API Publ 306-1991: An Assessment of the Use of Split-Sampling for Environmental Data Quality — Technical Overview and Implementation Guide

API Publication 306 (API Publ 306-1991), titled An Assessment of the Use of Split-Sampling for Environmental Data Quality, provides a structured methodology for implementing split-sampling programs in environmental monitoring. Split-sampling involves dividing a single sample into two or more portions, with one portion retained for the primary analysis and the other(s) submitted to a secondary (often independent) laboratory for comparison. This publication serves as a foundational reference for quality assurance/quality control (QA/QC) in environmental data collection, particularly where data defensibility and comparability are critical.

Scope of API Publ 306

The publication defines the scope of split-sampling as a tool for assessing and improving the quality of environmental data, especially in the context of petroleum industry monitoring programs. It addresses both field and laboratory aspects, including:

Field split-sampling: Division of a physical sample (e.g., water, soil) at the collection point.
Laboratory split-sampling: Subdivision of a received sample within the analytical facility.
Data quality indicators: Precision, bias, and comparability between analytical results.
Study design: Frequency, sample size, and statistical treatment.

The intended users include environmental managers, field sampling teams, laboratory analysts, and regulatory personnel involved in oil and gas operations, spill response, and site remediation.

Technical Requirements

Split-Sampling Protocol Design

API Publ 306 outlines a systematic approach to designing a split-sampling program. Key technical elements include:

Parameter	Requirement	Acceptance Criteria
Sample volume	Sufficient for all intended analyses including splits	≥3× the volume needed for primary analysis
Split frequency	Minimum 10% of total samples (or ≥1 per sampling event)	Based on project Data Quality Objectives (DQOs)
Homogenization	Before splitting, ensure sample is well-mixed and representative	No visible stratification; mixing time ≥2 min per liter
Custody and handling	Chain-of-custody records for each split	Full traceability: date, time, person, method
Laboratory qualification	Both primary and secondary labs must use accredited methods	ISO/IEC 17025 or equivalent
Result comparison	Calculate relative percent difference (RPD)	RPD ≤30% for organic analytes; ≤20% for inorganic

Field vs. Laboratory Splitting

The document distinguishes between two primary splitting approaches:

Field split: Performed at the sampling site using equipment such as split-core samplers, riffle splitters, or by coning and quartering. This method provides a measure of whole‐sample variability.
Laboratory split: The sample is received intact and then subdivided in the lab using techniques like rotary splitting, taking care to avoid bias. This isolates analytical variance from field handling variance.

The choice between methods depends on the project’s data quality objectives. API Publ 306 recommends including both types where possible to evaluate total measurement uncertainty.

Data Quality Objectives (DQOs)

Before implementation, the publication stresses that clear DQOs must be defined, including:

Acceptable limits for bias and precision between splits.
Statistical power required to detect meaningful differences.
The intended use of the data (e.g., compliance, risk assessment, trend analysis).

Tip: For high-profile projects (e.g., regulatory enforcement), consider using a third-party laboratory for the split sample to enhance transparency and credibility.

Warning: Incomplete documentation of the splitting procedure is the most common reason for rejection of split-sample data in litigation or regulatory reviews.

Implementation Highlights

Integrating Split-Sampling into a QA/QC Plan

API Publ 306 provides a template for incorporating split-sampling into an existing QA/QC program. Steps include:

Establish project DQOs — define the level of data quality required.
Design the split-sampling plan — determine number, type (field/lab), and frequency of splits.
Select qualified laboratories — ensure both labs have validated methods and comparable detection limits.
Conduct training — field and lab personnel must understand proper splitting techniques.
Perform splitting during sample collection/analysis — log all steps.
Evaluate results — calculate RPD, perform t-tests or ANOVA as appropriate.
Implement corrective actions — if criteria are not met, investigate root causes and revise protocols.

Statistical Evaluation

The publication recommends the use of simple statistical tests to evaluate split-sample comparability:

Relative Percent Difference (RPD): For paired comparisons.
t-Test (paired): To assess whether differences are statistically significant.
Control charts: For ongoing monitoring of split-sample performance over time.

Example RPD calculation:
RPD = (|C₁ − C₂| / ((C₁ + C₂)/2)) × 100%
where C₁ and C₂ are the concentrations from the primary and split analyses.

Success: Organizations that consistently implement split-sampling per API Publ 306 report higher data acceptance rates during regulatory audits and reduced re‑sampling costs.

Compliance Notes

Regulatory Alignment

While API Publ 306 is a non‑mandatory publication, its principles align with regulatory requirements under programs such as the EPA Resource Conservation and Recovery Act (RCRA), Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA), and state-level groundwater monitoring regulations. Many consent decrees and permits explicitly reference split-sampling as an acceptable method for demonstrating data quality.

Audit and Documentation

Compliance with API Publ 306 involves maintaining:

Complete chain-of-custody records for each split.
Standard operating procedures (SOPs) for splitting methods.
Training records for personnel performing splits.
Summary reports of split comparison results and any corrective actions taken.

Important: Failure to follow the documented split-sampling plan may result in the exclusion of split data from evidence in enforcement actions. Always adhere to the approved plan and document any deviations.

Frequently Asked Questions

Q: What is the main difference between a field split and a laboratory split as described in API Publ 306?
A: A field split is performed at the sampling site before the sample is shipped, capturing both field and analytical variability. A laboratory split is performed after the sample reaches the lab, isolating only the analytical variability. Both provide valuable but distinct information about data quality.

Q: How many split samples are typically required for a monitoring program?
A: API Publ 306 recommends a minimum of 10% of the total sample set or at least one split per sampling event, whichever is greater. The exact number should be based on the project’s Data Quality Objectives and the desired statistical confidence.

Q: Is API Publ 306 still relevant since it was published in 1991?
A: Yes. While newer QA/QC documents exist, API Publ 306 remains a foundational guide that many current standards and regulatory frameworks still reference for split-sampling protocols. Its principles are timeless and widely adopted in the environmental industry.

Q: What should I do if my split-sample results fail the acceptance criteria (e.g., RPD >30%)?
A: Investigate potential sources of error including improper homogenization, contamination, analytical error, or laboratory performance issues. Document the findings, take corrective action, and if necessary, resample. The publication stresses that a root cause analysis should be part of the QA/QC process.

📥 Standard Documents Download

🔒

Please wait 10 seconds, the download links will appear after the ad loads