What is it about?
After heavy rain, urban streams often see sharp spikes in fecal bacteria such as E. coli, which can make water unsafe for recreation. These spikes are hard to study because monitoring is sparse and the highest readings are rare, so a single average misses the real risk. We built a transferable modeling and analysis workflow that combines a watershed simulation model (SWAT) with statistics designed for skewed, storm-driven data, including methods that focus on the extreme upper tail of the distribution and account for how daily values are linked over time. We applied it to the Village Creek-Lake Arlington watershed near Fort Worth, Texas, using nine years of daily simulations and local monitoring data. Rainfall was by far the strongest driver of bacteria levels. We then tested six green infrastructure scenarios, such as treating stormwater from urban areas. Their effect on the long-term average was modest, but their effect on the worst days was large: treating all urban areas at high effectiveness cut the rarest, highest concentrations by roughly 89 to 90 percent and reduced the number of days exceeding recreational thresholds by up to about 48 percent. Green infrastructure works mainly by intercepting storm runoff, so it is most powerful against extreme events but cannot fix everything on its own.
Featured Image
Photo by Marek Lumi on Unsplash
Why is it important?
Bacteria are a leading cause of recreational water impairment, yet management decisions often rely on long-term averages that mask the storm-driven extremes that actually drive exposure risk. This study shows why that matters: green infrastructure barely moves the average but sharply reduces the worst days, exactly the days when people are most at risk. By pairing watershed simulation with tail-focused metrics and dependence-aware uncertainty estimates, the workflow provides water managers with a defensible way to evaluate stormwater investments against the thresholds used by regulations and public health advisories. It is transferable to other mixed land-use watersheds with sparse bacterial monitoring, and it supports more realistic TMDL and watershed protection planning. The key practical lesson: report upper-tail and threshold-exceedance metrics, not geometric means alone, and treat green infrastructure as one part of a portfolio that also tackles chronic sources.
Perspectives
What we find useful about this study is that it reframes a question managers ask all the time: does green infrastructure actually help with bacteria? The honest answer is "it depends what you measure." If you look only at long-term averages, the benefit looks small and easy to dismiss. But the days that close a swimming area or trigger an advisory are the rare, rain-driven spikes, and that is exactly where green infrastructure delivers. Building the analysis around the upper tail, rather than the average, changed the story. We hope the openly shared model and code make it straightforward for other watersheds to ask the same question of their own data.
Gehendra Kharel
Texas Christian University
Read the Original
This page is a summary of: Watershed-scale green infrastructure reduces storm-driven Escherichia coli extremes: Tail-aware modeling and dependence-robust inference for bacterial water-quality management, Cleaner Water, September 2026, Elsevier,
DOI: 10.1016/j.clwat.2026.100283.
You can read the full text:
Resources
Contributors
The following have contributed to this page
