Nitrates/nitrites concentration estimation in wastewater samples using transmittance curve models optimized by evolutionary computation techniques Pablo Lo ´ pez-Espı ´, Sancho Salcedo-Sanz, A ´ ngel M. Pe ´ rez-Bellido, Irene de Bustamante and Francisco Lo ´ pez-Ferreras ABSTRACT Pablo Lo ´ pez-Espı ´ Sancho Salcedo-Sanz (corresponding author) A ´ ngel M. Pe ´ rez-Bellido Francisco Lo ´ pez-Ferreras Departament of Signal Theory and Communications, Universidad de Alcala ´ de Henares, Campus Universitario, E-28871, Alcala ´ de Henares Madrid, Spain Tel.: +34 91 885 6731 Fax: +34 91 624 8749 E-mail: sancho.salcedo@uah.es Irene de Bustamante Department of Geology, Universidad de Alcala ´ de Henares, Campus Universitario, E-28871, Alcala ´ de Henares Madrid, Spain This paper presents an application of evolutionary computation algorithms in the estimation of the concentration of nitrates/nitrites in water. Specifically, we start from the measured transmittance curves of a water sample and a model of the curve consisting of a mixture of polynomial, Fermi and Gaussian functions. An evolutionary programming algorithm is then used to obtain the optimal parameters of the model which minimize the distance between the measured and the modeled transmittance curves. This process allows us to separate the modeled transmittance curve into several components, one of them associated with the nitrate/nitrite concentration. We can extract the nitrate/nitrite concentration of the water sample using this component of the transmittance model. We test our proposal on several laboratory samples and on three real samples measured in different locations around Madrid, Spain. Key words | evolutionary computation, evolutionary programming, nitrates/nitrites concentration estimation, ultraviolet spectroscopy INTRODUCTION Nitrogen is the most common pollutant found in ground- water. In many cases, nitrates are found in dangerous concentrations mainly due to agricultural activities: nitro- gen is a vital nutrient to enhance plant growth. This fact has motivated the intensive use of nitrogen-based fertilizers to increase the productivity of crops in many regions of the world (Almasri & Kaluarachchi 2004, 2007). Thus, it is known that agricultural activities are the main source of elevated nitrate concentrations in groundwater (Almasri & Kaluarachchi 2007). The problem is that, in many cases, application of nitrogen-rich fertilizers exceeds the plant’s demand, so the nitrogen reaches the groundwater. It has been reported that elevated nitrates concentrations in drinking water can cause different serious health problems in children and adults (Wolfe & Patz 2002), so the detection and control of these contaminants are very important tasks in which a huge amount of research is being carried out. Different methods have been proposed to accurately estimate the concentration of nitrates in water. We briefly summarize them in the next subsection, before stating the objective and structure of this paper. Measurement techniques for nitrates concentration estimation Ionic chromatography (IC) is one of the most used methods to estimate nitrates concentration. In IC, a process of ionic interchange is used to separate molecules with different charges. IC allows us to determine very low concentration values of contaminants, so it is really useful in water quality studies (Holm et al. 1997). Another interesting property is that IC is able to detect different ions from nitrates: however, the measurement time is relatively large, depend- ing on the type of sample (Holm et al. 1997). doi: 10.2166/hydro.2010.016 446 Q IWA Publishing 2010 Journal of Hydroinformatics | 12.4 | 2010 Downloaded from https://iwaponline.com/jh/article-pdf/12/4/446/476737/446.pdf by guest on 02 June 2020