Estimating Errors in Concentration Measurements Obtained from Image Analysis

Abstract

Image analysis (IA) techniques are increasingly being used in porous media experiments to measure system properties such as concentration and water content. The values of system properties estimated using IA techniques can be influenced by various types of experimental errors. These errors are generally quantified by using global mass balance calculations or by comparing the dispersion coefficient value obtained from the IA data against an accepted value. We used a theoretical test problem to show that both of these error quantification methods have severe limitations. Hence, we developed an alternative statistics‐based method for quantifying IA errors. The applicability of the new method was verified using the theoretical test problem. In addition to quantifying errors, the method can also be used as a design tool for selecting optimal concentration ranges for conducting contaminant transport experiments with minimal errors. We conducted a dense‐tracer transport experiment to demonstrate the use of the proposed error analysis method.

Automated image analysis for experimental investigations of salt water intrusion in coastal aquifers

“…It took around 30 minutes for the 141 beads to become fully saturated and required around 10 litres of saltwater solution for each 142 concentration. The aquifers were flushed with approximately 10 pore volumes to ensure full 143 saturation of the new concentration (Goswami et al, 2008). For these experiments, a power law was 144 adopted to relate light intensity ( ) to concentration ( ) in the form: 145…”

Thermo-Hydro-Mechanical-Chemical Processes in Porous Media

“…image analysis used for concentration measurements: see [160]. image analysis used for concentration measurements: see [160].…”

Modelación de parámetros físicos de la ecuación de advección-dispersión utilizando un tanque de experimentación de escala intermedia.

“…McNeil y Oldenborger (2006) desarrollan un algoritmo de análisis de imagen capaz de identificar automáticamente las zonas de la fotografía del medio que deben ser objeto de una calibración independiente debido a sus propiedades o a la falta de uniformidad de iluminación. Frente a otros estudios previos que utilizan dos (Oostrom et al, 2003), tres (Ursino et al, 2001), o hasta cinco Goswami et al (2009) realizan un detallado análisis de las fuentes de error y correcciones necesarias para obtener un resultado preciso eliminando las fuentes de ruido.…”