
build a year-by-year computational model of a groundwater basin that links water storage, people’s water use, stream habitat and a management technology, and use it to compare strategies.
Phase 3: Make Sense
Do Now. A bathtub holds 100 L. The tap adds 8 L a minute and the drain removes 11 L a minute. How long until it is empty? Show your work.
In the bathtub, which change would stop the water level falling?
A basin as a bathtub. A groundwater basin can be modeled like a bathtub with a tap and a drain. The tap is recharge: rain soaking in, river and canal seepage, and water deliberately spread on fields to sink in. The drain is discharge: pumping, plus natural outflow to rivers and springs. The water in the tub is storage. Conservation of matter gives the model’s one rule:
storage next year = storage this year + recharge − pumping − natural outflow
This is a stock-and-flow model, and it is the core of the computer models water agencies use, which divide a basin into thousands of cells and run the same rule in each.
| Model basin (sample data) | Value |
|---|---|
| Usable groundwater storage at the start | 2,000,000 acre-feet |
| Average recharge (rain, rivers, canal seepage) | 400,000 acre-feet per year |
| Pumping (farms and towns) | 550,000 acre-feet per year |
| Wet-dry recharge sequence | 700,000 in wet years, 200,000 in dry years; two wet years in every five |
| Stream-habitat rule | the valley’s spring-fed creek keeps flowing while storage stays above 1,200,000 acre-feet |
Sample data for a model basin. It is not a measurement of any real valley.
Worked example 1 (the model basin). Ignore natural outflow. Each year storage changes by 400,000 − 550,000 = −150,000 acre-feet. After 1 year: 1,850,000; after 5 years: 1,250,000. At this rate the usable storage is gone in 2,000,000 / 150,000 ≈ 13.3 years. To balance the budget, pumping must fall to 400,000, a cut of 150,000 / 550,000 ≈ 27%. A 15% cut (pumping 467,500) still leaves an overdraft of 67,500 acre-feet a year: slower decline, but still decline.
Adding reality. A useful model includes what matters and leaves out what does not. Four improvements make this one more realistic:
Worked example 2: a user-controlled variable. Jamal’s spreadsheet has one cell for “pumping cut (%)” and one for “managed recharge (acre-feet per wet year).” Every row reads those cells, so changing one number re-runs all 20 years. That is what makes it a simulation and not just a table.
Sustainability and limits. A basin is managed sustainably when, averaged over wet and dry years, storage stops falling and harms such as subsidence, dry wells and a dry creek are avoided. Every model simplifies: this one treats the basin as one well-mixed tub, but a real basin has areas where water tables fall fast and others where they barely move. Say what your model leaves out.
Build the model, then test it.
Design one strategy and test it in your model.
Design one management strategy for the model basin (a pumping cut, managed recharge in wet years, or a combination). State one criterion and two constraints. Run it for 20 years with the wet-dry recharge sequence, graph storage against year, and report the creek's habitat score and whether the strategy meets your criterion.
Name two things this model leaves out of the real valley and explain how each could change your conclusion.
Our basin now runs in a computer. Next lesson we pull the whole unit together and start brainstorming the district’s plan.
How confident are you that you can I can build a computational model of a groundwater basin and use it to compare management strategies.
Exit ticket: write the model's update rule in words and explain what each term stands for in the real valley.
Timing: spans two 50-minute periods: day 1 the base model and the three runs; day 2 the student-designed strategy, graphs and limits.
Materials: devices with JupyterLite or any spreadsheet; graph paper and calculators for the no-device route.
Answer check: base run storage first goes negative in year 14 (−100,000). A 27% cut leaves pumping at 401,500, so storage still slips 1,500 a year; only the exact cut of about 27.3% (pumping 400,000) holds it flat, a good talk about rounding. The wet-dry sequence averages the same 400,000 recharge, so with original pumping it still falls about 150,000 a year on average, just unevenly. No run with a 15% cut meets the criterion.
Listen for: treating the model as the real valley; forgetting the creek. The basin numbers are sample data.
Support / stretch: provide a spreadsheet with the first two rows built. Stretch: add the subsidence rule (lose 2% of capacity in any year storage is below 1,000,000) and compare.