290
J.C. Ritchie and F.R. Schiebe
Continuing development of new sensors with higher spatial and spectral (hyperspectral) resolution will improve our capability of measuring both the quality and
quantity of the radiance thus improving our ability to measure and quantify water
parameters. Middleton and Marcell (1983) provided a literature review of early
remote sensing studies of water quality. Gordon et al. (1983) discuss applications of
these techniques to ocean and estuarine waters. Kirk (1983) discusses the applications
of these techniques to freshwater systems. Dekker et al. (1995) have provided a
current overview of remote sensing of water quality in freshwater systems for
suspended sediments and chlorophyll. Fingas et al. (1996) provided an overview of
remote sensing of oils.
13.3 Application
A combination of ground (water) and remote sensing measurements are required to
collect the data necessary to develop and calibrate empirical and semi-empirical
models and to validate more physically based models. Water samples analyzed for the
substance of interest (i.e., suspended sediment, chlorophyll) should be collected at the
same time (or on the same day) that the remote sensing data is collected. Water
systems are very dynamic and rapidly change so that the substances in the water are
continuously changing. Certain water properties can be measured in situ while for
other properties, water samples have to be collected and analyzed in a laboratory
using standard techniques.
Location of sample sites should be determined with GPS (or other available technique) so that the correct data (pixel information) can be extracted from the remote
sensing data for comparison. Often 3 or 5-pixel arrays are averaged to obtain the
remote sensing data to account for the dynamic nature of the water body. Remote
sensing data must be converted to radiance or reflectance data (Ritchie et al. 1988)
if the algorithms developed are to be applicable to other conditions. Data from
satellite sensors and high altitude aircraft sensors should be corrected for atmospheric
interference. Several atmospheric correction models are available to make corrections
for atmospheric interference. Many water quality studies have used the dark pixel
technique to correct for atmospheric interference on the assumption that the dark pixel
value in a scene is due to atmospheric interference. Care should be used in the
application of the dark pixel technique in water quality studies since most often the
dark pixel is from a water body thus leading to overcorrection since all waters have
some reflectance. Once a data set is collected, empirical algorithms can be developed
by applying standard regression techniques to the data or the data can be used to
validate analytically based models.
Since the radiance/reflectance measured is predominately from the surface water,
the algorithms developed are good for estimating substances in the surface water.
Information about total load or depth distribution can only be derived by modelling
(knowing) the relationship between surface substance distribution and total distribution of the substance in the water column.
J.C. Ritchie and F.R. Schiebe
Continuing development of new sensors with higher spatial and spectral (hyperspectral) resolution will improve our capability of measuring both the quality and
quantity of the radiance thus improving our ability to measure and quantify water
parameters. Middleton and Marcell (1983) provided a literature review of early
remote sensing studies of water quality. Gordon et al. (1983) discuss applications of
these techniques to ocean and estuarine waters. Kirk (1983) discusses the applications
of these techniques to freshwater systems. Dekker et al. (1995) have provided a
current overview of remote sensing of water quality in freshwater systems for
suspended sediments and chlorophyll. Fingas et al. (1996) provided an overview of
remote sensing of oils.
13.3 Application
A combination of ground (water) and remote sensing measurements are required to
collect the data necessary to develop and calibrate empirical and semi-empirical
models and to validate more physically based models. Water samples analyzed for the
substance of interest (i.e., suspended sediment, chlorophyll) should be collected at the
same time (or on the same day) that the remote sensing data is collected. Water
systems are very dynamic and rapidly change so that the substances in the water are
continuously changing. Certain water properties can be measured in situ while for
other properties, water samples have to be collected and analyzed in a laboratory
using standard techniques.
Location of sample sites should be determined with GPS (or other available technique) so that the correct data (pixel information) can be extracted from the remote
sensing data for comparison. Often 3 or 5-pixel arrays are averaged to obtain the
remote sensing data to account for the dynamic nature of the water body. Remote
sensing data must be converted to radiance or reflectance data (Ritchie et al. 1988)
if the algorithms developed are to be applicable to other conditions. Data from
satellite sensors and high altitude aircraft sensors should be corrected for atmospheric
interference. Several atmospheric correction models are available to make corrections
for atmospheric interference. Many water quality studies have used the dark pixel
technique to correct for atmospheric interference on the assumption that the dark pixel
value in a scene is due to atmospheric interference. Care should be used in the
application of the dark pixel technique in water quality studies since most often the
dark pixel is from a water body thus leading to overcorrection since all waters have
some reflectance. Once a data set is collected, empirical algorithms can be developed
by applying standard regression techniques to the data or the data can be used to
validate analytically based models.
Since the radiance/reflectance measured is predominately from the surface water,
the algorithms developed are good for estimating substances in the surface water.
Information about total load or depth distribution can only be derived by modelling
(knowing) the relationship between surface substance distribution and total distribution of the substance in the water column.