Multivariate statistical approaches as applied to environmental physics studies

Abstract

Abstract:The present communication deals with the application of the most important environmetric approaches like cluster analysis, principal components analysis and principal components regression (apportioning models) to environmental systems which are of substantial interest for environmental physics -surface waters, aerosols, and coastal sediments. Using various case studies we identify the latent factors responsible for the data set structure and construct models showing the contribution of each identified source (anthropogenic or natural) to the total measure of the pollution. In this way the information obtained by the monitoring data becomes broader and more intelligent, which help in problem solving in environmental physics.

A Comparative Study on Hydrogeochemistry of Ken and Betwa Rivers of Bundelkhand Using Statistical Approach

“…It reduces a large amount of complex data to a more comprehensible form (Davis 2002;Jayaprakash et al 2008). It allows examining the underlying (or latent) relationships between the variables and their possible influence on the water quality (Lovchinov and Tsakovski 2006;Singh et al 2010). Multivariate analysis was performed on matrix of hydrogeochemical data.…”

Chemometric expertise of the quality of groundwater sources for domestic use

“…In order to avoid classification problems with objects described by variables of completely different size, in the preliminary step of the classification the input data matrix (n objects x m variables) is normalized to dimensionless values (by the use of autoscaling or z-transform, range scaling, logarithmic transformation) which replace the real data values, reducing them to close numbers. [9,19] Then, a similarity measure is applied to calculate the distance between all objects of interest. Very often the Euclidean distance (ordinary, weighted, standardized) is used as a reliable measure of similarity between the classified objects.…”

“…The PCA computational step is followed by regression of the sample mass (concentration) on these APCS to derive each identified source's estimated contribution to the total concentration of a certain chemical variable. [9,12,19,23] The main advantage of this modeling is its receptor orientation and the opportunity to estimate the source emissions without direct measurement. A complete description of the computational procedure can be found in the work of Thurston and Spengler.…”

“…It is required to take into account many environmental parameters and interpret information on the complex relationships between various chemicals in order to achieve a reliable pathway of sustainable development: what to do with a system, how to treat it, and what involvement of authorities is needed, etc. [9] Water quality monitoring and assessment is a process of analysis, interpretation and communication of those properties within the wider context of human activity and use and the conservation of the natural environment. [1] The classification, modeling and interpretation of the monitoring data could be a very important step in the complete assessment of the quality of the water intended for human consumption.…”

“…[1] The classification, modeling and interpretation of the monitoring data could be a very important step in the complete assessment of the quality of the water intended for human consumption. In the last decades, multivariate statistical methods (chemometrics and environmetrics) such as Cluster Analysis (CA), Principal Component Analysis (PCA), Discriminant Analysis (DA), Factor Analysis (FA), have become accepted in detecting similarities in site locations and identifying time trends and sources of pollution in various environmental systems [9][10][11] such as river water, [10,[12][13][14] potable water, [1,8,[15][16][17][18] marine sediments, [19,20] soils, [19,21] etc. Also, quality of surface waters have been assessed using Hasse diagram technique (HDT) and self-organizing maps of Kohonen (SOM).…”