As a data point / yard stick / discussion point my self installed ground mounted system is a year old on Oct-18. It is 800w peak1 and has produced 551kWh so far. It is of course south facing but not an ideal location, it is partially shaded in the middle of the day in high summer and from about 14:30 year round. Ideally sited or if I got busy with a chainsaw I'd be over 650kWh/pa.
On paper my return on investment time should be 11 years but as an unregistered system I do not get paid for excess production exported so it is higher. I could register it and get paid (pitifully, about a 1/4 of unit purchase cost) but if I did that I should probably double the system to 9m2, the max you can do ground mounted without planning permission. I'm not going to do that, I'm fitting an east/west roof mounted system instead.
Cost was £1100. If you can build a fence or small shed, fit a fused spur and deal with android and web apps you have the skills. At that price it was cheap enough for me to consider it a hobby, a geek interest project. Certainly far less than the hardtail 29er n+1 I had my eye on.
Cost for a ground mount 9m2 system would be about £1800 and produce, if unshaded, about 1.3MWh/pa. Registering is free, it's just paperwork. You can then estimate your ROI off your savings in power not purchased and power sold. If it comes out over 10 years I'll be surprised.
1The panels are 800W peak but the system isn't. The IQ7A microinverters max out at ~ 710W but to count as a paperwork free system it's peak panel power you must use. Why did I fit 800W panels if I can't use it all? The panels will degrade over time, the microinverters will not. At about their mid life point, 13 years, the panels will have degraded to match the microinverters.