Skip to the content | Change text size

Project Description

Project Title

Impact of Fire on Surface Heat and Moisture Fluxes in Australian Tropical Savanna and Feedbacks to Regional Climate


Fire is probably the greatest natural and anthropogenic environmental disturbance in Australian tropical savannas, with vast tracts burnt each year, particularly by pastoralists, Aboriginal land holders and conservation managers (Russell-Smith et al., 2000). For example, in the relatively mild fire year of 1992 74,000 km 2 (5.5% of the total land area) of the Northern Territory was burnt (Beringer et al ., 1995), by far the largest proportion being savanna landscape. This "poor" fire year consumed an estimated 29.5 x 10 6 tonnes of biomass, and was associated with a likely release of more than 13 Tg of carbon products to the atmosphere (Beringer et al ., 1995). Russell-Smith et al. (2000) provide an estimate of 244,000 km 2 and 242,000 km 2 for the total area of northern Australia (at least partially) burnt in 1997 and 1998, respectively. Hao et al. (1990) calculated that 40 to 75% of global savannas were burned annually between 1975 and 1980. Globally these savanna ecosystems account for 11.5% of the global landscape (Scholes and Hall 1996); burning them consumes three times as much dry matter per year as the burning of tropical forests (Levine et al. 1995), accounting for more than 40% of biomass burned globally (Hao and Ward 1993). The figures for Australia assume even greater significance given predictions of increases of up to 30% in seasonal cumulative fire danger (and a likely associated increase in fire activity) in parts of northern Australia under predicted climate change due to increased greenhouse gas emissions (Tapper, 2000; Williams et al., 2001).

Landscape fires that occur on the scale described above would have massive impacts on the regional water, energy and carbon dioxide exchanges and as a result are likely to have important feedbacks to the atmosphere and regional climate. In addition, different burning regimes can result in either a stable or dynamic change in vegetation composition and structure, which in turn can feedback to affect the atmosphere through changes in land surface properties. Most of these impacts are not well understood and an improved understanding of Australian savanna fires and their feedbacks to surface processes and local and regional climate is the major goal of this project. Some savanna fire-climate feedbacks are likely to occur through the direct radiative impacts of gaseous and aerosol emissions on the regional and global atmosphere, and the indirect role of emitted aerosol particles on cloud convection and precipitation processes (Cachier et al. 1991). It is emphasised that these are not research questions being directly addressed in this proposal.

Of significance for this study, there are other important impacts of fire on interactions between the land surface and the atmosphere. Fire is likely to radically alter the surface energy budgets of tropical savanna lands through reduced surface albedo, increased available energy for partitioning into the convective fluxes, and increased substrate heat flux into the soil. In addition, the aerodynamic and biological properties of the ecosystem can change, affecting surface-atmosphere coupling. Such direct fire impacts have scarcely been addressed or quantified around the world and never in the Australian environment. Scholes and Walker (1993) reported a halving of African savanna albedo to 0.06, immediately following fire, with a recovery to unburnt values after six weeks. Bremer and Ham (1999) observed similar impacts on the albedo of tall grass prairie. They showed increased available energy at the surface after fire, along with a three-fold increase in soil heat flux and reduced Bowen ratio (a relative increase in latent over sensible heat fluxes). In contrast, the only available data for Australian tropical savanna (Hutley et al., unpub.) showed a complete cessation of evapotranspiration in burnt savanna following a mid-dry season fire, with almost complete recovery within 60 days. This work implies a dramatic increase in sensible heat fluxes to the atmosphere, immediately post-fire. The long-term recovery of savannas following fire is also not well understood. Changes in leaf area index, transpiration, albedo, and vegetation composition are likely to alter the surface energy balance with subsequent consequences for land-atmosphere interactions. These parameters need further investigation over burns of different ages to understand the long-term response of savannas to burning.

