APPLICATION OF REMOTE SENSING FOR ADJUSTMENT AND CALIBRATION OF WATER QUALITY MODEL - Integrating GIS, Remote Sensing, and Mathematical Modelling for Surface Water Quality Management - page 188

8.7. UNCERTAINTIES WITHIN THE APPLIED TOOLS AND METHODS
Uncertainty analysis for model simulation is of growing importance in the field of water quality
management. In water quality modelling, sources of uncertainty could be; error due to inaccuracies of
the input data, model structure, uncertainty in the estimated model parameter values, and the
propagation of prediction errors. Added to these main sources, the complexity of the system under
study and the integration of different methodologies, modelling tools calibration and verification tools
could be another added indirect sources of uncertainty. The integration of knowledge in the form of
mathematical modelling, remote sensing modelling and earth observation data can be very useful for a
variety of reasons: i) such models and tools make it possible to test hypotheses on functional
interactions in the systems under investigation. (ii) they are tools for interpreting knowledge between
different users for management and decision support. (iii) they can be used for predicting future states
of the systems or its responses to assumed or expected changes in driving conditions. iv) they can be
used for forecasting of future extreme events. In the mean time the complex the modelling system and
tools gets the more uncertain predictions might be associated. In this research study all the mentioned
sources of uncertainty are involved within the used and/or developed tools, research methods and data
acquisition sources. The main driver of this research was to make use of integrating mathematical
modelling, GIS and remote sensing tools for surface water quality management in “data scarce”
irrigated watersheds. Data scarcity is the first uncertainty associated problem in water quality
modelling where very limited datasets were available but were not enough to conduct detailed
sensitivity analysis for all developed models. The used modelling tools are physically based models
and these models involve large number of variables modelling parameters and coefficients in both
hydrodynamic and water quality models which further contribute to uncertainty and error propagation.
The complexity of the lake water system and the numerous interacting eutrophication model
coefficients, variables, physical, chemical and biological processes are considered another important
source of modelling uncertainty. Remote sensing was used for data acquisition and for extraction and
prediction of water quality parameters that were used for calibration and verification of the
mathematical models. Remote sensing procedures and methods itself involves uncertainty that
includes temporal and spatial variability of input data, atmospheric correction accuracy, applied
models and algorithms, accuracy of extracted data and ground truth verification data. Uncertainty
analysis of the different modelling systems should be conducted for models fine-tuning and
verification.
Lack of continuous temporal measurements, neglected measurements of important eutrophication
indicators, lack of archived historical data of the lake system was a major constraint in the research
and in the mean time it was one of the driving forces to conduct the research based on the integration
of tools and data acquisition sources for calibration and verification of models. Although there is a
level of error propagation in the different linked modelling tools specially the eutrophication model,
the developed 2D hydrodynamic model, the basic water quality model and the screening
eutrophication models can be considered robust and reliable for their designated management
objectives since the calculated errors are within accepted ranges. Uncertainty analysis was not a in the
scope objectives of this research but it is a priority for the use of the system future water quality and
eutrohphication forecasting of the watershed lake system.