[ebildsten]

Satellites in LEO (including ours) travel at about 17,000 mph — drones can get nowhere close to that, which affects total coverage area possible per aircraft / spacecraft, which in turn will affect costs and operations.

To get a sense of what 17,000 mph ground speed looks like from a satellite's point of view: https://www.planet.com/pulse/la-to-vegas-in-52-seconds/

[sparkman55]

Ground speed doesn't seem like a feature to me - after all, the ideal is an unblinking 'eye in the sky' that provides high-resolution video for large regions a la ARGUS. A stationary observer seems like it would provide more valuable information than an orbiting one.

With orbiting satellites, most of the time the birds are over the ocean and empty stretches of land that most people don't care about. You're wasting most of your bandwidth and storage on places that someone might be interested in one day...

Don't get me wrong, the work you're doing is very cool, and global data sets are really fun! Unfortunately, I've found that the 'value density' of these global data sets isn't great.

[branchan]

You cannot keep a stationary drone in the sky for long stretches of time. Their field of view is also going to be massively smaller than a satellite. You would need a lot of them just to keep the same amount of coverage.

How do you know nobody is interested in the 'ocean and empty stretches of land'? It could provide a lot of meteorological and geological data for scientists for example.

[sparkman55]

Funny you should mention meteorological use cases. At a previous startup, we built global historical weather data sets (at 5km resolution, 30+ year hourly time series), accessible via a metered API. This took HPC, storage, and engineering investment about an order of magnitude less than Planet. We also had a whole room of tapes filled with satellite imagery from the world's governmental Met offices...

Our main use case was for wind and solar renewable energy, but we also entertained other uses, like architecture and agriculture. These data sets turned out to be difficult to monetize; while the continental US data set may have broken even, the rest of the globe never recouped the cost of storage, let alone the supercomputer time. It's not that people around the world weren't interested in the data, they just couldn't justify paying for it.