Another vital factor in the savanna landscape is the fire regime imposed by different land use. The frequency, timing and intensity of burning all have profound influence in determining the biome type in a region. Decades of fire suppression can alter the patch mosaic structure of natural landscapes (Baker 1994). This may arise as a result of chronic cattle grazing which can suppress fire and lead to encroachment by woody plants over several decades (Archer et al. 1995). Conversely, Russell-Smith et al. (1997) mapped fire frequencies in Kakadu National Park in Australia over the period 1980-94, based on a manual interpretation of Landsat MSS imagery. They found savanna vegetation could tolerate burning at 3-year intervals without a loss of current floristic diversity (Russell-Smith et al. 1998). However, frequent burning of catchments favours the development of understoreys dominated by the high water using Sorghum spp, which may result in a reduction in water yields relative to less frequently burnt controls. Recently, Hutley et al. (2000) have shown that the grass-dominated understorey is responsible for up to 75% of the total water and carbon fluxes during the wet season from tropical savanna ecosystems. Thus shifts in fire regimes which favour higher water using grasses (native or exotic, such as Gamba grass), could have hydrological impacts such as reduced deep drainage and recharge and could impact negatively on the flow dynamics of ephemeral streams, with significant reductions in riparian health. Zeng et al. (2000) recently discussed the role of vegetation-climate interactions in shaping African savanna.

The changes to surface energetics following fire that were indicated earlier would likely have significant impacts on local to regional scale atmospheric circulations and climate. At the local scale, depending on the aerodynamic changes to savanna vegetation following fire, enhanced sensible fluxes over patches of burnt landscape (on the order of 10 km in diameter) would be likely to produce localised areas of convergence and divergence and associated mesoscale circulation systems (Knowles, 1993). These circulation patterns may lead to an increase in spatially fixed convective cloud development and precipitation that could in turn influence the re-establishment of the burnt savanna. Tapper (1991) and Physick and Tapper (1990) have shown that landscape albedo differences elsewhere in Australia can produce quite strong mesoscale circulation systems. Such circulations are known to be quite capable of producing intense, spatially fixed cloud convection and precipitation under suitable environmental conditions (Keenan et al ., 2000, Beringer et al. , 2001a). The timing of the fire is important in the context of this study. Given the airflow patterns and atmospheric stability of the dry season (May to September) over northern Australia, the probability of generation of rain-producing cloud is low at this time and the likely impact of fires on precipitation is small. Similarly, the fire regime of northern Australia is distinctly seasonal (Tapper et al. , 1993), with early dry season (May-June) fires being generally of low intensity (approximately 1.7 MW m -1 ). The tree canopy of the savanna recovers remarkable quickly from such fires, lessening the likely radiative impacts of burning. However by the late dry season and pre-monsoonal period (August to October), fire intensity can be an order of magnitude greater (Williams et al. 1997). These fires tend to burn over very large fronts, are more damaging with crown scorch of > 80%, and occur during periods when pre-monsoonal cloud-formation is significant. At this time, significant modification of precipitation patterns might occur. Over the last 50 years or so, there has been a strong shift in the fire regime in northern Australia (Braithwaite, 1991), from an Aboriginal regime characterised by early dry season burning, to these more destructive late dry season fires (Williams et al. 1997).

In addition, such fire-induced landscape changes when integrated over the total area of northern Australia might also have impacts on regional scale climate. Recent work by Beringer et al. (2001b) suggests that differential heating resulting from quite modest albedo differences between boreal forest and tundra may be responsible for persistent large-scale circulation features in the Arctic. It is hypothesised here that burning of the scale seen in northern Australian savanna could have similar impacts, perhaps even extending to influences on the strength and southward penetration of the Australian monsoon. In a recent article in Science Johnson et al. (1999) argue that changes in vegetation across northern Australia resulting from human burning practices over the last 60,000 years reduced the effectiveness of the summer monsoon for central Australia. Equivalently, current burning practices along with potential changes in fire frequency as a result of climate change (Williams et al. 2001) could again cause widespread changes in vegetation with feedbacks to regional climate. The outcome of such a change in fire regimes under different climate/vegetation scenarios is of great importance to land managers in the region and is a critical part of this project. In savanna landscapes, an adaptive management strategy is essential to ensure appropriate and timely management decisions are made in the context of a highly variable environment (Parr and Brockett 1999, Richards et al. 1999).

