Models vs. Reality: Critical Omissions in SJRA Planning

11/7/26 – Computer models, if accurate, can aid in planning. But what happens if their assumptions are flawed.

On 11/5/26, some of the brightest engineers in the region trotted out a presentation that showed the impact of releasing water before floods from both Lake Conroe and Lake Houston. They used computer models to calculate how much the extra storage capacity in the lakes could reduce flooding under various conditions and assumptions.

The work was brilliant in many ways, except for several critical omissions They didn’t model:

Current conditions, i.e., they assumed new floodgates for the Lake Houston dam had already been built.
Impacts of pre-release on the area SJRA directly controls between the dams where most damage and deaths occurred.
Actual rainfall from Harvey and soil conditions before the storm. They only used half the actual rainfall where most damage and deaths occurred, and they assumed a drought before the storm. So, model inputs were vastly inaccurate.

Recommended Next Step

SJRA has three additional model runs that it can make under its current contract with the engineers. I suggest they need to examine what would happen to the West Fork watershed if no additional floodgates are installed on Lake Houston.

At the very least, that knowledge could help determine optimal operational procedures to help protect people who live downstream from the dam. Right now. Before additional gates. If additional gates ever get built. That project currently faces a funding shortfall approaching $200 million in a time of political upheaval and uncertainty. Why?

Illusion

Engineers and computer experts often become seduced by the illusion of precision that underlies computer modeling. But they and we should all realize that the illusion is just that.

Frequently, model results fail to predict reality. Case in point: Take the bar graph below. It compares actual temperature observations to predictions of climate warming trends produced by 36 different climate models between 1973 and 2022.

The U.S. Dept. of Energy produced the graph for a 2025 report called, “A Critical Review of Greenhouse Gas Emissions on U.S. Climate.” The actual “observed conditions” are shown by the tiny blue bar on the far right.

*Page 43 from U.S. Dept. of Energy’s “A Critical Review of Impacts of Greenhouse Gas Emissions on U.S. Climate*” *2025*

Computer models are useful for planning the future and predicting results, given certain assumptions. But if those assumptions are unreal, the results are worthless, potentially misleading, expensive and dangerous. In less elevated language, “Garbage in, garbage out.”

My point? The measure of a model should be its ability to predict the future accurately. Failing that, models are little more than a magic show.

What Next?

So, what should the SJRA do at this point? The engineers made a big point of saying that they wanted input from the audience on three additional scenarios to explore.

I suggest they explore the impact of pre-release on:

The one area they already control, the San Jacinto West Fork
Given: CURRENT conditions, i.e., NO additional floodgates; a flowage easement on Lake Conroe to 207 feet above mean sea level; and various release rates in storms ranging from 10 years to 500 years in magnitude.
Objective: Optimize gate operations to minimize upstream AND downstream flood damage.

My Gripe about SJRA

SJRA constantly complains about how little they control. It comes across as if they’re still fighting a PR battle to win post-Harvey lawsuits. In public meetings, they go out of their way to dramatize how much they don’t control. It makes them look powerless and defensive. Worse, it doesn’t mitigate flooding.

SJRA Pre-release meeting in Conroe — *Poster from SJRA Pre-release Meeting in Conroe. SJRA controls only the top gray area. But that affects the bottom gray and yellow areas.*

Harvey and other floods dramatized two critical gaps:

Need for Pre-Release Strategy Before More Gates. Despite computer models and gages galore, the person operating floodgates at Lake Conroe during a flood needs better knowledge about how their actions will impact lives and property.
Better Coordination with Lake Houston Operators. That person also needs to coordinate with Lake Houston Dam operators, who may be/will likely be constrained by their own lack of pre-release capacity. They need to act as one team.

Let’s plug those gaps. It’s nice to use computer models to design a safer world in the future; we need that. But until funding becomes available, let’s all live in reality. And try to make the best of it.

Let’s use those three remaining scenarios to help us NOW…before floodgates that may never be built.

Posted by Bob Rehak on 11/7/2026

3358 says since Hurricane Harvey