PJM's published solar forecast against what the fleet actually metered, spanning yesterday and the day ahead — two days of horizon on the fastest-growing thing on the system.
How much solar power will PJM get?
PJM solar output peaks at 11,080 MW on Fri, 12 PM, against a low of 0 MW — a swing of 11,080 MW across the window. Solar forecast is the heaviest area at 11,080 MW.
Every line above, and who produced it.
What settled — the realized values every forecast here is scored against.
The grid operator's own published forecast for these hours. Source: PJM.
Every settled hour scored against what each forecast said beforehand.
The window peaked at 12,311 MW on Thu, 10 AM; PJM solar forecast called it 10,469 MW — 1,842 MW low.
Each day of the window on its own.
Every hour of the window, shaded by load.
| 0 | 6 | 12 | 18 | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Thu, 8/27 |
| Series | Peak (MW) | Peak hour | Low (MW) | Average (MW) |
|---|---|---|---|---|
| Solar forecast | 11,080 | Fri, 12 PM | 0 | 4,422 |
| Solar settled (recent) | 12,311 | Thu, 11 AM | 60 | 4,796 |
| Forecast | Avg miss (MW) | Avg miss % | Bias | Hours |
|---|---|---|---|---|
| PJM solar forecast | 575 | 38.0% | −535 | 21 |
The latest published vintage of each forecast, against what settled. Smaller is better on every column. Bias is signed: positive means the forecast ran high.
| Forecast | Produced by | Method | Horizon |
|---|---|---|---|
| Solar forecast | PJM | PJM Data Miner 2 — Hourly Solar Power Forecast (utility-scale), against metered solar output from Generation by Fuel Type | About 48 hours ahead, hourly |
PJM publishes a point forecast with no confidence interval, and no per-zone breakout — this is an RTO-wide number. Convective cloud cover is the dominant error source. PJM retains only 30 days of vintages, so the forecast record here does not run deeper than that. The behind-the-meter forecast PJM publishes alongside this one is excluded: it has no metered actual to score against.
PJM publishes an hourly solar power forecast for the whole RTO, re-evaluated each hour and running about forty-eight hours forward. This page plots the most recent vintage against metered solar output from PJM's own generation-by-fuel feed, with the recent hours scored against what actually delivered.
PJM's metered solar peak has gone from 1,390 MW in 2019 to 13,698 MW in 2026 — close to ten-fold in seven years, and still climbing year over year. Through the summer of 2026 solar has averaged roughly twice PJM's wind output and peaked above 14,000 MW, which makes it the market's largest renewable by both measures and the reason PJM afternoons no longer price the way they did.
The line plotted here is utility-scale solar — the metered generation PJM dispatches against. PJM separately forecasts behind-the-meter solar, the rooftop output aggregated through Load Serving Entities that it never meters and that shows up on the system as demand that simply does not appear. That second number is real and growing, but it has no published actual to be scored against, so it is not added in here.
The scorecard below sizes forecast against metered output over the overlapping hours. Convective cloud — the pop-up summer storm — is the dominant error source across the Ohio Valley and Mid-Atlantic, and a storm line crossing the fleet can halve an afternoon against a clear-sky call.
curl "https://askthegrid.com/api/v1/grid/radar/curves?iso=PJM&horizon=extended"
Connect your agent to mcp.askthegrid.com and this forecast is a tool call away — the same host Claude and ChatGPT connectors already use.
Enough to meter a 13,698 MW peak in 2026, against 1,390 MW in 2019 — close to ten-fold growth in seven years. Through the summer of 2026 it has run at roughly twice PJM's wind output.
About forty-eight hours, hourly, for the RTO as a whole, re-evaluated every hour. PJM publishes a separate behind-the-meter forecast over the same horizon.
No. This page plots PJM's utility-scale forecast against metered utility-scale output, so the two sides measure the same thing. PJM's behind-the-meter forecast covers rooftop and other output it never meters — it reduces apparent demand rather than adding generation, and there is no metered actual to score it against.
Plotted window Aug 27, 6:00 AM EDT → Aug 29, 3:00 AM EDT. Re-evaluated hourly; the latest vintage available is shown.
| Fri, 8/28 |
|---|
| Sat, 8/29 |
|---|
Darkest cells are the tightest hours. The window peaks at 11,080 MW on Fri, 12 PM, local to the market.