Presentation is loading. Please wait.

Presentation is loading. Please wait.

Fire in the vegetation and peatlands of Borneo, 1997-2007: Patterns, Drivers and Emissions Allan Spessa [1]*, Ulrich Weber [2], Andreas Langner [3], Florian.

Similar presentations


Presentation on theme: "Fire in the vegetation and peatlands of Borneo, 1997-2007: Patterns, Drivers and Emissions Allan Spessa [1]*, Ulrich Weber [2], Andreas Langner [3], Florian."— Presentation transcript:

1 Fire in the vegetation and peatlands of Borneo, : Patterns, Drivers and Emissions Allan Spessa [1]*, Ulrich Weber [2], Andreas Langner [3], Florian Siegert [4], Angelika Heil [5] The peatland forests of equatorial SE Asia covers over 20 Mha with most located in Indonesia. Indonesian peatlands are globally one of the largest near-surface reserves of terrestrial organic carbon, with peat deposits of up to 20m thick and an estimated carbon storage of Gt. The destructive fires in Indonesia during the exceptionally strong drought of burnt large swathes of peatland forests, and caused some of the largest emissions events in recorded history. Past studies estimate that ca. 1Gt of carbon was released to the atmosphere from the Indonesian fires in 1997– equivalent to 14% of the average global annual fossil fuel emissions released during the 1990s. Previous studies have established a non-linear negative correlation between fires and antecedent rainfall in Borneo, with ENSO-driven droughts being identified as the main cause of below-average rainfall events over the past decade or so. However, while these studies suggest that this non-linear relationship is mediated by ignitions associated with land cover change, they have not demonstrated it. A clear link between fires and land cover loss in Borneo has been reported, but was restricted to eastern Kalimantan in This study assesses the relationship between fires, emissions, rainfall and land cover change across Borneo using available fine resolution data over a 11 year period. Introduction Conclusions [1] National Centre for Atmospheric Science, Dept Meteorology, Reading University, UK, [2] Biogeochemical Model-Data Integration Group, Max-Planck Institute for Biogeochemistry, Jena, Germany, [ Dept Environmental Studies, University of Tokyo, Japan, [4] Remote Sensing Solutions GmbH, Munich, Germany, [5] Research Centre Juelich, Germany, // * Methods Burnt area data were obtained from the Remote Sensing Solutions-Borneo-Version 1 fire database (RSSv1), , monthly at 0.01 degs (Siegert et al). RSSv1 is based on calibration of AVHRR (ca. 1.2 km 2 ), ATSR and MODIS hotspots on a large sample of Landsat ETM fire scars (resolution ca. 30 m 2 ). Rainfall datasets i) 0.5 degs, and ii) 0.25 degs. The overlap period between TRMM and CRU was used to calibrate CRU to be used to extend TRMM data back to January Peat areal extent, depths and burn depths were taken from Page et al. (2002 Nature),Wetland International Peat Map (2007), and Balhorn et al. (2009 PNAS). Tree cover for 1997 and 2007 were derived from global AVHRR tree product and a 2005/2008 forest/non-forest map of Borneo, after Langner et al. (2009 GCB). Statistics are based on 0.5 o resolution. We fitted 12 different regression models to each pixel of the form logit(PAB) = beta1 × RF_lag_k, where k = 0 (present month’s rainfall), …, and k = 11 (average of last 12 months rainfall, including current month’s rainfall). Highly negative beta 1 values indicate a very strong negative correlation between burnt area and prior rainfall. Beta1 of lag k = 4 rainfall (B1_lag4) gave the best overall R 2 sq values. We then examined change in B1_lag4 as a polynomial function of 1997 forest cover and 2007 forest cover and their interaction term (R 2 = F 5, 276 = 78.2) (Fig 3). Kalimantan peat fires represent a serious emissions perturbation to regional and global climate. Equatorial SE Asia, including Borneo, has been forecast by climate models to experience reduced rainfall in future decades (2007 IPCC 4 th Assessment Report), which implies an increase in future fire activity and emissions. However, as this study shows, fire activity is driven by climate variability and land cover change. Hence, forecasts of fire in the region under climate change should also account for future land cover change. Future demand for establishing pulp paper and palm oil plantations to replace native rainforests (especially on peatlands where tenure conflicts among land owners tend to be minimal) is forecast to increase in Kalimantan and other parts of Indonesia esp. Sumatra and Iryan Jaya (Page et al FAO paper). This will presumably result in more forest cover loss. The joint scenario of drier climate and reduced native forest cover implies there will be even more fires and emissions in this region in future than otherwise suggested by climate predictions alone. Results Burning predominated in the Indonesian part of Borneo viz south-west, southern and eastern Kalimantan and this is associated with ENSO-induced drought events (Figs 1, 2). Despite experiencing similar drought occurrences, burning in other parts of the island is greatly reduced (Sarawak and Sabah) or negligible (central Borneo and coastal regions) (Figs 1, 2). Burnt area per pixel in each month is easily described as a negative exponential function of prior rainfall, and this relationship shows marked regional variability. The negative exponential relationship between fire activity and rainfall is generally stronger (as indicated by the magnitude of the slope of the function) in pixels that have undergone a major reduction in land cover between 1997 and 2007 (Fig 3). Such pixels occur overwhelmingly in south-west, southern and east Kalimantan, and are relatively scarce in Sarawak and Sabah, central Borneo and the coast, where by comparison, significantly little or no deforestation and fire was observed during (Fig 4). Previous work in southern Amazon forests demonstrates that recurrent fires promote a change from tree- dominated to grass-dominated ecosystems which, in turn, promotes even more fires (Cochrane 2003 Nature). We show that recurrent fire and deforestation are also linked as part of a similar positive feedback process in Kalimantan. Our results support the detailed field observations carried out in in East Kalimantan by Siegert et al. (2001 Nature), and reinforce these findings across the island and over more than a decade. Carbon emissions from peat, forest and non-forest fires, show that emission spikes reflect fire activity ( ), and that fires in the carbon-rich peats of Kalimantan dominate the emissions profile overall Borneo, particularly during El Nino years ( , 2002, 2004 and 2006) (Fig 5). These peats are principally found in south-west and southern Kalimantan according to the most up-to-date peat map available- Wetlands International Peat Map (2007) (Fig 1). Kalimantan peat fires comprise between ca % of total annual carbon emissions across Borneo. We calculate that across the decade , carbon emissions from Kalimantan peat fires in El Nino years can be as high as around 10% of annual average carbon emissions from fossil fuels during this period. Further work is needed to quantify the compounding effect of uncertainties associated with peat extent, peat depth, depth of burn, organic carbon content, forest cover, burnt area etc on emissions. Figure 2 Time series of monthly burnt area and rainfall for 6 regions, Figure 4 Percentage tree cover Mean annual proportion of pixel burnt Percentage tree cover 2007 Figure 1 Borneo, the six analysis regions and extent of peat as decribed in Wetlands International peat map (2007). Figure 5 Annual total carbon emissions from peat burning versus annual total area burnt for Borneo. Figure 3 Relationship between forest cover in 1997, forest cover in 2007 and B1_lag4 across Borneo. B1_lag4 is the slope of the change in monthly burnt area wrt average rainfall during previous 4 months- see Data & Methods). Increasing Burnt Area wrt prior Rainfall


Download ppt "Fire in the vegetation and peatlands of Borneo, 1997-2007: Patterns, Drivers and Emissions Allan Spessa [1]*, Ulrich Weber [2], Andreas Langner [3], Florian."

Similar presentations


Ads by Google