Spire Plans To Use Tiny Satellites For More Accurate Weather Forecasts

Zothecula writes Weather forecasting is a notoriously inexact science. According to San Francisco-based tech startup Spire, this is partially because there are currently less than 20 satellites responsible for gathering all of the world's weather data – what's more, some of the older ones are using outdated technology. Spire's solution? Establish a linked network of over 100 shoebox-sized CubeSats, that will use GPS technology to gather 100 times the amount of weather data than is currently possible. The first 20 of those satellites are scheduled to launch later this year.

Comments

I work in Earth-observing satellite ground systems

[idontgno]

The last I looked, the state of remote-sensing algorithms for limb profiling (i.e., looking through the layer of the Earth's atmosphere over the limb of the planet from your orbital position) is something between bad and "are you kidding?".

I wonder what kind of secret sauce these Young Turks have that NASA and NOAA doesn't?

A sense of scale

[RightwingNutjob]

is missing in this notion. Meaningful Earth observation from space is done with cameras that take up more physical space than a cubesat. Yeah, you can squeeze several high definition cameras into a cubesat, but the moment you realize that you need something other than visible band, temperture control on the ccds, and the power-aperture to beam that stuff down to earth in a meaningful timeframe, you've built 1500lb worth of overhead around your tiny little cubesat and you're back in GOES and NPP land.

Re:A sense of scale

[mspohr]

OMG! Quick somebody tell these people that the experts here at /. have decided that they are making a big mistake. I'm sure they will be grateful for all of the expert opinions of our hive mind since it will save them from making a BIG MISTAKE.
(OTOH, you could RTFA.)

Re: A sense of scale

[rfengr]

They are not. The combination of L1 and L2 allows for estimation of the ionospheric delay since the delay is proportional to frequency (difference between L1 and L2). It's all moot now anyway with the availability of reference stations to obtain a correction.

Re:I work in Earth-observing satellite ground syst

[ShanghaiBill]

The last I looked, the state of remote-sensing algorithms for limb profiling is something between bad and "are you kidding?".

But they are not doing much "remote sensing". All they are doing is recording when a GPS signal is received. That's it. That shouldn't be too hard. The delay between when the GPS should have been received, and was actually received, will tell them the index of refraction of the atmospheric cord it passed through, and from that, a ground computer can calculate the humidity, temperature, and pressure.

Re:I work in Earth-observing satellite ground syst

[DerekLyons]

The last I looked, the state of remote-sensing algorithms for limb profiling is something between bad and "are you kidding?".

But they are not doing much "remote sensing". All they are doing is recording when a GPS signal is received. That's it. That shouldn't be too hard. The delay between when the GPS should have been received, and was actually received, will tell them the index of refraction of the atmospheric cord it passed through, and from that, a ground computer can calculate the humidity, temperature, and pressure.

That sound you heard is the OP's point whooshing over your head. Limb profiling (what you describe) is a remote sensing technique, and it's not one that works really well.

The problem isn't receiving the signal (well, it's not a problem in this sense though it has challenges of it's own), the problem is analyzing the signal. You have three different variables (all of which vary with altitude to boot), with no way to significantly constrain any one of them - meaning arriving at an accurate value for one (let alone three) is a Very Hard and Poorly Understood problem.

Re:I work in Earth-observing satellite ground syst

[fermion]

The first question that popped into my mind was did they have a new model that would take data from 100 satellites and produce a more accurate forecast. I don't think that satellites alone are not going to create a more accurate forecast. This reminds me when I was talking to a teacher back in the 80's. He mentioned that at one point it was believed that if we could create a dense enough network of satellites and sensors, we could forecast the weather with great accuracy and for arbitrarily long periods. Theoretically, given an infinite array of sensors, the forecast would be perfect and long range. But then actual science interfered as the work of Lorentz propagated through the ranks. The sensitivity to initial conditions, and the inherent limitations of data collection, made such claims of better forecasting theoretically impossible. I have to think that the current configuration of satellites represents some compromise between cost and benefit. Not to say that more satellites will not provide a benefit. Whoever contracts with the service will be able to claim 'We have better forecasting because we have more satellites', which will help with marketing. It will help push forward the cubesat business and will test out these new technologies, which is of great benefit. And it is an experiment that might succeed in producing useful data that might be able to be put into better models.