info:eu-repo/semantics/article
Fusing data of different orders for environmental monitoring
Fecha
2019-11Registro en:
Martínez Bilesio, Andrés Rogelio; Batistelli, Marianela; Garcia Reiriz, Alejandro Gabriel; Fusing data of different orders for environmental monitoring; Elsevier Science; Analytica Chimica Acta; 1085; 11-2019; 48-60
0003-2670
CONICET Digital
CONICET
Autor
Martínez Bilesio, Andrés Rogelio
Batistelli, Marianela
Garcia Reiriz, Alejandro Gabriel
Resumen
In the present work a novel application of data fusion to an environmental monitoring study is proposed. This paper involves the joint analysis of zeroth-, first- and second-order data measured on a particular environmental system. The main advantage of this methodology is the possibility of analyzing the relationships of the different order data provided by several analytical techniques. This approach enables to achieve new knowledge, in a way that would be not accessible if considering the information individually. Environmental monitoring databases usually generate large amount of data. Multivariate statistical techniques are necessary to process all this information and obtain a correct interpretation. The Ludueña Stream located in Argentina was chosen as the study system. Samples from different sites of the basin were taken periodically. Conductivity and pH (zeroth-order data) were fused with near-infrared (NIR) spectra of suspended particulate material (first-order data) and with fluorescence emission-excitation matrices of dissolved organic matter (second-order data). Different chemometric algorithms made it possible to extract and merge all the information in a new database, enabling its later analysis as a whole. This methodology allowed to successfully studying the behavior of dissolved organic matter together with suspended particulate material and other specific variables, showing links between them. Their distributions along the basin and their evolutions over time were possible to obtain. Therefore, a simpler interpretation to evaluate the system status was achieved. This model allowed differentiating the variables affected by anthropic activities from those with a natural origin.