittal of data too. Desk-top computers were very new, I can still remember in my university days going into a computer lab where there was a huge machine with vacuum tubes and fortran cards for programming. Bill Gates was probably hitch-hiking around Europe about the time I was working in these weather offices. But while some weather maps came in on the dreaded Alden FAX and could be studied, the practice was to plot and draw up (analyze) your own maps, then you would look at the guidance and draw up your own progs. Many of the concepts that are well known to thousands of weather enthusiasts, not to mention professional forecasters, nowadays were completely unknown back then, but I have to say that the accuracy of 24 and 48 hour forecasts (and maps) was not vastly different.

In fact, I have the feeling that being that hands-on made you more sensitive to nuances that you don't get by just looking at maps and charts, and it could lead to some pretty good forecasts. The accuracy often degraded faster than what we're used to seeing nowadays. It was considered a coup to get to the end of a long weekend without totally losing the plot, so to speak. Sometimes the guidance was horrendously inaccurate very early into a storm cycle, not often, but the higher resolution models we have now would not often do what the 1970s version models did about once a month, which would be to take a storm and track it way off its actual course. I can remember one low in April 1979 during a rather cold spring that was supposed to move due east at about 45-46 deg N and ended up going southeast into Tennessee and North Carolina. You can imagine the havoc that caused in short-range forecasts.

One of my best memories is the "Superstorm of January 26, 1978" which occurred on a day that I was working a shift. We knew the night before that something big was coming, but the models were a bit off on the details. As I got into work, I noted that the air conditioning unit from the roof of the office was now in my parking space thanks to the strong wind gusts. And as I drew up the map, I could not quite believe what I was seeing (we were very lucky to get the observations and to keep the power on). London Ontario had something like a south wind at 60 mph and snow from Lake Erie, and it was about -8 C, while in Toronto it was 4 degrees with rain and an east wind. And the pressure at a station on Lake Huron was 955 mbs. I still have that map around my home office here, quite a sight. I also had to draw up the blizzard off the east coast that followed in about two weeks.

So the tools have been described, but how about process? At first I was working as part of a two or three person team, or sometimes alone. So we would get around a table and discuss what was likely to happen in the forecasts of interest, as there were so few models it was not much of a debate, and most of the time it was all about interpretation issues. At Accu-weather, where you had a much larger group (30-60 forecasters all in one rather small room in what I believe was a converted church) the process was one of managed discussion. A team leader would get people to submit ideas on paper mostly, such as drawing up snowfall maps when there was a regional snowstorm. Then all of them would be pinned up and people would walk around like folks at an art gallery for a while, and have a conversation (an argument, whatever). Consensus was often something like a weighted average, the weight being experience rather than force of argument (sorry Joe, I had to say that).

My big coup in that period was to draw an 18" snowfall max on a map of the big Leap Year Day blizzard of 1980 and have it verify, although not necessarily accepted prior to the fact. After that some of the bigwigs took a bit more note of my eccentricities in precip forecasting which got a severe workout because in the spring of 1980 there was one closed low after another over New York, New Jersey and Pennsylvania where most of the clients were. We had one day with about five inches of rain in northern New Jersey. They were very meticulous about warning clients who needed to know flood potential about such matters. I think it was done as well back then as it has been done since.

A lot of other things in the weather field were quite primitive back then. Television graphics were at an early stage, most of the presenters drew maps either on paper (while you watched) or with chalk, some got adept at drawing numbers in mirror image so it would look right on the far side of a glass screen. Really, not making this up. Discussion of the coming weather was pretty basic and the forecasts were not all that detailed. Warnings were not as prevalent and as the audiences were often more limited in geographical spread, a big city TV station rarely said much about weather in the nearby rural areas or snow belts. There was much less awareness of lake effect snow storms and I did some pioneer research into patterns and forecasting (which some knob at the government agency stole and published as if he had done the work).

My interest in forecasting quickly morphed to global climate and long-range forecasting while at Accu-weather. Before that, I don't think I ever tried to make a monthly forecast. There were very few techniques available back in those days, people seemed to go mostly on pattern recognition and instinct. Some of the results were remarkably good considering that fact but it was the time of amplified patterns and big anomalies so perhaps it was easier than some recent years. By the middle to end of 1980 I had decided to go into that sort of work and I was able to sell long-range forecasts on a mail subscription basis. Publicity was always the most difficult aspect, I was able to retain about 80% of paying clients from year to year but how to reach thousands of them? Also mailing costs got rather prohibitive after a while. I had other work and let this "peter out" in the late 1980s. We could see that something like the internet was coming but didn't know exactly what shape it would take. Meanwhile, more for research than urgent work, I needed weather data and the only way to get it was either walk into an airport weather office and take the expired data printouts, or buy it over a primitive phone dial-up transfer. So I ended up spending many thousands of bucks that I had earned in my day job for weather data which was then absolutely free by 1996 or thereabouts in the very early stages of the public internet.

I have a feeling this post is already too long for boards but I will post it and finish up with a second post that should follow in a few minutes. If you feel moved to post a comment, could you hold off until you see a second post? Thanks.