Calibration Competition.

Your professor has built a regional aquifer with a known truth, then handed you only a handful of water-level measurements. Reverse-engineer the aquifer.

✎ Adaptable template⏱ 1–2 class sessions👤 Individual🏆 Scored & ranked📈 Intermediate

The objective

Recover the three parameters that govern regional flow — representative hydraulic conductivity, recharge, and streambed leakance — from sparse data.

What you learn

Inverse reasoning from residuals to physics, the non-uniqueness of calibration, and why a parsimonious three-parameter model is the right abstraction for a regional question.

How it works

Receive the sparse observation data and the model geometry (river, lake, boundaries, pumping).
Build your model and choose the three representative parameters.
Compare simulated heads against the observations — reason from the residuals, not by guessing.
Refine until residuals are minimized, then submit your parameter set.

How you win

Submissions are ranked by how well they reproduce the withheld truth — not just the sampled points, but the full head and streamflow fields.

Score = rank by residual error against the withheld truth (heads + streamflow)

Constraints

Only the sampled observations are given
Three representative parameters only — no over-fitting heterogeneity
Fixed geometry and boundary conditions

Deliverable

Your three calibrated parameters
A residual analysis
A note on what the data could — and couldn't — constrain

Make it yours — for instructors

Every number here is a starting point. Change the site, the budget, the deadline, or the scoring rule to fit your course — then publish your version to the instructor network as a new seed problem for others to adopt and adapt.

Before you start

New to this? Watch the Connection with Surface Water concept lesson first.

Launch IGW-NET — start the challenge →Watch the concept lesson