Sidenote, feedback on https://solcast.com/rooftop-solar/free-pv-system-performance... from a non-specialist:
I know the tool targets existing solar users, but as a potential user, it would help me answer the question "Should I get a 5kw solar panel?". For the same input on the left, I'd like to see historic monthly data for the past 3 years, telling me how much energy my potential panels would generate per year. Also, would be useful to include some info on how much of that energy would actually used when passing through batteries and all the system. If some panels 100kwh in a month, that doesn't mean I can simply subtract 100kwh from my energy bill, but how much could I subtract?
Appreciate the feedback and likely we'll have more tooling is this space coming out over the next 12-24 months, and the data for others to build these tools sooner than that. We need to get the data correct and easily accessible first and then the hope is we won't be the only ones building the tools on top and others like yourself can see gaps in what is useful and map that to data we have.
Specifically regarding "historic monthly data for the past 3 years..", you can tackle it multiple ways, but likely the most scalable for us is monthly averages of a "typical" year for a specific location rather than last X years we can look back on 10+ years to get a better idea "typical" and synthesize it down to monthly.
Combining cost comparison, batteries and individual energy consumption is another level of complexity again and not something that can give accurate answers to without knowing highly specific details. How much you could subtract you could take a guess at by looking at your electricity bills and having a guess that >50% of your consumption is probably during the day when the sun is out, but again a couple of heavy usage days when it is cloudy will throw off your averages. There are lots of websites about the economics of getting solar/battery, so while where I work will produce good data to help with those calculations, it will always be individual behavior that determines about worth while the purchase is.
As a rule of thumb (in Australia), it is around a 7 year payback at the moment, and batteries will not save you money at this stage.
It's not a hard problem, it's just tedious.
Working with live satellite imagery has a lot of challenges, and pulling clean data out of them consistently. Each satellite is different and has its own biases, operational issues that can be tricky but working on these problems I think is a really useful skill so not time wasted IMO.
Best of luck!
What could you use help with? If a software engineer could invest 2 weeks in producing something useful for your company (including adding a feature to an existing open source project), what would be a list of potential quick projects?