University of the Aegean
Αξιολόγηση της ποιότητας του πόσιμου νερού της Αθήνας, με στόχο την ολοκληρωμένη διαχείρισή του, μέσω στατιστικών μεθόδων
Abstract
dc:descriptionThe aim of this thesis was to explore statistical methods for the integrated analysis of the water quality parameters at all stages (raw water – treated water). Α great variety of statistical methods were extensively explored and then applied to the water quality data of the complicated water supply system of Athens with a view to gaining insight knowledge of the water quality behavior and the system’s integrated management. The study of seasonality and trend analysis of the toxicity data (2000-2006), for each raw water reservoir, resulted in no statistical significance. No statistical significant difference for the toxicity between the reservoirs was observed. The microbial load of the rivers that flow into the reservoirs was investigated by means of seasonal and local variability as well as trend analysis tests for the bacteria indicators of enterococci (1997-2009). Two of the rivers were characterized by exceptionally small microbial load and one by high amount of censored data at the highest reporting limit of the method. No trends for the enterococci were found in the rest of the rivers. Multivariate statistical techniques (Principal components Analysis / Factor Analysis, Cluster Analysis, Discriminant Analysis, Classification and Regression Trees) were applied to physical-chemical data (16 parameters) of each reservoir to explain the data structure and to assess temporal variations in the water quality. The resulting models presented high accuracy and allowed a reduction in the dimensionality of the complex dataset, delineating a few indicator parameters responsible for the variations in temporal water quality. Significant differences in the treated water physical-chemical characteristics due to different sources of surface water and differences in the water treatment facilities and operational procedures were observed. Discriminative models were developed for the treated water produced at the water treatment plants. Moreover, the predictive models were evaluated for the discrimination of network water samples according to their theoretically expected origin. The prediction of the water origin in network was very satisfactory (97.8% for the best model). Therefore, the model can be used in the establishment of safety protocols in the case of an emergency, where a possible deterioration of water quality in one of the plants could degrade the water quality of the related water supply network sections. Finally, control charts were constructed for the data that were obtained from the daily toxicological analysis of treated water from the treated-water tanks. The time-series based method captured the dynamic structure of the data and gave reasonable control charts that can be useful tools for the improvement of the water quality.
Degree
thesis:*- Grantor dc:publisher
- University of the Aegean
- Year dc:date
- 2010
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Smeti, Eleni
- Σμέτη, Ελένη
Subjects
dc:subject × 26- Ποιότητα νερού
- Ανάλυση τάσεων
- Εποχικότητα
- Αυτοσυσχέτιση
- Μη-παραμετρικές στατιστικές μέθοδοι
- Πολυμεταβλητές στατιστικές μέθοδοι
- Διαγράμματα ελέγχου
- Φυσικές-χημικές παράμετροι
- Μοντέλα χρονολογικών σειρών
- Τοξικότητα
- Μικροβιακή φόρτιση
- Water quality
- Trend analysis
- Seasonality
- Autocorrelation
- Non-parametric statistical methods
- Multivariate statistical methods
- Control charts
- Time series models
- Toxicity
- Microbial load
- Physical-chemical parameters
- Κοινωνικές Επιστήμες
- Οικονομικά και Επιχειρήσεις
- Social Sciences
- Economics and Business
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/28730