A solar farm is not supposed to be a hydrology problem.
The panels sit on posts, the ground underneath stays in vegetation, and nothing about it looks like a parking lot.
That appearance is what makes the permitting assumption easy to accept, and the assumption is that the site behaves roughly like the field it replaced.
Stormwater rules are written on that basis. A number goes into a spreadsheet, a ditch is sized from it, and the drawing gets approved.
Three sites in one state now have instruments in those ditches, recording what actually comes down them.
The number in the spreadsheet is the thing being tested.
A flat glass sheet does not absorb a raindrop
The mechanism has nothing to do with the panel being impermeable, which everybody already knows.
It has to do with where the water goes afterward. Rain lands on tilted glass, runs down it, and leaves along the lower edge in a line.
So the ground below that edge takes the rain that fell on the whole panel, arriving concentrated and fast on a narrow strip.
That strip gets hit hard enough to seal over, losing the open structure that lets water soak in, and the water that cannot soak in runs sideways instead.
Underneath the rest of it, the soil was driven over by pile rigs and cable trenchers during construction and never recovered.
The site drains the way pavement does, in patches.
The thing in the ditch that does the counting
The instrument is a flume, which is a steel throat of known shape set into the channel so that depth converts to flow.
These are 24 inch units, paired with rain gauges logging every five minutes down to four hundredths of an inch.
The catchment feeding each one was mapped with a handheld scanner walked across the site, because the area draining to a flume is the denominator in every figure that follows.
Storms were kept for analysis from about a sixth of an inch upward, which is a lower bar than erosion work usually sets.
Each solar catchment was matched against a reference catchment next door, under corn and soybean rotation, soybean alone, or a mixture of grass and forest.
Not pasture, and not one land use.
Time of concentration and a number called CN
The three sites run from 17 to 240 megawatts, across the coastal plain and the piedmont, monitored from January of 2024, September of 2024 and February of 2025 respectively until October of 2025.
Time of concentration, meaning how long the catchment takes to deliver its peak, came out between 1.2 and 2.1 hours under panels.
Row crop ground ran 1.7 hours. Grass and forest averaged 3.5.
Runoff ratios separated further. At one site the paneled areas turned 10 to 20 percent of the rain into runoff against 2 percent next door, and at another 15 to 66 percent against 3.
Curve numbers, the figure that actually goes into a permit, came out between 76.8 and 90.7 under panels against 61.8 to 75.5 on the references.
Eight of nine paneled catchments cleared 74.
The storms that would settle it have not come
The report is unusually direct about what it cannot yet say, and the biggest gap is the size of the rain.
Not one monitored storm reached the local ten year event. The largest were about 3.2, 3.0 and 2.7 inches, which are one and two year rainfalls rather than the design case.
Several numbers also rest on three or four storms each, one site had no reference catchment at all during this period because a permit was late, and a flume in a rock lined spillway may have let some water past it, in the way that a project in the same state or a replacement crossing is judged on numbers still arriving.
This is year one of six. Final recommendations are not expected until 2029, and the authors say so in the preliminary report itself.
Nothing here is a conclusion yet.
Why a curve number is worth arguing about
All of this matters for one boring reason, which is that the curve number is not a research output. It is an input.
It sets how big the basin has to be, how wide the ditch is cut and how much of the site gets given over to holding water back.
Pick it from the wrong land cover and the infrastructure downstream of it is undersized from the day it is built, and the shortfall only shows up in the storm nobody has measured.
That is the argument for putting flumes in ditches rather than modeling the question, and it is why the coverage of the first results drew as much attention from county boards as from hydrologists.
The solar farms are not behaving like crop fields.
Whether that matters is still open.