The important research questions touched on above can be framed in a set of research hypotheses which we will investigate using available theory, observational and modelling techniques.

Hypothesis 1: Fire causes a threshold change in savanna energy exchange with the atmosphere.

Immediately following dry season fires, there is a measurable decrease in savanna albedo, a decrease in evapotranspiration and increase in soil heat and sensible heat fluxes, followed by an extended recovery phase.

Hypothesis 2: Alterations in savanna energy exchange with the atmosphere following fire have important local-to-regional scale impacts on atmospheric circulation and water balance.

Changes in atmospheric heating rates above burnt and unburnt savanna and associated horizontal pressure gradients will produce atmospheric motion at a range of scales. In addition, local-to-regional scale circulation changes associated with burning will modify patterns of precipitation and affect the strength of the Australian monsoon. Changes in evapotranspiration following fire will also impact on soil moisture at equivalent scales

Hypothesis 3: Future climate change will alter the north Australian fire regime, with consequent feedbacks to vegetation and climate.

The climate associated with a projected doubling of CO2 will lead to a more extreme fire regime for northern Australia and a likely increase in area burnt each fire season. This will lead to dynamic changes in vegetation and climate in a series of complicated feedbacks.

In this study we will use contemporary theoretical understandings, observations and modelling approaches to investigate how current burning practices impact the atmosphere. Additionally we will investigate how current and future fire regimes and management practices might alter vegetation type, area burned and impacts on the atmosphere. A combination of modelling approaches will be well suited to addressing these questions. A transient vegetation model incorporating the effects of fire would enable a number of scenarios to be defined. These scenarios would then be used in a mesoscale numerical model to investigate land-atmosphere feedbacks to provide increased understanding for scientists and guidance for land managers.

Significance and Innovation

Tropical savanna covers a large proportion of the Australian landscape and is important for its economic, conservation and cultural value. Through the processes of evapotranspiration, photosynthesis, respiration and biomass burning savanna plays a highly significant role in regional and continental cycles of water and carbon. Savanna is fire-dominated, indeed fire-dependant, and yet an understanding of direct feedbacks of savanna fires to the atmosphere is virtually non-existent.

This study is a unique opportunity to develop, virtually from a non-existent base, a scientific understanding of the impact of savanna fire on the local-to-regional scale climate. Hence the work is entirely innovative and draws on theoretical, observational and modelling approaches. Such understanding will be an essential part of the management of Australia's tropical savanna lands into the future. In addition to direct observation and modelling of surface fluxes and atmospheric motion, we will also simulate processes such as the ignition and spread of fire across a heterogeneous landscape, under a variable climate. This will enable us to fully explore the long-term consequences of alternative management strategies on climate and vegetation in this critical biome.


The overall objective of our proposed research is to develop a predictive understanding of the major feedbacks between Australian tropical savanna fires and climate as a basis for future management of this important landscape. Direct measurements of radiation, heat and moisture fluxes and wind fields will be made over burnt and unburnt savanna. The surface flux observations along with other relevant meteorological data will be incorporated into a flexible mesoscale numerical model to quantify impacts of fire-altered energy, water vapour and momentum exchange on the regional hydrology and local and regional climate. A transient vegetation model that incorporates fire explicitly will be used to examine changes in biomes under likely and future fire regimes and under varying climate change scenarios. Simulations will be used in the mesoscale model to examine the effects of long term changes in fire regimes on climate.

