Coastal carbon opportunities: changes in the distribution of mangrove and saltmarsh across South Australia (1987-2015): Technical Report

Abstract

Mangrove and saltmarsh ecosystems play a vital role in healthy, functioning coastal systems and have the capacity to sequester large amounts of carbon, both through uptake of atmospheric carbon dioxide during photosynthesis and by trapping organic materials from sea and land based inputs. Because of their ability to provide a range of ecosystem services, not least carbon capture and storage, there is a need to better understand the distribution of saltmarsh and mangrove throughout the state, as well as how this may be changing in response to human activities (e.g. coastal development, illegal dumping, changing land use) and challenging environmental conditions (e.g. pollution, coastal squeeze, invasive species and climate change). As part of our Coastal Carbon Opportunities project, we aimed to assess spatial and temporal changes in the distribution of mangrove and saltmarsh vegetation throughout South Australia using historical and new imagery to look at changes in area coverage of these two coastal vegetation communities. The primary data set that we used to assess this was the South Australian Department for Environment and Water’s (DEW) SA Land Cover dataset, which was released in 2018. The SA Land Cover dataset is a modelled output, based on classification of spectral information from the Landsat satellite remote sensed imagery. The dataset covers the period from 1987 to 2015 and is provided as six composite epochs (1987 – 1990, 1990 – 1995, 1995 – 2000, 2000 – 2005, 2005 – 2010, 2010 – 2015), so can be used to assess change through time. Using the SA Land Cover dataset, we estimated area coverage of 164.2 km2 for mangrove and 197.6 km2 for tidally influenced saltmarshes in 2015. We found that there had been a net increase in the area of both saltmarsh and mangrove ecosystems between 1987 and 2015, with a greater increase in saltmarsh (16 km2, or an approximately 9% increase since 1987) than mangrove (7.9 km2 or a 5% increase since 1987). However, it should be noted that a broad scale loss of these coastal ecosystems is likely to have occurred prior to the commencement of the Landsat satellite data coverage in 1987 (particularly in urban, industrial and agricultural areas). Therefore, the relatively small increases in area coverage reported here for the period between 1987 – 2015 should be viewed with that in mind. There are some uncertainties around how well saltmarsh and mangrove are identified in the model-classified SA Land Cover dataset. For this reason, we carried out an external evaluation based on assessing change in the distribution and area coverage of mangrove and saltmarsh using manual digitisation of aerial photographs in two areas where changes had previously been reported in the literature (Torrens Island and Middle Beach, both on the east coast of Gulf St Vincent, north of Adelaide). We compared the results of the aerial photo change analysis with those based on the SA Land Cover dataset and found the area coverage estimates from the SA Land Cover dataset were 5% higher for mangrove and 25% lower for saltmarsh for the period 1987-2015. We propose that the aerial photograph based assessment is likely to be more representative of changes at a scale that is relevant to local and regional authority management activities. However, the SA Land Cover dataset is the best available state-wide mapping product to use for baseline carbon stock assessment and for the identification of broad scale gains and losses in saltmarsh and mangrove ecosystems. We therefore suggest a multi scale approach, which involves further local-scale external validation of the SA Land Cover dataset’s classification of saltmarsh and mangrove (e.g. for other sites across the state using alternative data sources and comparing patterns of change). We believe this is a practical way forward, especially when relating the area and changes in the distribution of these ecosystems to carbon stocks and accumulation, as the results will improve accuracy at state scale and be more robust at validated sites. If the SA Land Cover dataset is found to be consistently unreliable in the mapping of saltmarsh and mangrove (after further external validation), a dedicated mapping and modelling program for these difficult-to-classify coastal vegetation types may be the best way forward.