What weather apps are you guys using? Do any do well with cloud cover predictions?

[BrettS]

So now that I’m paying more attention to clouds and cloud cover I’ve also been paying more attention to the cloud cover predictions in Dark Sky, which is the weather app that I’ve been using for years. I have discovered that as much as I like Dark Sky their cloud cover predictions are wrong way more often than they are right. They will also change vastly from day to day. For example I’ll go to bed on a Monday evening with Dark Sky showing that Tuesday, Wednesday and Thursday will be clear and sunny and then wake up with it showing that Tuesday, Wednesday and Thursday will be overcast and rainy.

I think some of this might just be the nature of the weather in Florida. I believe that because Florida is a thin little strip of land between two very large bodies of water the weather tends to be less predictable and changes quickly, but I’m also wondering if there are other apps that can do a better job with their predictions.

A few days ago I checked Dark Sky in the morning and it reported that it would be overcast all day, but the whole day was actually bright and sunny and one of my best production days in the past few weeks. I’ve had other days where I checked in the morning and Dark Sky said they would be clear, but then I barely saw the sun at all as the day went on.

 

[zƬesla]

Good ol' national weather service; check out Sky Cover below (blue line)
Hourly Weather Forecast for 28.44N 81.31W (Elev. 89 ft)

There is even an API

 

[wjgjr]

The NWS sky cover predictions for us are okay, but it does seem like something that is difficult to forecast - particularly for purposes of solar production. I also had spent some time trying to find historical data for cloudiness/sky cover (to try and determine if, for example, this September was more or less cloudy than average, to get an idea of how my solar production should compare to PVWatts estimates.) I did find a single, daily number published by NWS, but it seems to be rounded to the nearest 10% and doesn't have hourly data, so harder to assess the relative cloudiness and, importantly, when the clouds occurred during the day.

 

[Rockster]

Not trying to be contrarian, just honestly curious. I have had solar for going on two years and it never occurred to me to try to predict (or even care) what cloud cover is forecast. It's going to be what it's going to be and I don't think about it. What's the appeal?

 

[getakey]

Not trying to be contrarian, just honestly curious. I have had solar for going on two years and it never occurred to me to try to predict (or even care) what cloud cover is forecast. It's going to be what it's going to be and I don't think about it. What's the appeal?

Agree. It is what it is

 

[zƬesla]

New Not trying to be contrarian, just honestly curious. I have had solar for going on two years and it never occurred to me to try to predict (or even care) what cloud cover is forecast. It's going to be what it's going to be and I don't think about it. What's the appeal?

Agree. It is what it is

Clearly not enough of data geeks

 

[Rockster]

I do like to go back and compare my system's output with what my pre-installation estimates showed I would produce. Thus far I haven't produced quite as much as the online tools predicted but it's been pretty close.

 

[wjgjr]

I do like to go back and compare my system's output with what my pre-installation estimates showed I would produce. Thus, far I haven't produced quite as much as the online tools predicted but it's been pretty close.

For me, that is the main curiosity (which is why I was interested in history as much as forecasts.) Since I only have 2.5 months of data, if I could say that the weather has been sunnier or cloudier than the averages (which PVWatts is presumably using) that could help get me a better idea if production is as expected. Right now, I am above the Tesla estimates, but September will be decently below PVWatts (August was almost spot on.) I do think clouds are probably the answer as opposed to a system issue, but it would be nice to have data to support that.

With the forecasts, I would say that I am more likely now to notice when they are off due to looking at solar production numbers.

 

[BrettS]

Not trying to be contrarian, just honestly curious. I have had solar for going on two years and it never occurred to me to try to predict (or even care) what cloud cover is forecast. It's going to be what it's going to be and I don't think about it. What's the appeal?

I think a lot of it is because this is still new to me. My system is still a couple of months old and I’m still definitely in the micro managing stage. While I’m operating on grid you’re right, it’s kind of an it is what it is sort of situation.

However, if I do need to operate off grid for an extended period of time it is useful information to have. If I know what the cloud cover is going to be then I will know if I will have excess power available to charge my car or do a load of laundry or if I’m better off conserving as much power as I can. My system is big enough that on a sunny day I can cover my normal daily power usage with some power to spare. But on a cloudy/rainy day I can easily use more power than I can generate.

Even though it doesn’t really matter now while I’m on grid, if I spend time learning how the system behaves with and without cloud cover and I’m better able to predict how much power I’ll be able to generate on a given day then I’ll be better prepared for the time when a hurricane comes and knocks out my power for several days.

 

[getakey]

I just look at the daily production. I'm down 15% in August and 20% in Sept because of smoke

 

[woferry]

Powerwall reserve management is one reason. Most of the year I can leave my reserve at the minimum and not draw anything from the grid, because the PW won't run out before the sun comes back up to recharge it. But if the next day is going to be very cloudy it would be best to raise the reserve the night before, resulting in some grid draw but having enough left in reserve to still handle a power outage. Right now I handle this manually based on TV weather forecasts in the winter, but I'd like to be able to automate it.

Of course the past month has shown that smoke can be an equally big issue in this area, the worst days a few weeks ago were as bad for production as the worst overcast/rainy winter days I've had. And I'd imagine that's not something the NWS predicts or includes in that sky cover value.