BLOG & INSIGHTS

Five Things Everyone Believes About Field Screening That Aren't True

By: TRIUM Environmental
October 9, 2026
3
min read

Very few people would defend the bag-and-PID method in a technical debate[cite: 1]. Almost everyone still uses it to decide where to stop digging[cite: 1]. That gap between what practitioners believe and what practitioners do is held open by a handful of assumptions that rarely get examined, because they sound like common sense[cite: 1]. Here are five of them, and what the evidence actually says[cite: 1].

1. "High is high and low is low"

This is the workhorse defence of conventional vapour screening[cite: 1]. Nobody claims the bag method is quantitative; the claim is that it is directionally reliable — that a high reading means more contamination than a low one, which is enough to steer a dig[cite: 1].

The trouble is that a headspace bag reading is driven by variables that have nothing to do with contaminant concentration: how much soil went in the bag, how much air is above it, how warm the sample was, how long it sat, how vigorously it was worked, and how the probe was inserted[cite: 1]. Vapour pressure is strongly temperature-dependent, so an uncontrolled temperature means an uncontrolled measurement[cite: 1]. Reported correlation to laboratory data for these methods is typically below 25%, and frequently below 10%[cite: 1].

A head-to-head field study makes the point without any interpretation required[cite: 1]. Sixty-two samples were screened by both the conventional bag method and a temperature-controlled, standardized system, using the same detector on the same soils, with 31 sent for laboratory confirmation[cite: 1]. The standardized system tracked laboratory Total BTEX at R² = 0.94[cite: 1]. The bag method, on those same samples, returned R² = 0.002 — and a regression slope statistically indistinguishable from zero (p = 0.91)[cite: 1]. For Total PHC the bag method was worse still, at p = 0.995[cite: 1].

A slope that is statistically indistinguishable from zero means the reading carries no information about concentration[cite: 1]. Not weak information[cite: 1]. None[cite: 1].

High is not reliably high[cite: 1]. Low is not reliably low[cite: 1]. On any given sample, most of what the needle is telling you is how that sample was handled[cite: 1].

2. "Field screening is trying to replace the laboratory"

This is the objection that ends most conversations before they start, and it misreads the purpose entirely[cite: 1]. No credible field programme proposes to replace accredited analysis, and no regulatory closure requirement changes because a field instrument was used[cite: 1]. Confirmation samples are still submitted; criteria are still met with laboratory data[cite: 1].

What changes is what happens between excavation and confirmation[cite: 1]. Under conventional screening the practitioner has two options: wait days for laboratory turnaround before deciding where to dig next, or proceed on an unreliable reading and hope[cite: 1]. Reliable field data removes that dilemma[cite: 1]. The laboratory then does what it is uniquely good at — analytical precision and regulatory-grade documentation — for samples that were chosen because they answer specific questions, rather than because they smelled the worst[cite: 1].

The relationship is complementary and it compounds[cite: 1]. Each laboratory result is compared against the field reading for the same sample, quantifying site-specific bias and calibrating the model, so the next deployment predicts better than the last[cite: 1].

3. "If the field result and the lab result disagree, something went wrong"

Intuitive, and mostly false[cite: 1]. EPA guidance on matrix effects documents that soil heterogeneity influences result variability roughly nineteen times more than the choice of analytical method[cite: 1]. Two subsamples of the same interval are, chemically speaking, two different samples[cite: 1].

Laboratories live with this openly[cite: 1]. Duplicate QA/QC tolerances of ±30–40% are standard, and relative percent differences above 50% are common in heterogeneous soil[cite: 1]. If the accredited laboratory can disagree with itself by half, expecting a field instrument to reproduce a laboratory number exactly is not a meaningful standard[cite: 1].

There is a better way to read a disagreement: as information about the site rather than a verdict on the instrument[cite: 1]. A pair that disagrees sharply is telling you that the material at that location is variable — which is precisely the kind of thing you want to know before you draw a boundary through it[cite: 1]. Systems that collect field duplicates and triplicates convert that scatter into a stated probability rather than leaving it as noise[cite: 1].

4. "More samples means more cost"

Only if every sample goes to a laboratory[cite: 1]. The instinct to minimize sample count is a rational response to per-sample laboratory pricing, but it optimizes the wrong stage of the project[cite: 1].

