The Prediction Lab & The City of Salem

Who we are

Who we are

The City of Salem Public Works Technical Group and The Prediction Lab produce weekly forecasts of harmful algal blooms in Detroit Lake, OR. We use Big Data collected from the lake, including water samples taken directly from the lake, as well as information on local and regional weather. We are also working to incorporate satellite imagery to measure changes in the color of the lake. All this information is fed into machine learning algorithms that help us:

  • make 7 day forecasts about cyanobacteria concentrations and algal toxin levels
  • estimate levels of uncertainty about our predictions
  • learn about what drives the occurrence of harmful algal blooms

See the side menu for more information about this project, especially the Forecast Archive for previous blog posts about Detroit Lake harmful algal blooms.


The City of Salem Public Works Technical Services Group is a technical and innovation center for the City of Salem’s Public Works Department. It provides operational intelligence by implementing a variety of cutting edge technologies to turn data into insight for decision makers. The group manages the department’s GIS, real-time analytics, unmanned aerial system (UAS) and flow monitoring programs.


The Prediction Lab LLC (TPL) is a technical consultancy that solves hard problems with math and computer science. Our expertise lies in finding the signal in the noise: working with messy data, and using cutting edge data science tools to make predictions about the likelihood of future events. In this project, TPL has worked with the City of Salem to develop new prediction algorithms for Harmful Algal Blooms in Detroit Lake, OR. TPL uses data collected by the City of Salem from the lake, along with other weather and satellite data to predict the level of toxins and certain algae in the lake at short, medium and long-term time horizons.

How we work together

The Technical Services Group synthesizes as much data as possible about Detroit Lake. These data are then transferred to The Prediction Lab, who use the data in their models to predict harmful algal blooms.