Skip to main content.

13 Integration of the LCDB with the CMS (Stephens et al. 2001)

The LCDB1 has been used to establish a 1996 baseline estimate of carbon in soils, in indigenous forest and in scrub. Error estimates associated with mapping the aerial extent of indigenous forest and scrub as well as carbon calculation have been determined.  A map depicting the distribution of national soil carbon and carbon in indigenous forest and scrub has been produced. The total carbon stored in indigenous forest and scrub for 1996 was estimated at the 95% confidence level to be 1149 (± 50) x 106 tonnes 153 (± 80) x 106 tonnes, respectively.  A high error was found to be associated with the estimation of carbon in scrub because a limited number of plots were used for the calculation.

It is expected that a second LCDB will be established over the next 18 months.  Since this second database (2001/02 LCDB2) would be backward compatible (with respect to the land cover classes mapped previously) with the first (1996) LCDB, it is important that the carbon monitoring system uses the LCDB1 and not the VCM as the source of land cover information for accounting purposes.  Several key issues associated with using the LCDB for operation carbon monitoring have been addressed, including: the number of scrub plots; mapping arable cultivation; a land-use activity that affects soil carbon; the over-mapping of bare ground; and the generalised LCDB vegetation classes.

For national carbon accounting purposes, a verifiable 1990 baseline will be required.  Such a baseline could be derived from the 1996 LCDB. A recommended method would use sampling of aerial photography and satellite imagery to quantify land-cover change between 1990 and 1996, and would combine these changes with the LCDB to get a 1990 estimation of land cover.  The 1990 land cover would then be combined with the GLM model of soil carbon and an equivalent regression model of indigenous forest and scrub carbon to produce a 1990-carbon budget. As the transition of scrub to indigenous forest is too subtle for detection by remote sensing, a simple growth model and scrub age distribution will also be required. Special attention will have to be given to using a confusion matrix of the LCDB to counteract random and systematic errors of the LCDB.

 

[ |