Screen two hundred locations in the field and submit the twenty-five most informative, and the laboratory bill falls sharply while the conceptual site model gets stronger, not weaker[cite: 1]. More importantly, the costs that actually destroy project margins are not analytical[cite: 1]. They are the supplemental mobilization to chase an impact that was missed, the clean soil that was excavated and hauled because nobody knew where the boundary was, the failed audit that reopens a closed file, and the closure timeline that stretches while liability sits on the books[cite: 1].

Every one of those has the same origin: a decision made without enough data[cite: 1]. That is the inversion at the centre of conventional practice — cost and time are optimized at the sample-collection stage, which defers uncertainty rather than eliminating it, and deferred uncertainty always costs more to resolve later than it would have cost to resolve in the field[cite: 1].

What gets counted[cite: 1] What doesn't[cite: 1]
Per-sample laboratory cost, field days, equipment rental.[cite: 1] Supplemental mobilizations, over-excavation and disposal, under-excavation and re-opening, failed audits, extended closure timelines, unnecessary laboratory samples submitted blind.[cite: 1]

5. "Regulators won't accept it"

This one is worth stating precisely, because it contains a real question wrapped in a false premise[cite: 1].

The false premise is that a screening method requires regulatory approval[cite: 1]. It does not[cite: 1]. Field screening is not a compliance determination — closure is still demonstrated with accredited laboratory data[cite: 1]. Choosing a better screening method is an operational decision made by the client and the consultant, in the same way that no one seeks agency sign-off to use better field equipment generally[cite: 1].

The real question underneath is whether a programme built on high-density field data is more defensible or less[cite: 1]. The regulatory record answers clearly[cite: 1]. The EPA's Triad approach explicitly recommends real-time measurement technologies and dynamic work strategies[cite: 1]. The Data Quality Objectives process asks practitioners to define the decision, the tolerable decision error, and the data needed to meet it — which is exactly the false-negative and false-positive framework a well-run screening programme reports against[cite: 1]. CCME Volume 1 recognizes headspace vapour and solvent-extraction testing as field analytical methods and expects field precision assessment, duplicates and triplicates included, as standard professional practice[cite: 1]. CSA Z769 places responsibility for the adequacy of the sampling programme on the professional overseeing it — a responsibility that cannot be discharged by following a fixed protocol that ignores what the site is showing[cite: 1].

Read together, these say something uncomfortable for the conventional approach: the greater professional risk lies in under-sampling, not in over-characterizing[cite: 1]. A programme that screened densely, documented its QA/QC and adapted to what it found is easier to defend than one that took ten samples and asked the reviewer to accept the model on faith[cite: 1].

The One Thing to Remember[cite: 1]

None of these five beliefs is unfounded — each was reasonable when the only field tools available were qualitative[cite: 1]. What has changed is that the alternative now exists, which turns each of them from a description of reality into a decision to keep working with less information than the site is willing to give you[cite: 1].

Sources referenced[cite: 1]

  • TRIUM. AISCT® Aurora — Comparative Field Study (62 samples, 31 laboratory-confirmed).[cite: 1]
  • CCME. Guidance Manual for Environmental Site Characterization, Volume 1 (2016), Section 5.5.1 and Suggested Operating Procedures.[cite: 1]
  • U.S. EPA. Best Management Practices: Systematic Project Planning Under a Triad Approach (EPA 542-F-10-010, 2010).[cite: 1]
  • TRIUM. The Why — Physical AI for Complex Environments (technical bulletin, 2026).[cite: 1]

Join the AISCT Insight List

New technical content, case studies, and regulatory updates delivered monthly

We respect your privacy, unsubscribe anytime.
YOU'RE SUBSCRIBED!
Oops! Something went wrong while submitting the form.

Why work with TRIUM?

Proven Technology

AISCT systems are field-proven, scientifically defensible, and regulatory-aligned.

Export Support

Our team brings deep environmental and analytical expertise to every project.

End-to-End Partnership

From project planning to data delivery, we're with you every step of the way.

Need Immediate Assistance?

Reach out and a member of our team will respond back to you as soon as possible.

1 (403) 932-5014

Office Hours

Monday - Friday

8:00 AM - 5:00 PM

Local Time

Sustainability Commitment

TRIUM Environmental is committed to responsible innovation and advancing environmental outcomes through technology