Once upon a time, only wise men could foretell weather, and they did it with a mix of gut feel and observation of animal behaviour. How often they got it right is hard to tell, but they were the only ones to rely on. Later, until the turn of the 19th century, they compared the day’s weather with historical weather records, and came up with a forecast, assuming that history would repeat itself. In the early years of the 20th century, they realised this method was just not accurate, and thus began the search for newer ways of predicting the weather.
In 1922 — an epochal year in weather forecasting — an English mathematician called Lewis Fry Richardson (LFR) announced he had found a way, which was by the use of — guess what? — mathematics. If you look up LFR’s book on the internet (Google Weather Prediction by Numerical Process by Lewis Fry Richardson) and flip the pages, unless you are also one of those mathematical whiz kids, you’ll faint — I almost did. Page after page is filled with equations. I understand he used partial differential equations.
64,000 ‘computers’!
LFR tried to sell his method to the authorities to get a forecasting contract. His idea involved breaking the world into 12,800 boxes between latitudes and longitudes, measuring some physical parameters, and using 64,000 computers to do some tedious calculations and then say whether the sun would shine or not in the immediate future. (Ha, ha, you think you caught me on the wrong foot, don’t you. You spotted a fatal flaw in my narrative. Back in 1922 where were the computers? Nay, LFR’s ‘computer’ meant ‘man who computes’. Yes, he wanted to employ 64,000 people for the job.)
He didn’t get the contract. It was not only way too expensive relative to the benefits of forecasting but, more importantly, by the time his ‘computers’ came up with the prediction for the future, the future was already upon the world, rendering the whole exercise futile.
But LFR is regarded as a pioneer in numerical weather prediction, the method still in vogue. Only, far more advanced supercomputers do the job that LFR wanted his 64,000-strong team to do, and they are quick. According to an estimate (provided by one Dr Karl Gutbrod, CEO of a Swiss forecasting company called Meteoblue) says a good weather computer requires 500,000 lines of code, generates a lot of heat and consumes a lot of electricity. They process 8.4 billion data-points per day. And today, we can get a good forecast not for a region, but for an area as small as 3 sq km. All thanks to Lewis Fry Richardson.
Prediction is crucial
Now, why are we discussing weather forecasting in a Cleantech column? Because forecasting is becoming integral to clean technologies. First of all, with climate change making the weather more fickle than ever, prediction and post-prediction action becomes imperative. Also, we are adding renewable energy so much and so fast that, without proper weather forecasts, running the electricity grid would be impossible because of the yes-no nature of renewable energy. For example, erratic winds will cause wind power generation to zig-zag, and that’s a problem. Can you — to give you an oversimplified example, run a factory if the machines keep slowing, stopping or running very fast?
If you want to have more wind and solar in your energy mix, you can’t do so without accurate forecasting which, fortunately, is possible today. In India, we tend to ridicule our India Meteorological Department. (Reminds me of a cartoon showing a row of people standing at a bus stop, waiting for the bus. All but one have come prepared for the rains, with umbrellas and raincoats. The only one who is unprepared and standing in the drizzle looking miserable is from the Met department.) But if you have a friend in, say, the United States, ask him or her about weather forecasting there. If the weatherman tells you there will be rains from 4 pm, it will begin raining at (or around) 4 pm.
India is cruising towards that level of accuracy. Companies like Meteoblue are getting into action. (‘Meteoblue, close to you’, is its tagline.) The wind industry badly needs weather forecasts to calculate wind speeds, which determine the output of energy.
Weather management
Today, India has 25,000 MW of wind power and the government wants to raise this to 60,000 by 2022. But that level of ramp-up can just not happen without wind power producers being able to tell how much they will generate in every 15-minute interval the next day. The government wants wind power companies to do what is known as ‘scheduling and forecasting’. (Even solar would need weather forecasts, but not as much as the wind power guys do.) State electricity regulators are framing laws for scheduling and forecasting — at present, Tamil Nadu, Karnataka and Madhya Pradesh are ahead of others.
A good guess is that the entire country will be covered by such laws and all wind power generators will be required to state their expected output to the grid operators. Once you know how much of a product is coming and when, it is possible to develop a market for it.
Thus, the importance of ‘weather forecasting’ is growing by the day. I wouldn’t be surprised if, in the not-too-distant future, a university or institute offers an MBA in Weather Management.