How a Chart Solar forecast gets its numbers.
Every number on a Chart Solar forecast traces back to a model, a source, and an assumption you can challenge. This page is the index; the deep-dive lives in Field Notes.
Hourly physics, not annual averages
We simulate your roof one hour at a time — all 8,760 hours a year, every year for 25 — on satellite sunlight records for your address (NSRDB in the US, PVGIS in Europe, Open-Meteo elsewhere). The physics is the same family NREL uses in PVWatts, run through the open-source pvlib engine. Satellite records and real rooftops disagree a little, so we measure how much from published validation studies (cited on your forecast) and fold that margin into the band.
A measured range on every assumption
Your installer prices a forecast at 'utility raises rates 4%/yr forever.' We don't. Every uncertain input carries a measured range: how far your utility's rate rises have wandered (its own filed history over 3-to-25-year lookbacks, widened to county, then state, then the national series only when one utility's record is thin), how fast panels fade (NREL's degradation compendium), how far 'typical' weather misses real weather (published validation studies), and how well we know your usage. We nudge each input, watch how far the 25-year cashflow moves, and combine the wobbles the way independent errors combine (for the initiated: a finite-difference Jacobian with sigma-squared quadrature). The output is the P10–P90 band: the run of outcomes from the pessimistic tenth percentile to the optimistic ninetieth.
Your money's other options are in the model
Solar has to beat the boring alternatives. The landing chart draws it against a high-yield savings account; the paid workshop adds paying down your mortgage and the S&P 500, at fixed long-run rates. The question the model answers: would this money do better on your roof or somewhere else?
We document what we don't know
The battery model follows simple rules at launch (charge at midday, discharge into the evening peak); a true optimiser lands later. Without hourly usage data from your utility (Green Button), we assume a typical household's daily pattern. The federal residential credit isn't modeled for systems placed in service after 2025. Each report calls these out by name (KNOWN / PARTIAL / UNKNOWN), with a confidence band on the headline.
Export-policy compensation regimes — net metering, net billing, avoided-cost — are sourced from DSIRE (CC-BY-SA 4.0), attributed on each forecast beside the resolved policy; the full source list is at /credits. Per-source attribution (NSRDB year, pvlib version, tariff dataset, climatology window) is coming; until then, this overview plus the Field Notes archive is the canonical reference.