Meteorologists Failed To Predict A Single Hurricane Right In 2013 So Why Should We Trust Them When

Meteorology: Why is still so hard to predict weather accurately and precisely?

As you're leaving for your office, you peep outside the window and see that its not raining. You decide not to carry the raincoat as it adds extra carriage.You head to the parking, and kick-start that bike you so adore. A couple of minutes into the journey, the rain begins. You reach the office completely drenched.This was just an example from our everyday lives. I was just trying to explain how weather is a very unpredictable phenomenon.Mathematics and statistics have helped us understand the physics involved in atmospheric phenomena. This has allowed us to predict beforehand for the convenience of the society.What should be understood is that we cannot play god. Nature will always have an upper hand. No matter how much we predict, nature will always have a couple of surprises for us.That being said, the IMD hasn't been that bad lately, actually their current forecasts are really good.Their cyclone warnings have been accurate and precise in the past few years. Had the IMD not given advance warnings, storms like Phailin and Hudhud would have left behind huge loss of lives. It is the Regional Specialized Meteorological Centre for the North Indian Ocean cyclone basin, and this honor is bestowed upon by none other than the World Meteorological Organisation (WMO).The monsoon long range forecasts end up to be near-perfect, thanks to recent advancement in research and advanced computing capabilities.There is no debate for accuracy. Being a national agency the recorded temperatures/rainfall adhere to international standards. They don't just happen to be a WMO recognised agency.What the problem is:Lack of good weather coverage. India has reached a point where it can have a 24x7 weather reporting television channel, like the United States does.Lack of awareness among the common public. Weather does not matter to us unless we are affected by it.Lack of faith in the agency among the common public. This question wouldn't have been asked otherwise.

If nobody can predict what the weather will be tomorrow with any degree of certainty, what makes them think they can predict what our weather will be like in the coming centuries?

They do not say they can predict the weather in coming centuries. They say they can predict the climate, which is average weather like annual average rainfall or temperature, in coming centuries.There is considerable difference between trying to predict a very specific event, like the weather, and predicting large scale events like climate. If I toss a coin on a table, I cannot predict whether it will come up heads or tails. However, if I toss 1000 coins on table, I can predict that the number of heads facing up will be very close to 500. It may not be exactly 500, but it will be within a percent or two of 500, whereas it would be pure luck if I correctly predicted which way one coin would land. To name another example, I cannot predict with any accuracy whether a specific person will die in the next year (unless, of course, they have a fatal disease), but I can predict with considerable statistical accuracy how many people will die in the US over the course of the next year just by looking at trends in total death tolls.The weather in a specific place at a specific time is a chaotic system that depends on thousands of conditions like temperature, humidity, wind direction and speed, etc. in all the surrounding areas. It is impossible to measure all of these accurately in time to then compute the complex interactions in time to give an accurate forecast. The climate. however, is the long term average or large areas. This can be calculated by calculating the total inputs of conditions like sunlight, heat trapping gases like CO2 in the atmosphere, etc.

When someone says there is a 30% chance of snow, what is the probability of that 30% happening?

According to Nate Silver in The Signal and the Noise (chapter 4), it depends on the someone.If "someone" is the US National Weather Service, the probability of snow, given that they state at 30% confidence of snow, is about 30%. In other words, if you look at all the times they said there's a 30% chance of snow, it snowed very nearly in 30% of them. This is called being "well-calibrated". It's not the entire story behind good forecasting, because perfect forecasting is when you always know exactly what's going to happen, but being well-calibrated is one benchmark forecasters can shoot for. Incidentally, meteorologists' forecasts are subjective opinions based on a combination of computer models and their own expert human judgment.If "someone" is a commercial weather forecasting service, like the Weather Channel, it's a bit different. These services are usually not well-calibrated, and this is deliberate. Commercial services are in the business of making people trust them. If the Weather Channel predicts 10% chance of rain, then it rains, people think the forecast failed them and their trust will drop, so the Weather Channel compensates by overstating the chance of precipitation. By a chart in Silver's book, the effect is strongest for forecasts in the range of about 20%. When the Weather Channel predicts at 20% chance of rain, it only rains about 8% of the time. The Weather Channel exaggerates like this mostly for low probabilities; by 30% the effect is mostly gone. When they predict 30% chance of rain, it rains roughly 24% of the time. I imagine they treat snow and rain similarly.Finally, if "someone" is a random Joe making a prediction based on the color of the sky or the aching feeling in a trick knee, then it's anybody's guess. Most people are very poorly-calibrated on most tasks. For example, if you ask people to answer a bunch of trivia questions and state their confidence in their answers, it's very common that when they say they are 90% confident, they are only right about half the time. You can train yourself to be well-calibrated, though. This is explored in How to Measure Anything by Douglas Hubbard, and you can practice by playing The Credence Game, by CFAR - an overview.