Digital Signal Processing Reference
In-Depth Information
8.1.
INTRODUCTION
An important reason for integrating remote sensing into the water quality management information
system is the scarcity of field monitoring measurements due to the absence of water monitoring
stations in the Edko lake area. To overcome the lack of calibration data, remote sensing provides an
alternative means of water quality monitoring for a range of temporal and spatial scales (Yang et al
2000).Therefore this chapter focuses on the development of different procedures based on remote
sensing techniques for the validation of the hydrodynamic model and the calibration and enhancement
of the water quality mathematical model developed for Lake Edko. The use of remote sensing data
driven from both satellite images and spectral in-situ measurements can assist in developing spatial
and temporal data sets that can be directly used for model calibration and enhancement. In this
research, the complexity of the physical and ecological properties of the lake system was sufficient
reason to explore different methodologies and procedures for model calibration using remote sensing.
Here, the role of remote sensing is considered of great importance in order to compensate for missing
in situ measurements.
The calibration methodologies used are categorized as qualitative and quantitative. The different
techniques of remote sensing for estimating water quality parameters have their own limiting factors.
In this study calibration was based on two parameters: total suspended sediments and CHL-aorophyll-
a, Many satellite sensors are potentially suitable for estimating the concentration of these parameters.
The basis for sensor comparison and selection is the spectral, spatial and temporal resolution, in
addition to factors related to the quality of selected images and cloud coverage, which may affect the
selection of images. As mentioned in
Chapter 5
, two types of sensor data are used for the calibration
process: the Moderate-resolution Imaging Spectroradiometer (MODIS) and the SPOT-5,
characteristics of these data sets are given in Annex (A1).
8.2.
DESCRIPTION OF THE APPLIED REMOTE SENSING METHODS
Comprehensive research on the quantification of water quality parameters reflectance spectra began in
the early 1970's. Remote sensing of freshwaters can basically be done through two different
approaches or through a combination of the two: The empirical or (statistical approach) and the
analytical or bio-optical modeling approach (Gordon and Morel, 1983).
8.2.1.
Empirical (Statistical) Approach
The empirical approach is based on the calculation of a statistical relation between the water
constituent concentrations and the radiance measured by the sensor. Normally in the empirical
approach, remote sensing data is related by regression analysis to
the lake in-situ
measurements of
water quality parameters. This approach needs extensive field work and logistics since samples have
to be collected from the water body simultaneously or near simultaneously with the overpass of the
sensor, which in practice is very difficult to achieve. A review of the literature on empirical algorithms
for estimating water quality parameters shows a vast variety of algorithms proposed. They start from a
simple linear regression between reflectance and water constituent concentrations to non-linear
multiple regressions between a combination of band ratio(s) and the concentrations. The advantage of
using the empirical approach is that the algorithms are straightforward and easy to use. The
disadvantages are that false results may occur while using this method, because a causal relationship
does not necessarily exist between the parameters studied (Hogenboom and Dekker, 1999).
Search WWH ::
Custom Search