Specifically, the approach to Hypothesis 1 will be as follows. Two years of field observations will be employed. In Year 1 Dr Lindsay Hutley and the other CIs will select an appropriate savanna site within 150 km of Darwin with the active assistance of the Bushfires Council of the Northern Territory. Two 20 metre instrumented towers will be used in the field observations. Campbell Scientific (CSI) eddy correlation (3-D sonic) along with full radiation budget sensors will be mounted above the savanna vegetation on each of the towers to evaluate fire impacts. Measurements will include up- and down welling short and long wave radiation, along with sensible, latent and soil heat fluxes to/from the surface. The turbulent fluctuations of H 2 O and CO 2 at greater than 10 Hz will be measured with a LiCor open path infra-red gas analyser. The extra effort to obtain measurements of net CO 2 exchange from the burnt and unburnt savanna would be trivial and would provide an excellent resource for a separate research project or for other researchers. For example we would provide the data to researchers involved in the North Australian Tropical Transect (NATT) and OZFLUX initiatives (Egan and Williams, 1996; Leuning et al , 2001). A full suite of standard meteorological (including wind speed and direction) and soil moisture measurements (CSI MetOne stations) would also be maintained alongside each tower. Initially, as a check of instrument bias, both towers will be established for one month in the middle of the dry season at the same reference location. One mobile tower will then be moved to a nearby location earmarked for burning where the paired measurements would continue for one further month. The instrument array would then be removed and the site burnt, with the instruments being re-established immediately afterward. Observations would then continue well into the subsequent wet season to observe post-fire recovery. In Year 2 the emphasis would be on undertaking a similar set of measurements over a chronosequence of savanna fire scars. Ideally as many as five fire scars of appropriate dimension ranging in age from freshly burnt to ~10 years old would be co-located around an area of unburnt savanna. The control measurements would be made continuously through the mid- to late dry season over the unburnt savanna, with the mobile tower being moved through the chronosequence. Such approaches have been successfully used in examining Arctic land atmosphere interactions (Eugster et al ., 1997). This approach may have to be applied flexibly depending on the availability of appropriate chronosequences. During this time regular calibrations of sensors will be performed. Raw eddy covariance data will be collected at greater than 10Hz and stored via laptop PC. Time series analysis will be performed using the RAMF processing software (Flinders Institute for Atmospheric and Marine Sciences - Flinders University). Spectral analysis will be conducted for each site for quality control and corrections will be performed on closed path H 2 O and CO 2 data. Processed data will be the 30-minute fluxes, turbulence parameters and meteorological data.

Hypothesis 2 will be approached through a combination of observation and modelling. Observable differences in boundary layer heating rates and structure (if any) resulting from differential heating over burnt and unburnt savanna will be determined using two identical tethersonde systems owned by Monash University and the University of Canterbury. These systems will allow simultaneous measurements of wind speed, direction, temperature, humidity and pressure to heights of up to 2000 metres. These systems were successfully used in a similar study of differential boundary layer heating over Arctic tundra and spruce forest (Beringer et al. , 2001b). Over a period of one month, spanning the late dry season to the monsoon transition period (i.e. September - October), approximately 15 days of paired observations made at 2-3 hourly intervals would be obtained over the burnt and unburnt savanna landscapes. Observations would be made over at least 18 hours to capture most of the diurnal signal from pre-dawn until about midnight. If environmental conditions became suitable, further measurements to determine the possible presence of a mesoscale circulation would be made through the establishment of one system on the boundary between burnt and unburnt savanna and the other over unburnt savanna. The paired burnt-unburnt atmospheric profiles will be examined for any evidence of differential heating rates and for any differences in maximum boundary layer development. These differences will also be related to observed differences in sensible heat inputs from the surface. It is possible that the Monash/Canterbury systems could be augmented with profiles from an innovative atmospheric sounding system provided to the project by the Atmospheric Technology Division (ATD) of the National Center for Atmospheric Research (NCAR), USA. TAOS (Tethered Atmospheric Observing System) allows continuous measurement of meteorological information (wind speed, direction, temperature, humidity and pressure) at eight levels to a height of 1000 metres. Professor Nigel Tapper is spending May 2001 at NCAR familiarising himself with TAOS and its capabilities and exploring possibilities for its deployment in this project.

