
Renewables are no longer a niche choice. Across Europe, more and more of our electricity comes from the sun and the wind, and Denmark is one of the clearest examples: wind supplies more than half of its power, and solar is growing fast.
That shift changes what information is worth. When power comes from nature, output depends on the weather, and the weather changes by the hour. A cloud can cut a solar park's output in minutes. Anyone who produces, buys or balances that power needs to know how much will arrive, and how far to trust the estimate.
That is true at every scale. A household with a rooftop system wants to know whether the battery will be full by evening. A developer needs a credible yield estimate before a bank will lend. A utility has to balance the grid against what the weather actually delivers. The more complete the information, the better those decisions get.
To show what that looks like in practice, we took the Meteosource renewables API to western Denmark for the working week of the summer solstice and compared its output with the solar generation the grid reported.
What the renewables API gives you

The solar endpoint returns an estimate of power output for a specific PV system, together with the weather that drives it, for any location, for dates up to 3 years back, extendable to 20 years with the historical data add-on, and forecasts for up to five days. You describe the system and the API does the rest.
Your system, not a generic one. You set the module and inverter power in kW, the tilt of the panels, and their orientation. A flat panel and a steep south-facing one at the same site produce very different curves, and the API accounts for that.
Production. The estimated electric output of your system, step by step through the day. This is the number you plan, trade or settle against.
Three kinds of sunlight. GHI (global horizontal irradiance) is all the sunlight reaching a flat surface. DNI (direct normal irradiance) is the direct beam from the sun, which clouds cut sharply. DHI (diffuse horizontal irradiance) is light scattered by the sky and clouds, which makes up most of what reaches the ground on an overcast day. Seeing all three tells you not just how much sun there is, but what kind.
Cloud cover by layer. Total, low, middle and high cloud, as percentages. Thick low cloud blocks far more light than thin high cloud, so the split matters.
Supporting weather. Temperature (panels lose efficiency as they heat up), humidity, dew point, pressure and snowfall.
15-minute resolution. The PRO version delivers all of this in 15-minute steps, matching the resolution used on European power markets.
A companion wind endpoint does the same for wind farms, with turbine model, hub height, terrain roughness and capacity as inputs. We focused on solar for this test.
The benchmark
• Location: Esbjerg, on Denmark's west coast (55.4703 N, 8.4520 E), inside the DK1 grid area.
• Dates: Monday 22 to Friday 26 June 2026, the working week of the summer solstice, including Sankt Hans on 23 June.
• Our side: the renewables API at 15-minute resolution, for a 1 kW south-facing system at 30° tilt, run for each of the five days.
• Grid side: solar generation for DK1 as published on the ENTSO-E Transparency Platform, also at 15-minute resolution.
The two are on very different scales: one rooftop-sized system against a whole region's solar fleet, which peaked at about 3,100 MW on 25 June. So we compared the shape of each day, scaling both curves so that each day's peak equals 1. The question is simple: does the API's daily solar curve rise, peak and fall the way the grid's did?
Results

Across the five days, the API's curve followed the grid's closely, with an average correlation of 0.99. The average difference between the two curves was about 5% of the daily peak.
|
Day |
Correlation |
Average difference (% of daily peak) |
|
Mon 22 June |
0.99 |
3.7% |
|
Tue 23 June |
0.99 |
4.7% |
|
Wed 24 June |
0.98 |
6.6% |
|
Thu 25 June |
0.98 |
7.5% |
|
Fri 26 June |
0.99 |
4.6% |
The closest days were 22 and 26 June, when the two curves stayed within a few percent of each other from sunrise to sunset. The largest difference came on 25 June.
Most of the difference sits in the late afternoon. On 23, 24 and 25 June, the API's curve fell away earlier than the grid's between about 15:00 and 19:00. Mornings and midday matched closely on most days.
How to read these numbers
These are not errors, they are differences. Neither curve is a known-true value. The grid figure is not a perfect measurement. Energinet, the Danish grid operator, receives real-time data from a few thousand larger solar plants, but small rooftop systems do not report, so the national total is partly estimated. Part of any gap can sit in the reference itself.
There are other limits worth stating. We compared one site with a whole region's fleet. We used a single generic setup, a south-facing system at 30° tilt, while the real fleet mixes many tilts and orientations. East- and west-facing panels shift output earlier or later in the day, which can influence the morning and afternoon differences. We compared shapes, not megawatts. The API was run for past dates, so this shows how its model reproduces a week that already happened, not how a forecast issued in advance performed. And it is one working week, in the sunniest part of the year.
Within those limits, the result is clear: for a single, configurable system, the renewables API reproduced the daily solar curve of an entire grid region closely, through clear days and cloudy ones.
Try it on your own system
Want to see the curve for your own site first? Try the solar forecast for your PV system: enter your location, system size, tilt and orientation, and see the estimated production in seconds.
Ready to build it into your workflow? The renewables API, including the 15-minute PRO version, is available on our paid plans. See pricing and sign up to get your API key.
Whether you are sizing a rooftop installation, planning a battery or managing a portfolio, you can test it on your own site and your own dates.











