Delineating high-density areas in spatial Poisson fields from strip-transect sampling using indicator geostatistics:...

Area of Science:

• Geostatistics
• Spatial analysis
• Environmental science

Background:

• Spatial Poisson fields are common in modeling discrete event occurrences.
• Delineating high-density clusters from limited transect data is challenging.
• Accurate mapping of anomalies is crucial for site remediation, such as unexploded ordnance (UXO) removal.

Purpose of the Study:

• To develop an optimized approach for delineating high anomaly density areas.
• To apply this method to spatial mixtures of Poisson fields using strip transect data.
• To improve the accuracy of identifying clusters for practical applications like UXO site cleanup.

Main Methods:

• Transformation of sampled anomalies to anomaly count data.
• Application of indicator kriging to estimate probabilities exceeding a threshold.
• Determination of an optimized threshold value derived from the cumulative distribution function (cdf) of the background Poisson field.
• Utilizing a low-pass filter to enhance segmentation of high-density areas.

Main Results:

• Successfully developed and demonstrated an approach for delineating high anomaly density areas.
• The indicator kriging and optimized threshold method effectively identified clusters.
• Low-pass filtering enhanced the segmentation of transect data for improved delineation.

Conclusions:

• The developed method provides an effective means for delineating high anomaly density areas within complex spatial fields.
• This approach is particularly valuable for applications requiring precise identification of anomaly clusters, such as UXO remediation.
• The combination of indicator kriging, optimized thresholding, and filtering offers a robust solution for limited spatial data scenarios.