Aeolus and His Bag of Hot Air
(Originally posted August 21, 2016 on Blogger)
I was driving through the wide-open desert this morning, under dark skies with curtains of virga draping overhead.. I was listening to some old Cretan music and decided to swap the CD out for another disc. As I was swapping out CDs, the radio station was finishing up a weather forecast with the words, "...as we can expect temperatures to start cooling off over the next two weeks."
I popped in the other CD, continued down the long, unbending stretch of wonderfully empty highway and thought about how generally silly it is to give a forecast much beyond 6 or 7 days. And then I remembered a friend of mine who was just recently asked what the weather would be like for a wedding date MONTHS in advance!
How often have we heard folks mocking forecasters for getting it wrong, and not knowing what they're doing, or perhaps most commonly, exclaiming the faults of a so-called modern society that cannot accurately predict the weather more than a week out. I'm sure you've heard it before... "You'd think with the advances in technology that we'd at least be able to predict the weather a couple weeks out! We're living in the 21st century aren't we?!" To the latter, yes (according to Pope Gregory XIII, and yes, you too Julius). But to the former, let's set aside blind technological optimism in lieu of some reality, and by reality, I mean the reality of a world that has yet to create a true quantum computer capable of tackling some complicated equations in what would need to be in real-time, or very near it at least.
Let me explain, perhaps crummily (nod to Shakespeare with that made-up word! That's one! 1799 to go!), but I'll attempt to explain nonetheless because that's what these blog entries are all about... explaining things terribly for your amusement!
When I delve into the forecast models in meager attempts to predict where tornadoes might form, I generally have an exponentially-decreasing interest in models as the days go out from the now. Anything forecast beyond 6 days I nearly completely ignore as insignificant and, well, for all intents and purposes, meaningless. By day 5, if I see something, I keep an eye on it with cursory interest. Day 4, if whatever awesome weather pattern I saw on day 5 is still there and looking as good if not better, I become genuinely interested but reserved. Day 3, if all looks good, I start to feel some excitement. Day two, looking good? Truly excited and unable to sleep. Day 1... pure adrenaline (after my morning coffee of course)!
And though not all models are constructed the same (some use single vector, others bred vector perturbations for instance), they all are victim to that thing Lorenz realized... chaos. What he saw in fluid dynamics (mainly wind in our case) was its inherent chaotic nature. Think of the butterfly effect; a monarch flutters its wings and half way across the world it causes a hurricane. Ok, actually don't think that! That's ridiculous, BUT, consider that even minute errors in temperature, wind velocities, and humidity to name a few parameters needled throughout the complicated algorithms these geniuses compose, can result in massive errors over time.
More specifically, and according to Lorenz, these tiny errors amplify, and in the world of weather forecasting, they amplify and double every 5 days, thus making anything much beyond that nonsensical at best. 14-day forecasts, the farmer's almanac... snake oil my friends! Pure, slick, well-pitched snake oil!
Indeed, we are in the 21st century, and we have some incredible supercomputers running right now as I type this. They're all over Earth churning away. China's Tianhe-2 has a computing speed of 33.86 petaflops (PFLOPS). That's 10^15 floating point operations per second! Thirty-three quadrillion, eight-hundred, sixty trillion per second! Just to give an idea of what this number represents... A million seconds is just over 11 days. A billion seconds is over 31 years! A trillion seconds is over 31,700 years! A quadrillion seconds? Forget about it!
And yet, this Goliath of computing prowess is not fast enough to handle the equations necessary to extend forecasts, or even to compute initial uncertainty in these complex models when they're set loose. The non-linear partial differential equation necessary (Liouville's equation) to do this are so complex AND require being calculated in real-time (as we are talking about weather here), that only a quantum computer could tackle them fast enough to be useful for forecasting... again, these things need to be done in real time... and yes, technically even the act of reading these words is not truly real time... but you catch my drift! :)
So, until a quantum computer is developed (which actually is a bit scary considering the habits of societies these past 10,000 years), I doubt we'll ever see forecasts worth a darn beyond a week. At best, and what is used now, are ensemble forecasts. Because of this inherent uncertainty, this chaos, the best we can do (and by we I mean not me) is to utilize stochastic simulations to obtain probabilities of different outcomes. To modify a well-known phrase, one might be correct to say that in a chaotic world, the stochastic Monte Carlo simulation is king.
