Validation — AERMOD 24142

Version 24142 is the AERMOD version this platform launched with, and its validation is the strictest we run: every EPA test case was built through the product itself — the same pages, forms, and map you use — then run on the production cloud, and the output compared to EPA's published reference, design value by design value and receptor by receptor.

This is a stronger claim than re-running EPA's cases on a bench copy of the model. It validates the entire platform end-to-end: the inputs you enter on screen, the input file the platform builds from them, the cloud execution, and every number shown back to you.

Re-validated July 19, 2026 — full library re-run on the current build; all 40 comparable cases matched again, most exactly. The meteorology chain was re-verified the same day: AERMET reproduced all 8 EPA reference cases byte-for-byte, and AERSURFACE matched EPA's reference exactly.

Result: every comparable case validated

EPA's 24142 test set gives 41 cases. Of these, 40 ship with usable reference outputs — all 40 validated:

• 39 cases reproduce EPA's reference fully — every design value, every receptor cell, and every output row identical.
• 1 case (capped, capped/horizontal stacks) matches all 162 design values exactly; 6 of its 23,328 grid cells differ within 4% on a single non-design statistic, within the cross-compiler envelope described in the 26135 report.
• The remaining case, scimtest, has no usable reference to compare against: the reference output in EPA's own published bundle terminates in a runtime error before producing summaries. The case still builds and runs to completion on this platform.

What each case runs

Each row is one EPA case, built as a user and run on the platform. Together they cover all 13 source types, the five NO₂ chemistry options in both background unit systems, building downwash, urban dispersion, hourly and variable emissions, gas and particle deposition, PSD credit groups, and every receptor network type.

Why this matters

A modeling tool can run the model correctly and still give a wrong answer — through a mis-mapped unit, a dropped option, or a coordinate shifted on import. Because these 41 cases were entered through the same screens you use, agreement with EPA's references validates the whole chain: what you enter, the model input built from it, the cloud execution, and every number displayed back. The suite is re-run as the platform evolves, so that agreement is maintained, not just demonstrated once.

For version 26135, see the companion report Validation — AERMOD 26135, which covers EPA's 53-case transition suite for the current regulatory version.