When you search your address on Solrova, the number that comes back, your roof's estimated annual kWh production, anchors every dollar figure we show you. That number isn't a guess. It comes from satellite roof imagery and an NLR production model, run on documented assumptions we publish on this page. Here is how the calculation works, where it falls short, and how to read the result before you talk to an installer.
How the satellite roof analysis works
We start with Google Solar API, which turns aerial imagery of your home into a usable 3D model of the roof. Google has trained machine learning models on satellite imagery of nearly every U.S. roof, and the API returns four pieces of information we care about.
First, it identifies the individual roof planes on your home, since most houses have several slopes facing different directions. Second, it estimates the usable area of each plane after subtracting skylights, vents, HVAC equipment, and roof edges that can't hold panels. Third, it estimates annual sun hours on each plane, factoring in your region's typical cloud cover. Fourth, it maps shading from nearby trees and buildings across the day and across the seasons.
The output is essentially a high-resolution survey of your roof's solar exposure built entirely from satellite imagery. No one has to climb a ladder, and the analysis runs on every Solrova estimate. The trade-off is what satellite can't see: interior structural details, very recent tree growth, or obstacles tucked behind chimneys. That is why an installer's site visit is still the final check. Google's API also flags roofs the model isn't confident about, typically very small homes, complex multi-tier roofs, or addresses where the satellite tile is old. Those are the addresses where the on-site visit matters most.
How sunlight becomes a kWh estimate
Knowing your roof gets sun isn't the same as knowing how many kWh you will produce. Google's model estimates the yearly energy each panel position on your roof would make. We pair it with PVWatts V8, NLR's reference model for residential solar systems. NLR is the Department of Energy lab in Golden, Colorado, and it was called the National Renewable Energy Laboratory until December 2025. PVWatts is a common baseline because its methods are public and documented, and anyone can run it.
PVWatts takes the solar radiation data for your location, runs it through a simplified photovoltaic system, and accounts for the real losses every system experiences. Hot panels are slightly less efficient than cool ones. Dust and pollen reduce output. Wiring losses, inverter conversion losses, and small panel-to-panel mismatch each take a few percent off the theoretical maximum. Combined, these losses bring a modern residential system to roughly 80 to 85 percent of its theoretical output.
Google's per-panel sunlight data sets the annual kWh for the panels in your design, after a 15% allowance for system losses such as wiring and inverter conversion. PVWatts, run with your roof's actual tilt and direction, sets how that total spreads across the twelve months at your location. When Google has no roof data for an address, the estimate falls back to regional sun-hour data for your state. For a typical 10 to 12 kW system in the continental United States, the model lands between 12,000 and 18,000 kWh per year depending on local sunlight. Cloudier markets like the Pacific Northwest produce closer to 10,000 kWh. Sunnier markets like Arizona or southern California push past 16,000 kWh.
The assumptions we publish and why we picked them
Every solar estimate sits on top of assumptions, and we publish ours so you can sanity-check them against your situation. Panel degradation runs at 0.5% per year, a rate supported by NLR field studies and standard manufacturer warranties. A panel making 100 kWh in year one makes about 99.5 kWh in year two, and so on for the warrantied 25-year life.
System cost uses your state's cash purchase price per watt, before loan dealer fees, from Lawrence Berkeley National Laboratory's Tracking the Sun (2024 Edition, host-owned residential systems, about the 25th percentile). States missing from that dataset, Nevada among them, use the national figure of about $3.32 per watt. The median of all reported prices in the same data, including financed systems, is about $4.18 per watt. Prices also vary widely by installer: Berkeley Lab's 2026 Data Update found 2025 median prices across the 100 largest residential installers ranging from about $2.40 to $6.30 per watt, with loan-financed systems priced well above cash purchases. The same update found host-owned home prices fell about $0.50 per watt in 2025, so current quotes can come in below our figure. The quote you receive replaces this assumption with a binding number for your address.
We do not model loan payments or APRs. Financing terms, including any dealer fee, come from the lender and the company selling the system, so the Studio shows the cash price and leaves payment math to the actual offer. Utility rate escalation defaults to 2.6% per year, the national long-term trend in U.S. Energy Information Administration residential rate data (Form 861), and you can switch to your own state's 20-year trend instead. If your local utility raises rates faster, the dollar value of your kWh comes out higher than our estimate. If it raises them more slowly, the value comes out lower. The 25-year savings figure is the gross value of the electricity your panels make at that escalating rate, and what you pay for the system comes out of that value. Lease and PPA rates appear as regional ranges compiled from provider filings, labeled as marketplace estimates rather than quotes, because no government dataset of those rates exists.
One assumption we don't carry: the 30% federal solar tax credit. Section 25D, which covered cash and loan purchases, expired December 31, 2025, so we model owned systems without it. Section 48E, the credit that applies to leases and PPAs, can still be claimed by the company that owns the system. For solar projects whose construction begins after July 4, 2026, current law generally requires the system to be placed in service by December 31, 2027, and other eligibility rules apply. A lease or PPA provider can price part of that benefit into your rate, rather than hand you a credit to claim. The Solar Design Studio reflects both rules in its cost panel.
Where the estimate is strongest, and where it leans
Solrova's estimate is most useful at the annual scale. Month-to-month output swings with the weather, and the annual total smooths those swings out. We have not yet compared our estimates with metered output from installed systems, so we do not claim an accuracy percentage, and you should read the number as a model result rather than a measurement.
Some assumptions lean cautious and one does not. The 0.5% degradation rate is the standard figure, not an optimistic one. The 2.6% utility escalation is the long-term national trend, below the faster increases some markets have seen in recent years, so real savings can beat our estimate where rates climb fast. The cash price leans the other way. It reflects a purchase with no dealer fees at about the 25th percentile of reported prices, which keeps our payback on the short side of what a financed buyer will see.
What satellite analysis can't see
Remote analysis has hard limits, and an installer's site visit is how those gaps get closed. Satellite imagery can be months or years old, so a newly planted tree or a deck added since the last flyover won't appear in our model. Shading estimates assume average tree height for the season. If your yard has unusually mature canopy, real production can run noticeably below the estimate.
We also can't see structural details that affect what fits. A second-story dormer, a chimney shadow, or unusual roof pitch can reduce the panel area Google flagged as usable. Microclimates aren't captured either: a home in a coastal fog belt produces less than the regional weather data suggests, even if a home two miles inland does fine.
For all these reasons, we treat the Solrova number as a screening estimate. It can tell you whether solar deserves a serious look. The installer's on-site design and proposal is the document that becomes binding. If its production figure differs noticeably from ours, ask the installer to explain the difference before you sign anything.
What this means in practice: Solrova's job is to give you a credible enough estimate that you know whether to keep going. If our number says solar pays back in 8 years on your roof, that's a strong signal worth pursuing. If it says the payback is 19 years, you may still want an installer's perspective. Local incentives or your utility's particular rate structure can shift the math more than a satellite model can pick up.