Projet de recherche S0/02/030 (Action de recherche S0)
Context and objectives
Even though land degradation is considered a major problem in the arid and semi-arid grasslands in Australia, and more research is devoted to sustainable development of these pastures, little is known on the actual extent of land degradation. From many different regions under grazing is known that they are partially degraded, but until now there is limited success in determining whether the negative effects are permanent or not. As many pastures show a non-equilibrium behavior land degradation is hard to quantify. Under a non-equilibrium behavior the short term rainfall variability causes extremely high changes in the vegetation. This is masking the downward (or upward) trend in the land condition except for the most extreme situations, and thus makes it difficult to determine whether the trend is persisting, stabilizing or improving. Moreover, due to the strong seasonal fluctuations it is not obvious whether the degradation is caused by grazing or by natural factors. Finally determining land degradation is even more compounded by the spatial variability of pasture quality and the non-uniform pattern of cattle grazing.
This project aims at evaluating and exploring the possibility of satellite images for monitoring past and present pasture condition on a property situated in the Mithchell grasslands bioregion in Australia’s Northern Territory. In first instance, the vegetation change over the last fifteen years was to be determined, while in a second step the composition of the grassland, this being the main productive vegetation type, had to be analysed.
Methodology
Both low and high resolution images are used, in combination with field measurements to support the interpretation. The high resolution images were used to classify the study area and as a basis for change detection, investigating the evolution of the vegetation since the purchase of the property by CPC. The current paddock condition was assessed quantitatively using the grazing gradient method
Since grassland composition cannot be deducted from single spectral observations, different methods were tested to separate annual and perennial grasses. The spectral behaviour throughout time was analysed based on a long term time series of NOAA AVHRR imagery. The data was upgraded through a two step compositing algorithm and using a Savitsky-Golay post-composting filtering algorithm. Different seasonal parameters for benchmark areas of annual and perennial sites were extracted. This was followed by a more in depth analysis using multi-resolution wavelet analysis.
Results
In general there is little change over the past 25 years on Newcastle Water Station. The biggest changes occurred between 1980 and 1985 where some encroachment of hummock grasslands and grasslands took place. Other observed changes were mainly related to differences in the total amount of rainfall. These were most obvious around the wetland area of Lake Woods and in the swampy areas of the station. Fire in 2004 burned parts of the Eucalypt with hummock understorey and the Acacia with grassland understorey. The grazing gradient calculated on the satellite imagery generally fitted with the ground measurements. Radial gradients were detected in a number of paddocks that have been heavily utilised in the past. A less strong gradient was observed in the cell paddocks, probably because the cell grazing management systems inherently causes a more uniform grazing pattern and thus also reduces the grazing gradient effect. These findings were also reflected in the corresponding Cover Percentage Loss values. The Studs paddocks were not used as heavily as the former paddocks. This is also reflected in the lack of grazing gradients within these paddocks.
From the processed NOAA AVHRR composites, time-series for typical annual and perennial dominated grassland sites were derived. Subtle differences between NDVI time-series of annual and perennial grasses were detected. Results obtained with both methods confirmed former findings on the higher variability of annual grass responses as well on the short term, (inter-seasonal) as on the long term (intra-seasonal). However the observed differences were small, and were partly masked by spatial and seasonal variability of rainfall. Next to rainfall, fire and management strategies also have a strong impact on the vegetation development in the growing season(s).
Products and services
Updated vegation cover map of the study area
Change detection procedure for the station
Basis for a permanent vegetation composition monitoring tool
