532
5.1.6 Satellite Remote Sensing
Satellites provide routine digital reflectance at several wavelengths which can be
trained to detect and monitor salt-affected land (McFarlane et al. 2004 ; Spies and
Woodgate 2005 ). Usually it is the poor condition of vegetation that is detected and
this can result from multiple causes so other data sets are required to reduce the
errors of commission (false positives) and omission (failing to detect salt-affected
land).
Saline land has low or no productivity over long periods of time so analysing
multiple images can ensure that temporarily bare areas are not confused with saline
areas. Dryland salinity caused by high groundwater levels are the most common in
low lying topographic areas and in where water accumulates in the landscape. These
attributes have been used to increase correct classification levels to more than 95 %
once image-training areas that are known to be salt affected (often by aerial photo-
graph interpretation) are used to train the classification. These methods are repro-
ducible over large areas at low costs (a few cents per hectare) and have been applied
to map and monitor changes in dryland salinity of areas of about 18 M ha (Caccetta
et al. 2009 ).
Expert hydrogeologists estimated the area that was salt-affected in south-western
Australia to be 1.8 M ha based on studies of selected catchments and monitoring of
observation bores over several decades (Ferdowsian et al. 1996 ). However, a more
repeatable, accurate and comprehensive mapping of saline areas using multi-
temporal Landsat Thematic Mapper (TM) images and digital elevation models
showed that its extent was less than 1 M ha (McFarlane et al. 2004 ). The expert
method was also affected by site selection bias where estimates were based on stud-
ies of saline catchments which may over-estimate salinity in other catchments.
5.1.7 Most Feasible Methods on Monitoring Changes in Salinity
Over Time
Methods that are quantitative, reproducible and relatively cheap are more likely to
be adopted and to provide accurate assessments of the change in salt-affected areas
over time. These are important for assessing the effectiveness of management inter-
ventions relative to the impact of climate variability, especially rainfall.
Monitoring changes in groundwater level or ground-based geophysical surveys
or transects can be used at the local level but for large areas satellite imagery prob-
ably can give the most cost effective regional coverage. Such monitoring can be
used to assess the condition of perennial vegetation and crop yields as an additional
benefit of producing interpreted images.
D.J. McFarlane et al.