Mapping and monitoring land degradation in southern New Mexico using Landsat data
This research aims to determine the effectiveness of satellite remote sensing, in particular
the technique of linear mixture modelling, in mapping and monitoring land degradation
in the Jornada Basin, Southern New Mexico, USA.
Land degradation in the Jornada area is characterised by shrub encroachment,
deterioration of soil resources (i. e. increased erosion, changes in soil texture, and
decreases in soil nutrient status and water-holding capacity) and often a reduction in
In this research, linear mixture modelling, regression models, and spectral vegetation
indices are applied to Landsat TM imagery in order to assess their utility for mapping
and monitoring land degradation. Linear mixture modelling has been used to estimate the
proportions of green vegetation, dry vegetation, shade, and soils. The results of this study
indicate that mixture modelling is a reasonably accurate technique to measure these
materials. The correlation of the field-measured and model-estimated green vegetation,
dry vegetation, and total vegetation proportions are 0.93,0.86, and 0.93 respectively.
Moreover, the results of mixture modelling improve when the shade element is removed.
In contrast, many of the correlations between vegetation indices and the various
vegetation parameters are not significant or result in low R2 values. Indeed many of
SVIs' R2 values are even lower than those provided by regressions of the field data to
individual TM spectral bands.
The application of mixture modelling to multiple-date imagery suggest that mixture
modelling can identify successfully the patterns and the extent of extreme change and
thus shows potential for monitoring of rangeland resources. Furthermore, the maps of
vegetation and soil types provided by mixture modelling have been manipulated to
estimate the shrub to grass ratio, a soil degradation index, and an index that combines
both these indicators of land degradation. These indices have been found to be more
sensitive to change than any of the individual mixture maps.