To further elucidate the influence of fire scars on local to mesoscale atmospheric processes, several modelling approaches using the PSU/NCAR mesoscale model (MM5) would be undertaken that would complement the observational data. Modelling would focus on both the local and regional scales. The critical modelling work will be undertaken using MM5, a limited-area, non-hydrostatic model capable of simulating (predicting) mesoscale and regional-scale atmospheric circulations (Anthes and Warner 1978). MM5 is a community based model (dedicated to the public domain) and was developed by scientists at Pennsylvania State University and NCAR, Colorado. The model has been used for a broad spectrum of theoretical and real-time studies, including studies of land/sea breezes, mountain winds and urban heat islands. Recently, the modelling system has been coupled to global climate and hydrological models for studies into regional scale climate prediction. The source code is extremely portable, which combined with diversity in application, makes the modelling system the numerical tool of choice for 658 institutions and universities in 33 countries. The MM5 model will initially be run at the Centre for Atmospheric Research at the University of Canterbury while the Monash MM5 modelling capability becomes established. The modelling methodology will be broadly as follows.

A process based modelling study will be undertaken to examine the influence of burned savanna patches of varying sizes on local to mesoscale circulation, convection and precipitation. Observations of radiation, flux and other surface based measurements (soil moisture, vegetation, etc.) from fire scars will be used directly to parameterise the land surface component of the mesoscale model (NCAR MM5) for areas of burned and unburnt savanna. A 500 x 500 km domain comprising an area on the northern territory will be chosen for the experiments. Output data from the CSIRO and ECMWF numerical models from a single year will be used to force the regional model. Validation of the model for this area can be achieved using the flux and tethered balloon observations. The influence of fire scars on precipitation will be more likely during the period of transition to the wet season, a time when the fire scars are still relatively fresh and when atmospheric moisture is increasing. Modelling will concentrate on such conditions. A series of experiments will be conducted using various burned patch sizes including a control (unburnt) and 1, 10, 100, 250 km 2 patches. Circulation patterns, along with convective development and precipitation will be compared to the control simulation to determine the local to mesoscale hydrological influence and the potential for feedbacks between fire scars, precipitation and the re-establishment of savanna.

To investigate regional scale effects of burning, the domain of the MM5 model would be expanded to incorporate the region of northern Australia, north of the Tropic of Capricorn. Once again circulation patterns (especially monsoon), along with convective development and precipitation will be compared to the control simulation to determine the regional influence and potential feedbacks between fire scars and climate. In the regional scale experiment a control simulation with no imposed burning would be contrasted with the contemporary regional burning regime (from the NT Bushfires Council). An additional scenario would include a simulated fire regime and change in savanna vegetation under the doubled CO 2 scenario using a frame-based transient vegetation modelling approach (see Hypothesis 3).

In relation to Hypothesis 3, the climate change associated with a projected doubling of CO2 will lead to a more extreme fire regime for northern Australian savannas and most probably an increase in the area burnt in each fire season (Williams et al., 2001). This will lead to changes in vegetation and climate in a series of complicated feedbacks. We plan to work with output from the CSIRO5 Global Climate Model under the doubled CO2 scenario (IPCC) to provide a climatic driver for simulations. A frame-based modelling paradigm will be used to simulate the outcomes of possible changes in fire frequency in the spatial domain under a doubled CO2 scenario. Frame-based models are quantitative versions of conceptual state and transition models. Frames represent distinct vegetation states, and rules based on key processes such as fire are established to switch between frames. Frame-based modelling has been used successfully to simulate the interactions between elephants, woodlands and fire management in Zimbabwe (Starfield et al. 1993), and to simulate long-term responses of plant species composition of South African rangelands to cattle, goats and fire (Hahn et al. 1999).

