New Framework Helps Climate Modelers Integrate Indigenous Community Input Into Simulations

Advanced computer models play a crucial role in understanding climate change. These Earth system models help scientists simulate everything from rainfall and river flow to vegetation shifts, wildfire risk, and wildlife habitats. But despite their sophistication, these models often fall short in one key area: making the results truly useful for the communities who live with the consequences of climate change every day.

A new research framework developed at the University at Buffalo aims to change that. Published in the December issue of AGU Advances, the study provides practical guidance for climate modelers on how to meaningfully integrate Indigenous community input into complex climate and environmental simulations. What makes this work stand out is that it approaches community collaboration from a modeler’s perspective, focusing on how real technical decisions inside large computer models can reflect local priorities.

Why Community Input Matters in Climate Modeling

Earth system models are powerful, but they are also built through countless decisions made by scientists behind the scenes. Choices about spatial resolution, time scales, calibration targets, and which processes to emphasize all shape the final results. Traditionally, these decisions are driven mainly by scientific norms and computational constraints, not by the needs of people on the ground.

The new framework argues that this approach limits the real-world value of climate science. Climate data is most effective when communities can act on it, whether that means managing water resources, protecting ecosystems, or planning for wildfire risk. When models ignore local concerns, their outputs may be technically impressive but practically disconnected from daily decision-making.

The research emphasizes that co-designing models with communities makes climate science more relevant, more trusted, and more usable—without sacrificing scientific rigor.

A Framework Built on Real-World Experience

The framework is based on the experiences of Yifan Cheng, assistant professor in the University at Buffalo’s Department of Earth Sciences. Before joining UB, Cheng worked as a postdoctoral fellow and project scientist at the National Center for Atmospheric Research, collaborating closely with Indigenous communities in Alaska and Northern California.

Rather than proposing abstract principles, the study draws directly from these large, complex modeling projects. The result is a realistic guide that acknowledges both the possibilities and the limitations of community-driven climate modeling.

Four Levels of Co-Design in Climate Models

At the heart of the framework are four levels of co-design that climate modelers can adopt depending on project scope, resources, and community needs. Importantly, the framework does not suggest that every project must reach the highest level.

1. Jointly configuring the model setup
   This involves working with communities at the beginning of a project to decide what the model should focus on. This could include which variables matter most, what geographic areas to prioritize, and what types of outputs would be most useful.
2. Fine-tuning models to reflect community priorities
   Here, model parameters are adjusted to better capture what communities care about most, such as seasonal water availability or fire behavior under traditional land management practices.
3. Incorporating local knowledge directly into models
   Indigenous observations, long-term environmental knowledge, and place-based insights can be used alongside conventional scientific data to improve model accuracy and relevance.
4. Developing new model functions
   In some cases, existing Earth system models simply lack the ability to represent processes that matter deeply to communities. This level involves building new components into the model to capture those missing processes.

The framework makes it clear that lower levels are not inferior. Even limited collaboration can significantly improve the usefulness of model outputs.

Large-Scale Collaboration in Alaska and Canada

One of the key case studies described in the research involved modeling climate and hydrology across Alaska, the Yukon Territory, and parts of the Northwest Territory in Canada. This region includes 229 federally recognized tribes and more than 40 First Nations, making it one of the most complex social and geographic landscapes for climate modeling.

To engage communities across such a vast area, the research team created an Indigenous Advisory Council, distributed surveys to Indigenous governments and organizations, and hosted an in-person community summit. These efforts helped identify shared concerns and priorities, such as changes in hydrology and long-term climate trends.

However, the project also faced real constraints. The model required approximately 30 million CPU hours, limiting how much it could be adjusted based on feedback. This highlighted a key reality: large-scale models can incorporate community priorities, but they often require compromises due to computational and logistical limits.

Deep Collaboration With the Karuk Tribe in California

A second project in the Mid-Klamath region of Northern California showed what deeper collaboration can look like on a smaller scale. In this case, the modeling team worked closely with the Karuk Tribe and its partner organizations.

Because the project focused on a single region and community, the model could be tailored much more directly to tribal priorities. These included watershed runoff, streamflow, salmon populations, and cultural and prescribed burning practices.

One notable moment came when tribal partners pointed out that the fire model being used could not simulate dynamic fire spread. Rather than ignoring the issue, the team adjusted course, choosing to simulate the effects of static fire scenarios instead. This decision aligned better with the tribe’s land management questions.

Another critical decision involved streamflow accuracy. When model tuning revealed a trade-off between accurately simulating summer or winter flows, it was the tribe who chose to prioritize summer streamflow, given its importance for salmon habitat and spawning.

These examples underline a central message of the framework: there is no one-size-fits-all approach to co-design.

Trust as a Foundation for Collaboration

The research also highlights the importance of long-term relationships. Several co-authors had already spent years working with Indigenous communities before these modeling projects began. This existing trust made communication smoother and collaboration more effective.

Trust is especially important when working with complex models that can feel opaque or intimidating. When communities understand how models are built—and see their priorities reflected in the results—they are more likely to use and trust the outputs.

Broader Implications for Climate Science

This framework arrives at a time when climate science is increasingly focused on adaptation and resilience, not just prediction. Indigenous communities often possess generations of place-based environmental knowledge, offering insights that complement scientific data.

By showing how this knowledge can be integrated into Earth system models, the study provides a roadmap for making climate science more inclusive and impactful. It also challenges modelers to rethink their role—not just as technical experts, but as partners working alongside the people most affected by climate change.

Looking Ahead

As climate impacts intensify, the demand for locally relevant, actionable climate information will only grow. This new framework offers a practical way forward, helping scientists build models that are not only accurate but also meaningful to the communities they aim to serve.

Research paper:
https://doi.org/10.1029/2025AV001921