Methodology

Phase II Investigation Design: Applying ASTM E1903-19 on Korean Sites

First in a methodology series on how we actually perform the work. A Phase II program lives or dies at the design stage — before a single boring is drilled. ASTM E1903-19 exists because "go take some samples" is not a method; here is the design discipline the standard imposes, where it interlocks with Korean statutory procedure, and the design mistakes we see repeated.

What the standard actually asks for

Stripped to its logic, E1903-19 requires an investigation to be an argument, not an errand. The elements that matter in practice:

• Objectives before holes. The program starts from explicit questions — typically inherited from the Phase I's findings and the client's decision at hand. "Is the former tank area a source?" is a designable objective; "characterize the site" is not, and the standard's discipline is largely a machine for refusing the second phrasing.
• A site model doing the aiming. Sampling locations, depths, and parameters follow from a working understanding of sources, pathways, and geology — the conceptual site model, which gets its own treatment in the next piece in this series. No model, no aim.
• Lines of evidence, plural. Analytical results, field observations, geophysics, historical records — the standard treats these as converging evidence rather than expecting laboratory numbers to carry the conclusion alone.
• Iteration as a feature. The standard anticipates staged work: screen, interpret, refine, and either close the question or drill the next round with better aim. A single mobilization that answers everything is the exception, not the design target.
• Documentation of the reasoning. Why here, why this depth, why these parameters — recorded, so the program can be defended later by someone who wasn't there.

Where it meets the Korean system

E1903 is a process guide with no legal force anywhere — in Korea, results acquire legal meaning only through the Korean apparatus. Three interfaces decide whether a well-designed program also produces a legally usable one:

• Numbers land on Korean tables. Analytical results matter when compared against the soil concern standards and their land-use categories — so parameters, methods, and reporting limits must be chosen with the Korean comparison in mind from the start, not converted afterwards. A result the Korean official method cannot reproduce is an expensive opinion.
• Who samples matters, not just how. Korean statutory processes recognize work by designated institutions; a program intended to feed an official proceeding — or the statutory Soil Environment Assessment — must be executed by the right hands to count. The dual-frame program uses E1903 logic for design and Korean designation for execution, a split we covered in the Phase II overview.
• The iteration gates map onto real decisions. In Korean transactions the E1903 stages align naturally with deal gates — screen during exclusivity, confirm before signing, characterize after closing if needed — and with the statutory escalation from finding to detailed investigation. Designing the rounds to land on those gates is what makes the iteration principle commercially real, as the timeline piece lays out.

The mistakes that repeat

1. The grid reflex. Evenly spaced borings look rigorous and answer nothing in particular. A hypothesis-driven pattern — dense at suspected sources, sparse where the model says clean — answers the actual question at half the cost. Grids are for characterization after liability is settled, not for diligence.
2. Depths by habit. Stopping at a standard depth regardless of contaminant behavior — light products smearing at the water table, dense solvents running deeper — produces false negatives with perfect paperwork. Depth is a per-target decision, the same logic that governs groundwater screen settings.
3. The copied parameter list. Last project's analyte suite, reused. Money burns on irrelevant parameters while the one that matters — the plating shop's chromium, the fab's chlorinated solvents, the foam area's PFAS — sits off-list. The site history writes the parameter list; the lab catalogue doesn't.
4. No decision rule. The most subtle failure: nobody wrote down what any given result would change. If a hit and a non-hit lead to the same next step, the sample was decoration. Every designed point should carry an if-then that connects it to the deal or compliance decision it serves.
5. Designing for one audience. A program that satisfies the US reviewer and is invisible to Korean procedure — or the reverse — does half its job. The design review question is always: who must accept this data, and does the design satisfy all of them at once?

Last reviewed: 31 August 2026 · Dime Works