The utility of NOAA AVHRR data for vegetation studies in semi-arid regions : a minor field study in the Hoanib Catchment of Namibia

Abstract: The relationship between field measurements of vegetation in a semi arid region and
NDVI from the 1 km NOAA AVHRR was investigated within this study. The Hoanib
catchment in the northwestern Namibia was chosen as study area. The field
measurements were carried out during the month of February 1999. Different NDVI
values and combinations of NDVI were used and it was found that the integrated
NDVI until the time of the field measurements correlated best to the grass vegetation.
No significant correlations were found with the woody vegetation.
The utility of non-destructive measurements of vegetation was also examined. Since
destructive measurements of woody vegetation are both time-consuming and harmful
to the environment, two different non-destructive methods were compared. It was
found that the result of the two methods differed considerably for species and growth
forms that were not recorded in the areas where the equations were established. The
relationship between NDVI and grass measured with destructive or non-destructive
methods was also compared. It was found that a non-destructive method returned
almost as high correlation values as the destructive method.
The relationship between NDVI and vegetation for all plots (with a large precipitation
gradient) were compared to the relationship for the most easterly plots (with a small
precipitation gradient) to draw attention to the fact that a trend in the data set
improves the correlation values. This was also the case in the study.
Finally, a Monte Carlo simulation was used to study the sensitivity of the calculated
net primary production, using the Monteith equation, to variations in the e, a and b
parameters. It was found that the NPP was very sensitive to variations in e but less
sensitive to variations in a and b.