406
T. J. Ballatore et al.
water balance are strongly reflected in changes in lake extent. The strong contrast
between light reflected from land vs. water, along with the complete and fine spatial
coverage offered by sensors such as the Multi Spectral Scanner (MSS), Thematic
Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+), onboard the Landsat satellite series, mean that this technique is not limited to large lakes and can be
applied, in theory at least, to lakes of 1 km
2 or less.
Lake Chad has drawn much attention in the literature given its dramatic decrease
in extent. Mohler et al. (1989) digitized photographs taken with onboard, handheld
cameras during manned space flights (Gemini, Skylab, and the Space Shuttle) dating
from 1966 to 1985 to document an approximate 90 % decrease in lake extent. Two
main cameras were used: a 70 mm Hasselblad and a 230–460 mm Large Format
Camera (LFC) and, except for one black-and-white photo and one color-infrared
photo (both taken with the LFC), the images were captured on natural color film.
After digitization, spatial resolution was reported as 11 m for LFC and 29 m for the
Hasselblad. Water and land were classified with a maximum likelihood algorithm.
It is important to note that the 1966 Gemini image, which showed an extent of
22,772 km
2 , pre-dated the era of non-manned satellite observing systems which
began with Landsat MSS in 1972, a time when the lake had already undergone most
of the areal change. Wald (1990), building on the work of Mohler et al. (1989), used
images from Meteosat in the visible and NIR range (0.4–1.1 μm) to document the
lake’s further shrinkage to 500 km
2 in April 1989.
Schneider et al. (1985) used the NIR Band 7 (0.8–1.1 μm) from Landsat MSS
(spatial resolution ∼ 80 m) to document the lake area changes from 1972 to 1982.
They also used the visible and NIR Channels 1 and 2 (0.55–0.68 and 0.725–1.1 μm)
of the Advanced Very High Resolution Radiometer (AVHRR), onboard the NOAA7 satellite, (spatial resolution 1.1 km
2 ) to develop a simple Normalized Difference
Vegetation Index (NDVI) to detail open water and vegetation along transects of key
portions of the lake from November 1981 to November 1982. Despite its lower spatial
resolution, the higher temporal resolution of AVHRR (daily vs. 18 days for MSS)
was an advantage in mapping the rapidly changing water area and vegetation.
Birkett (2000) used the AVHRR’s NIR Channel 2 (0.725–1.1 μm) to map lake
extent (inundated area) over a 4-yr period starting in 1995. A simple histogram
technique of the Digital Number (DN) values allow classification of inundated areas
with an overall relative error of ∼ 5 % (calculated from results of different algorithms
given the lack of field data) with lower errors during peak flows and higher errors
during other times, due in part to the presence of pools smaller than the AVHRR
spatial resolution. The paper also made extensive use of satellite altimetry data.
Turada (2008) used a selection of visible and NIR bands from the Medium Resolution Imaging Spectrometer (MERIS), onboard the European Space Agency (ESA)
satellite Envisat, (300 m spatial resolution) to look at inundation and land cover
change in the Lake Chad basin through Spectral Mixture Analysis (endmembers:
soil, vegetation and water). The MERIS-based classification was validated with
data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer
(ASTER), a high resolution imaging instrument that is flying on the US National
Aeronautics and Space Administration (NASA) orbital platform named Terra, (15 m
T. J. Ballatore et al.
water balance are strongly reflected in changes in lake extent. The strong contrast
between light reflected from land vs. water, along with the complete and fine spatial
coverage offered by sensors such as the Multi Spectral Scanner (MSS), Thematic
Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+), onboard the Landsat satellite series, mean that this technique is not limited to large lakes and can be
applied, in theory at least, to lakes of 1 km
2 or less.
Lake Chad has drawn much attention in the literature given its dramatic decrease
in extent. Mohler et al. (1989) digitized photographs taken with onboard, handheld
cameras during manned space flights (Gemini, Skylab, and the Space Shuttle) dating
from 1966 to 1985 to document an approximate 90 % decrease in lake extent. Two
main cameras were used: a 70 mm Hasselblad and a 230–460 mm Large Format
Camera (LFC) and, except for one black-and-white photo and one color-infrared
photo (both taken with the LFC), the images were captured on natural color film.
After digitization, spatial resolution was reported as 11 m for LFC and 29 m for the
Hasselblad. Water and land were classified with a maximum likelihood algorithm.
It is important to note that the 1966 Gemini image, which showed an extent of
22,772 km
2 , pre-dated the era of non-manned satellite observing systems which
began with Landsat MSS in 1972, a time when the lake had already undergone most
of the areal change. Wald (1990), building on the work of Mohler et al. (1989), used
images from Meteosat in the visible and NIR range (0.4–1.1 μm) to document the
lake’s further shrinkage to 500 km
2 in April 1989.
Schneider et al. (1985) used the NIR Band 7 (0.8–1.1 μm) from Landsat MSS
(spatial resolution ∼ 80 m) to document the lake area changes from 1972 to 1982.
They also used the visible and NIR Channels 1 and 2 (0.55–0.68 and 0.725–1.1 μm)
of the Advanced Very High Resolution Radiometer (AVHRR), onboard the NOAA7 satellite, (spatial resolution 1.1 km
2 ) to develop a simple Normalized Difference
Vegetation Index (NDVI) to detail open water and vegetation along transects of key
portions of the lake from November 1981 to November 1982. Despite its lower spatial
resolution, the higher temporal resolution of AVHRR (daily vs. 18 days for MSS)
was an advantage in mapping the rapidly changing water area and vegetation.
Birkett (2000) used the AVHRR’s NIR Channel 2 (0.725–1.1 μm) to map lake
extent (inundated area) over a 4-yr period starting in 1995. A simple histogram
technique of the Digital Number (DN) values allow classification of inundated areas
with an overall relative error of ∼ 5 % (calculated from results of different algorithms
given the lack of field data) with lower errors during peak flows and higher errors
during other times, due in part to the presence of pools smaller than the AVHRR
spatial resolution. The paper also made extensive use of satellite altimetry data.
Turada (2008) used a selection of visible and NIR bands from the Medium Resolution Imaging Spectrometer (MERIS), onboard the European Space Agency (ESA)
satellite Envisat, (300 m spatial resolution) to look at inundation and land cover
change in the Lake Chad basin through Spectral Mixture Analysis (endmembers:
soil, vegetation and water). The MERIS-based classification was validated with
data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer
(ASTER), a high resolution imaging instrument that is flying on the US National
Aeronautics and Space Administration (NASA) orbital platform named Terra, (15 m