In 2024 I wrote a paper called “Exploring Water Quality of Georgetown, Texas.” One section described the confluence of the two forks of the San Gabriel River in Georgetown. Heavy rain fell on April 9, 2024. The North Fork then ran clear with a green tint. The South Fork ran murky and brown. The two flows met in a sharp line.

A concrete quarry borders the South Fork upstream of the city. I pointed at that quarry as the source of the sediment.

The body of the paper hedged. It said that the quarry may add to the sediment load of the South Fork during heavy rain. The caption under the photograph dropped the hedge. It stated that the sediment came from the South Fork. It named the quarry in the same sentence.

A reader takes the claim from the caption. I now correct that caption.

The paper gave no measurement for that claim. My sample data covered one site near my house. It ran from September 2023 to April 2024. It recorded dissolved oxygen and conductivity. None of those numbers compared the two forks. The evidence for the quarry was one photograph and a story that sounded correct.

A dam explains the same photograph with no reference to the quarry. I did not test that explanation. That was a mistake. The flood of July 5, 2025 showed me the error.

What happened on July 5, 2025

Heavy rain fell across the region on the night of July 4 and the morning of July 5. The National Weather Service recorded 10 to 15 inches across parts of Burnet, Travis, and Williamson counties. Emergency calls about water over the roads started in the Liberty Hill area before 2 a.m. (KXAN, 2025). Liberty Hill sits on the South Fork.

The South Fork rose fast. The gage in Georgetown has a flood stage of 9 feet. The National Weather Service forecast a crest near 29.7 feet in the middle of the day on July 5 (Hello Georgetown, 2025).

The water destroyed sections of the pedestrian crossing near the confluence in San Gabriel Park. It toppled the limestone blocks of a retaining wall. It also damaged the rails at Blue Hole Park (KXAN, 2025). The mayor signed a disaster declaration the same day.

The North Fork behaved in a different way. Lake Georgetown sits about 3.5 miles above the confluence. The reservoir held the storm runoff of the North Fork basin. The Corps of Engineers started a limited release on July 7. On July 8 the lake stood 14 feet above its conservation pool. The Corps planned to lower the lake about one foot each day for about two weeks (Community Impact, 2025).

The order of events is the important part. The South Fork crested on July 5. The dam released water on July 7. The two peaks never arrived at the confluence together.

The structure I ignored

My paper read the confluence as though the two forks were a matched pair. The two basins differ in size. They also differ in control. The numbers make that clear.

The North Fork drains about 246 square miles above North San Gabriel Dam. Lake Georgetown began to impound water on March 3, 1980. The conservation pool sits at elevation 791 feet. The top of the flood pool sits at elevation 834 feet. At that level the lake holds 130,800 acre-feet. The Texas Water Development Board (TWDB) records these numbers (TWDB, n.d.).

The flood pool is therefore 43 vertical feet deep. It holds about 93,000 acre-feet above the conservation pool. The Corps operates that space under a published water control manual (U.S. Army Corps of Engineers, n.d.).

The South Fork drains about 133 square miles above the gage in Georgetown (USGS, n.d.). It has no reservoir. Rain that falls on the South Fork basin reaches the confluence within hours.

Any comparison of the two forks at the confluence compares a regulated river against an unregulated one. On July 5 the lake rose 14 feet into a flood pool that is 43 feet deep. The reservoir took most of the North Fork peak and kept it.

Why the dam changes the sediment

A reservoir works as a settling basin. The channel widens into still water. The flow velocity drops. The suspended load falls to the bed. Dams trap coarse material almost completely. They also trap a large share of the fine material.

The water that leaves a dam often runs clear for this reason. Engineers call the result a clear-water release (Kondolf et al., 2014). TWDB also surveys Lake Georgetown and measures the sediment on the lake bed (TWDB, 2017).

So the North Fork reaches the confluence after 3.5 miles of travel below a settling basin. The South Fork reaches the confluence after 133 square miles of direct runoff. A visible difference in sediment at that confluence is the expected result of the dam alone. My paper treated that expected result as evidence of an upstream pollution source.

The colors in my own photograph agree with the dam. Clear water with a green tint is the usual appearance of a reservoir release. Still water grows plankton. The release carries that plankton downstream. Brown water is the usual appearance of direct storm runoff.

I also took the photograph during heavy rain. Heavy rain is the exact condition where the dam holds the most runoff. I chose the moment of the largest dam effect. Then I read the result as evidence about a quarry.

What the flood showed downstream

Below the confluence the combined river still flooded. The damage varied with elevation. I watched neighborhoods that sit about 40 feet above the river stay dry. The flood put water in the homes lower on the bank. That observation is mine. I did not measure it.

The operation record supports a modest claim. The dam held the North Fork peak on July 5. It passed that peak two days later as a controlled release. A crest downstream is the sum of the water that arrives from both forks at the same time. The dam removed one of those two flows on the day of the flood. That removal lowered the crest below the confluence.

I cannot say how much lower. That answer needs the reservoir inflow record for July 4 and July 5. It also needs a routing model for the reach below the confluence. I have neither one. I am confident about the direction of the effect. I cannot state its size.

The strongest objection

The rain did not fall equally across the region. More rain may fall on the South Fork basin than on the North Fork basin in a single storm. In that case, part of the contrast on July 5 shows the storm rather than the dam. Reported totals differed greatly over short distances during this event. A fair test needs the basin-average rainfall for each fork. I do not have those numbers.

The flood agrees with the dam explanation. It does not prove it.

The same caution applies to the sediment claim. One observation at a confluence cannot separate the quarry from basin geology, land use, and channel erosion. The South Fork basin has its own soils and its own pattern of development. My correction replaces one untested cause with a second untested cause. I want to be plain about that.

One more gap sits in the middle of this argument. I did not record the color of the North Fork during the July 2025 flood. I did not measure its turbidity either. That reading is the closest match to the April 2024 photograph. The 2025 flood gives me the timing of the two peaks. It does not give me a second observation of the color.

What a real test looks like

The design is simple. I did not follow it in 2024.

• Sample the South Fork above the quarry discharge.
• Sample the South Fork below the quarry discharge on the same day.
• Sample the North Fork above the lake.
• Sample the North Fork below the dam on the same day.
• Repeat each pair at base flow.
• Repeat each pair during a storm.
• Record turbidity continuously in place of single grab samples.
• Compare the basin-average rainfall of the two forks for each event.

The two explanations predict different results. If the quarry drives the sediment, the paired samples above and below the discharge will differ. The North Fork will then change little across the dam. If the dam drives the difference at the confluence, the North Fork will change sharply across the dam. The South Fork will then stay similar along its length.

Why I published the correction

The paper is still on my academic papers page. The page now carries a note about this problem. I left the paper in place because the record of a wrong conclusion is useful. The observation was real. The step from that observation to a cause was weak.

This gives two lessons. Before you name a pollution source, account for the structures that control the flow. Then write the caption to the same standard as the body. A hedge in only one of the two does no work.

A 165-foot dam sits about 3.5 miles above the confluence. It is the largest single feature in this watershed. My paper did not account for it.