Khalil, Bahaa
(2010).
Méthodes statistiques pour l'évaluation et la reconfiguration des réseaux de suivi de la qualité de l'eau de surface.
Thèse.
Québec, Université du Québec, Institut national de la recherche scientifique, Doctorat en sciences de l'eau, 333 p.
Résumé
This study addresses the assessment and redesign of surface water quality monitoring (WQM)
networks. The design of WQM networks depends primarily upon the objectives of the
monitoring program and the characteristics of the monitored region. Despite several statistical
approaches that have been proposed for the assessment and redesign of long-term WQM
networks, several deficiencies in these approaches exist.
The main goal of this study is to propose statistical approaches for the assessment and redesign
of WQM networks that overcome existent deficiencies in the currently applied approaches. In
addition, this study intends to introduce an innovative approach for the estimation of the water
quality characteristics at ungauged sites. Four objectives are specified: (i) to review the current
applied statistical approaches for the assessment and redesign of surface WQM networks; (ii) to
develop a new statistical approach for the rationalization of water quality variables; (iii) to
develop a new statistical approach for the assessment and redesign of WQM locations; and (iv)
to introduce a statistical methodology for the estimation of water quality characteristics at
ungauged sites.
In this study, statistical approaches used for the assessment and redesign of surface water quality
monitoring networks are first reviewed. In this review, various monitoring objectives and related
procedures used for the assessment and redesign of surface WQM networks are discussed. For
each approach, advantages and disadvantages are examined from a network design perspective.
The literature review reveals that correlation-regression is the most common approach used to
assess and eventually reduce the number of water quality variables in WQM networks. However,
several deficiencies in this approach are identified. Based upon these identified deficiencies, a
new statistical approach is proposed for the rationalization of water quality variables. The
proposed approach overcomes deficiencies in the conventional correlation-regression approach
and represents a useful decision support tool for the optimized selection of water quality
variables. It allows for the identification of optimal combinations of water quality variables to be
continuously measured and those to be discontinued.
To reconstitute information about discontinued water quality variables, four record extension
techniques are examined. Ordinary least squares regression (OLS), the line of organic correlation
(LOC), the Kendall-Theil robust line (KTRL) and KTRL2, which is a modified version of the
KTRL proposed in this study. The advantage of the KTRL2 is that it includes the advantage of
LOC in maintaining variability in the extended records and the advantage of KTRL in being
robust in the presence of extreme values. Monte-Carlo and empirical studies are conducted to
examine these four techniques for bias, standard error of moment estimates and a full range of
percentiles. The Monte-Carlo study showed serious deficiencies in the OLS and KTRL
techniques, while the LOC and KTRL2 techniques have results that are nearly similar. Using real
water quality records, the KTRL2 is shown to lead to better results than the other techniques.
The literature review also reveals that several deficiencies in the approaches proposed for the
assessment of monitoring locations exist. The deficiencies vary from one approach to another,
but generally include: (i) ignoring the characteristics of the basin being monitored in the design
approach; (ii) handling multivariate water quality data sequentially rather than simultaneously;
(iii) focusing mainly on locations to be discontinued; and (iv) ignoring reconstitution of
information at discontinued locations. A methodology that overcomes these deficiencies is
proposed. In the proposed methodology, hybrid-cluster analysis is employed to identify groups
of sub-basins with similar characteristics. A stratified optimum sampling strategy is then
employed to identify the optimum number of monitoring locations in each of the sub-basin
groups. An aggregate information index is employed to identify the optimal combination of
locations to be discontinued. Results indicate that the proposed methodology allows the
identification of optimal combinations of locations to be discontinued, locations to be
continuously measured and sub-basins where monitoring locations should be added.
To fulfill the last objective, two models are developed for the estimation of water quality mean
values at ungauged sites. An ensemble artificial neural network (EANN) model is developed to
establish the functional relationship between water quality mean values and basin attributes. The
second model is based on canonical correlation analysis (CCA) and EANN. CCA is used to form
canonical attributes space using data from gauged sites. Then, an EANN is applied to identify the
functional relationships between water quality mean values and the attributes in the CCA space.
A jackknife validation procedure is used to evaluate the performance of the two models. The
results show that the developed models are useful for estimating the water quality status at
ungauged sites. However, the CCA-based EANN model performed better than the EANN model
in terms of prediction accuracy.
Type de document: |
Thèse
Thèse
|
Directeur de mémoire/thèse: |
Ouarda, Taha B. M. J. |
Co-directeurs de mémoire/thèse: |
St-Hilaire, André |
Mots-clés libres: |
statistique; réseaux de suivi; qualité; eau de surface; estimation; site non jaugé; rationalisation des variables; refonte |
Centre: |
Centre Eau Terre Environnement |
Date de dépôt: |
17 févr. 2014 21:44 |
Dernière modification: |
28 janv. 2021 15:06 |
URI: |
http://espace.inrs.ca/id/eprint/1789 |
Actions (Identification requise)
 |
Modifier la notice |