A preliminary assessment of the novel application ASMITAS using sediments from Matlab, Bangladesh

Sammanfattning: Most of the drinking water supply in rural Bangladesh comes from groundwater collected using shallow tubewells. The tubewells, usually shallow because of the increased cost involved in deeper tubewells, have been installed by local drillers. A Sediment Color Tool was developed, with input from local drillers, that associated the arsenic concentration with specific sediment colors, in order to help the drillers install safe tubewells. This tool was digitized into the phone application, ASMITAS, to reduce subjectivity in sediment color determination due to human error or surrounding conditions, when used with a color sensor. The purpose of this study was to carry out a preliminary assessment of the application performance and usability, and the results provided by the application for color identification. 35 sediments were used and assigned into 4 different data sets to allow for comparison. Two data sets were assigned a Munsell color manually, while two were assigned the Munsell Soil Color (or Red-Green-Blue color) through use of the digital app. The sensor, the Nix Color Sensor Pro 2, was validated through a literature review and is considered accurate in identifying the color of the soil sediments. The data sets were compared based on the Delta E 2000 formula to determine the color difference between the data sets. The most relevant result of this method was between the Red-Green-Blue that the Nix Sensor originally provided to the application versus the closest matching Munsell code that the application could provide. It showed that the library from which the Munsell color was drawn was not yet expansive enough to accurately identify all sediments that may be scanned. Cyan-Magenta-Yellow-Black color comparisons were made to ascertain which aspects of the color are the most difficult to identify. It was found that both the sensor and the human eye had difficulties in identifying differences in the yellow percentage of several of the samples. The results showed that there may be greater need for distinction of which yellow percentages of Cyan-Magenta-Yellow-Black belong to which color sediment. Overall, the application appears to have a small number of less prominent features and functions to improve on prior to the publication of the application. At this stage of development, the main goal lies in the improvement and building of the Munsell color code reference library and the library of arsenic concentrations associated with each sediment color within the application, in order to improve the accuracy of the results.