Research

Forecasting Glossary

Before we dive in, let’s get acquainted with some terms.

Ecosystem Forecasting

An ecosystem forecast is a prediction of future environmental conditions that includes quantified uncertainty

An ecological model is a mathematical representation of an ecological system (ranging in scale from an individual population, to an ecological community, or even an entire biome). We develop models to better understand the real, complex ecological systems.

Digital Twin

The goal of ecological models are to create “digital twins” - representations of nature that we can manipulate in ways that wouldn’t be possible in the real ecological system. This lets us understand future conditions given warming scenarios, extreme events, and more.

Uncertainty

How much confidence do we have in a forecast? For example, how likely is an algae bloom in reservoir over the next two weeks expresses our confidence in a forecast.

Right now, environmental decision-makers are operating with incomplete information.

We can no longer use historical baselines to predict what tomorrow’s environmental conditions will be. This means that ecological forecasting is more important than ever! We focus on forecasts for critical ecological services: e.g., drinking water quality, carbon stocks, and more.

Research Themes

While we work across ecosystems and services, our forecasts fall into 5 major themes:

Ecosystem Science

At our core we define ourselves as ecosystem scientists and advancing fundamental ecosystem knowledge is a major goal of our work.

Models Meet Data

We are passionate about using cutting-edge statistics and computation to allow the in silico world and real world to learn from each other.

Open Science

We develop methods, software, and tools that are reusable, open, and reproducible. Our goal is to lower the barrier for others to forecast using our infrastructure.

Uncertainty

We embrace the inherent uncertainty in ecological systems and create new approaches for quantifying uncertainty in our forecasts, data, and models.

Translational Ecology

We develop forecasts that are actionable and co-produced with managers and decision-makers.

Water Quality Forecasting

Water quality forecasting To date, we have deployed our FLARE (Forecasting Lake And Reservoir Ecosystems) system on nine lakes across the U.S. and have been able to successfully forecast water temperature, dissolved oxygen, methane emissions, and algal blooms in drinking water supply lakes and reservoirs. FLARE is foundational to multiple Center projects that include the Virginia Reservoirs Long-term Research in Environmental Biology, Rules of Life Algal Bloom Forecasting, and Global Center for Freshwater Forecasting projects. This work has provided fundamental knowledge advancing ecosystem science, as well as critical decision support tools for managers.

More About FLARE

imagery of circuit board superimposed over leaves

NEON Ecological Forecasting Challenge

We are galvanizing the ecological research community to predict NEON data before it is collected to discover patterns of predictability across ecosystem systems and scales. As the lead of the Challenge, we have developed the common framework, cyberinfrastructure, and training associated with a massive 5-year forecasting competition. To date, we have received >1000 submitted forecasts for tick abundance, beetle diversity, lake and stream water quality, forest phenology, and terrestrial carbon storage for sites across the US. In organizing the Challenge, we are empowering ecologists across the US to generate forecasts and lowering the barriers to doing so.

More About This Project

closeup of electronic blueprint superimposed over yellow leaves

Macrosystems EDDIE: the first undergraduate curriculum in ecological forecasting

Our interdisciplinary team has developed flexible classroom modules that introduce undergraduate students to the core concepts of ecological forecasting while teaching critical skills in data analysis, modeling, and visualization. Each module utilizes long-term, high-frequency, and sensor-based datasets from diverse sources, including the Global Lakes Ecological Observatory Network, the United States Geological Survey, the Long Term Ecological Research Network, and the National Ecological Observatory Network. The modules are part of the Macrosystems EDDIE (Environmental Data-Driven Inquiry & Education) project. We are studying how best to teach ecological forecasting similar to how we make them: iteratively, with lots of feedback and data!

Explore Training

imagery of clouds superimposed over tech

QUALITY DATA. ITERATIVE MODELS. ACTIONABLE TOOLS.