Before calculating: a boundary card and four acceptance checks

Steps
1. Choose one catchment and one period for precipitation P, evapotranspiration ET and runoff Q. Use depths in millimetres and catchment area in square kilometres. Annual precipitation and monthly runoff cannot share one residual.
2. List transfers that the three-term model omits: imports, abstractions or groundwater exchange may matter. The residual cannot identify which omitted process occurred.
3. Run the four cases below. Positive, zero and negative residuals are all allowed. A negative precipitation or a zero area is invalid for this teaching tool; correct it before interpreting a number.
4. Save the inputs and output together. When using real measurements, record their uncertainty and sampling period. Additional decimal places in the output do not increase the quality of the observations.

Worked record
P=800, ET=450, Q=300, area=10: ____________________
P=800, ET=450, Q=350, area=10: ____________________
P=800, ET=450, Q=400, area=10: ____________________
P=−1 or area=0: ____________________
Unit check: ____________________

P=800, ET=450, Q=300, area=10: ΔS=+50 mm; equivalent +500,000 m³
P=800, ET=450, Q=350, area=10: ΔS=0 mm; equivalent 0 m³
P=800, ET=450, Q=400, area=10: ΔS=−50 mm; equivalent −500,000 m³
P=−1 or area=0: Invalid input; no interpreted balance
Unit check: 1 mm × 1 km² = 1,000 m³; 50 × 10 × 1,000 = 500,000

The worked cases test arithmetic and input handling. They do not validate a real catchment, predict flooding or establish sustainable withdrawals. If your boundary contains an unmeasured transfer, record that limitation beside the residual rather than silently treating the transfer as zero.

USGS: Water Budgets, Circular 1308 — https://pubs.usgs.gov/publication/cir1308