Correlaciones fisicoquímicas del agua con imágenes específicas preliminares

Abstract

The present work consolidates a visual-physicochemical repository of nine types of water (waste, natural and drinking) sampled for four weeks, with the purpose of evaluating the relationship between visual characteristics and water quality parameters. The capture of optical images, photographs of nitrocellulose membranes, and UV–Vis spectra (250–600 nm) was standardized, along with measurements of pH, conductivity, chlorides, chemical oxygen demand (COD), turbidity, and apparent color. The results showed very high correlations between the area under the spectral curve and color (r = 0.995) and between the area under the curve and COD (r = 0.965), which demonstrates the correspondence between visual patterns and the organic load of the water. The comparative analysis made it possible to accurately differentiate wastewater, natural and drinking water, validating the use of the image as an exploratory indicator. The generated repository constitutes a robust and reproducible database that could feed future artificial intelligence models aimed at the preliminary estimation of water quality parameters using photographs. This study contributes to the development of accessible and low-cost monitoring tools, useful for communities without access to specialized laboratories, as long as their limitations are recognized and they are complemented with conventional analyses.