Seasonal Cover Disturbance Index and Image (SCDI)
Background
The Seasonal Cover Disturbance Index/Image assists in identifying areas of grassland modification. The importance of this product has become increasingly evident since protection of native grasslands became recognised within State vegetation legislation and subsequently the need for reliable methods of detection.
The process which Tim Danaher, Catherine James and Dr Jim Watson have developed is based on the concept that cultivated or modified pastures can be distinguished from other non woody land cover types by identifying how the composition of green and non-green vegetation cover changes over time.
The SCDI products do this by identifying the pattern of change in vegetation cover over multiple seasons and across multiple years. For example, land that has been cultivated will show much greater variability in the amount of green and non-green vegetation cover on the land, compared to a lower level of variability in vegetation cover that native grasslands or other native vegetation typically display.
The SCDI products are designed to be used in conjunction with aerial imagery, SPOT and other satellite imagery to guide interpretation.
How
Landsat derived fractional cover seasonal composite images spanning seasons from summer 1988 to winter 2017 were used to identify areas that have experienced change in, or disturbance of, non-woody vegetation.
The algorithm for the SCD images is based on the following observed trends:
cultivated areas show significantly more variation in the level of cover and the relative proportions of green and non-green vegetation cover due to the cropping cycle
pasture areas, in particular native pastures, are relatively stable over time, with a higher proportion of non-green vegetation cover at most times and less fluctuation in the level of green cover, when compared to cultivated areas
modified pastures generally have a greater proportion of green cover when compared to native pastures.
To highlight the variation of these different cover types in the green/non-green vegetation cover space, these cover values were transformed into proportional measures of cover and several time-series statistics were calculated.
Three of the statistical images were combined to create the colour composite image known as the seasonal cover disturbance image (SCDImage). SCDImage products were developed for 4 time periods. One is based on the entire 1988–2017 time series and there is a set of three images, each covering approximately 8 years of the full time period.
The SCDIndex was developed later in the mapping program after developing a set of training data based on high-resolution satellite imagery, visual interpretation of time-series Landsat imagery covering 1987–2017, and roadside survey observations. The index was developed using this training data and more time-series images than used in the SCDImage. It classifies disturbance into a 0–1 range that can be displayed as a number of disturbance classes using a support vector regression (SVR) method. This approach makes the interpretation of disturbance less subjective, but the SCDIndex was still used in a multiple lines of evidence approach with other data sources.
Products
Regulatory environmental indicators
Catchment management
Vegetation models
An input to vegetation condition models
Where to next
We plan to expand the training / validation data set to provide an estimate of uncertainty for the SCDI index. The possible use of the SCDI for monitoring change in disturbance over time will also be examined.
Acknowledgements
NSW Government - Department of Planning and Environment
To find out more
Tim Danaher: Tim.Danaher@environemnt@nsw.gov.au
Jim Watson: Jim.Watson@enviornment.nsw.gov.au
Catherine James: Catherine.james@uq.edu.au
Links
Native vegetation regulatory map method statement
version 1.0