And so it is.
But fret not, for the European Space Agency (ESA) is finally (10 years late but ever-so-needed) nearing completion of Aeolus! A new satellite that will be capable of revealing a high-resolution 3-dimensional wind profile of Earth's atmosphere from the surface up to 30km! It's no quantum computer, and won't make 30-day forecasts reliable (I'm pointing at you weathertab.com!), it will improve things considerably, as well as increase our understanding of Earth's climate.
The satellite was named after Aeolus, and let me say that Aeolus, the wind bearer of Greek mythology, is as difficult to figure out as Herakles (for me anyway). Pinning Aeolus down as to who he was, is as difficult as pinning Herakles down as to WHEN he was! The Aeolus after who the satellite was named, is Aeolus, son of Hippotes. The one Odysseus received a leather sack filled with all the winds but the west one (lest Odysseus end up back on the shores where Polyphemus was no doubt still calling out to his father for revenge).
I just read an article on the BBC website yesterday (or was it the day before?) about Aeolus (the satellite), or more properly ADM-Aeolus (the ADM standing for Atmospheric Dynamics Mission). It was approved in 1999, and contracted in 2003, but its launch date is 10 years behind schedule. It will use a type of lidar, which in and of itself is not a new technology, but the ability for this laser to survive for years on end in space, the method for accomplishing that... that is new. According to the article, and I'm paraphrasing here, the laser was degraded by its own optics. Attempts to operate the laser in a vacuum proved impossible, and 'removing' the vacuum allowed contamination in the form of tiny particles. The article went on to say there are over 100 optical surfaces through and off of which the laser interacts, and those surfaces failed as particles caught in or by the laser burned and blackened optical surfaces. The end result was a laser that worked, but only for a short time, and scientists needed this system to last at least 3 years in space.
The article went on to say the issue was resolved with great effort by gently flushing the system with oxygen, though I can only imagine how much more complicated it is than that. At any rate, they have the lidar up and ready to run (a marathon as it were).
As I said, lidar isn't new, it's been used to map out all sorts of things from ice sheets to forest canopies, and is even utilized in those self-crashing, er, driving cars. Lidar is like radar, but instead of using sound to map distances (surfaces), it utilizes light. Given the 'speed of light', mapping can be done in real-time... hence the ability for a car to drive itself, if I must personify it so.
The lidar on Aeolus is the direct-detection type, and operates at a wavelength of 355 nanometers (UV). This should allow its 'light' to reflect off surfaces of dust, ice, water droplets and shiny bald heads. It's all part of one of the major components of Aeolus called Aladin.
Aladin: Atmospheric LAser Doppler INstrument is a high-tech wind probe, but not just a stationary wind probe sitting atop a tower... a global wind probe orbiting Earth and mapping out our atmosphere from 'here' to Timbuktu! Assuming no thick clouds, it can interpret distance down to Earth's surface, in clouds it can interpret their tops. Aladin has a flying carpet, no wait.. wrong Aladin... Aladin has a back-scattering receiver particular to Mie and Rayleigh scattering. Mie scattering of visible light wavelength (think bright...and dark... puffy... and non puffy clouds), and Rayleigh scattering in that shorter spectrum (think blue skies, blue because, as it turns out, scattering at 450 nanometers is about 9 times greater than at 700 nm.... or red'ish.... at equal incident intensity.
Aladin works alongside a nice 1.5 meter diameter Cassegrain telescope to process these backscattered signals to produce "line-of-sight wind-component profiles". Now this is really exciting for many reasons. For those that have read my storm chase entries, will know how heavily lowly amateur storm chasers like myself rely on the vertical wind profile when attempting to forecast severe weather. In fact, for me anyway, it's the first thing I look at. But not just likely improvements of weather forecasting will result from the launch of Aeolus... a more robust understanding of Earth's changing climate awaits as well.
The data from Aeolus will be integrated into the forecast models... which models? Well, I would think the ECMWF certainly, but I expect others as well. So I wait, as they load this 400 million Euro satellite onto a Vega rocket and launch it into space for its maiden orbit around the globe.