Something I was pondering and tinkering with briefly just before leaving for Tioman was how to gather weather imagery (e.g., weather radar, surface wind predictions, etc) while underway, in a way that’s reasonably bandwidth friendly since bandwidth is almost always at a premium when underway.
Some people might point to GRIB files as the solution to these problems, and we did use them during our trip, but they are far from accurate and they certainly failed to predict many of the conditions we found ourselves in. I’d say they’re a good supplement to sensor data like radar imagery, and finer-grained predictions from local meteorologists of stuff like surface winds, wave height, and so on.
I had some ideas for compression of meteorological imagery as typically generated by government agencies like NEA, NOAA, etc. This is an obscure enough problem that I wasn’t able to find all that much existing work of interest, although I’d certainly be glad to hear of any. My approach relies on the fact that most of the pixels in a weather imagery series are pretty static (e.g., the background map) and not related to the weather data itself at all. If you download this background data when you’ve got the bandwidth, and keep it around for when you don’t, then you can subtract it from the weather imagery to cut down the size pretty nicely. I was able to compress surface wind prediction images from NEA to about 1/8 their original size.
Anyway, I wrote up some details of the approach in the boatlogger wiki; probably I’ll attempt to formalize the idea a bit more and come up with a more robust implementation soon.