To investigate regional feedbacks of fire and vegetation to climate we would use the MM5 model to characterise the temperature, moisture and wind fields under three different scenarios, with particular reference to monsoon indices, which may mark changes in monsoon influence under this scenario. The north Australian domain (outlined in Hypothesis 2) will be used for the following three simulations.

A. Contemporary vegetation in the absence of fire.

B. Contemporary vegetation including current fire regime.

C. Simulated fire regimes and changes in savanna vegetation under a doubled CO2 scenario based on the frame-based transient vegetation model.


January-July, 2002 ; order, assemble and check all equipment at Monash prior to installation in Northern Territory; award Ph.D. Scholarships at Monash (earlier) and appoint Research Assistant in Northern Territory (later). Set up modelling domain and initiate early model testing.

August-December, 2002 ; install equipment in the Northern Territory, rotate and maintain equipment as described, continuing observations into the early wet season with regular down-loading and archiving of data. Undertake profiling operations. Integrate CSIRO and ECMWF model data for forcing of MM5.

January-July, 2003 ; check and calibrate equipment as necessary. Undertake processing of previous season's field data (flux and profiling) including quality control and corrections to derive reliable data sets. Undertake model validation using observational data. Initiate frame-based vegetation modelling work.

August-December, 2003 ; re-install equipment to establish fluxes over a range of chronosequences, rotate and maintain equipment into the early wet season with regular down-loading and archiving of data. Undertake modelling experiments on burn patches v's control, concentrating initially on local scale simulations, but then scaling up to regional scale impacts of burning. Continue with implementation of frame-based modelling approach, concentrating on vegetation changes under climate warming. Begin dissemination of first results, initially through conference presentations.

January 2004 and beyond; Complete processing and analysis of observational data. Continue modelling efforts, incorporating information from the frame-based model to concentrate on potential fire impacts under a doubled-CO2 scenario. Publication and dissemination of results by all participants along with the successful completion of the two Ph.Ds.

National Benefit

Australia will benefit from this project in a number of ways. The Australian environment is strongly impacted by human burning practices, nowhere more so than in its tropical savanna regions. There is a broad understanding of the background to this burning and the resulting emissions, but there is currently no understanding of the direct atmospheric impacts such as are postulated here. To our knowledge, none of the work proposed here is being done elsewhere in Australia or overseas. This study will lead to unique new scientific understandings that will be of wide interest, and it will also contribute to the development of management strategies for Australia's tropical savanna lands in a changing climatic environment. In addition, the flux measurements proposed here will contribute directly to the OZFLUX initiative (Leuning et al. 2001) which seeks to provide estimates of net carbon exchange between land and atmosphere at regional to continental scales. Finally, through inclusion of the Ph.D. scholarships this project will provide post-graduate research training in micrometeorology and numerical modelling for two young Australian scientists.

Communication of Results

Preliminary results and project summaries and reports will be made available on the WWW through a dedicated web site to be developed under the project. This web site will provide information to collaborators as well as a general public audience. Final results and conclusions will be published in high profile Australian and International Journals and we expect to produce 4 to 6 publications in refereed journals such as Boundary Layer Meteorology, Journal of Geophysical Research and Climatic Change. In addition, results will be communicated in a timely manner through attendance at national and international conferences by one or more investigators. Such conferences could include the American Meteorological Society meetings as well as the Ecological Society of America meeting and equivalent Australian conferences such as the annual Australian Meteorological and Oceanographic Society meeting.

Description of Personnel

Chief Investigator (Nigel Tapper - Monash University): Has primary responsibility for conception of the project and will be responsible for overall coordination of the project. He will assist in the establishment and maintenance of the flux observations and will take primary responsibility for atmospheric profiling operations. Nigel will co-supervise one Ph.D. student working on observational aspects of this project.

Chief Investigator (Jason Beringer - Monash University): Responsible for the field component of project including the establishment of sites and instrument towers. Jason will be responsible for the collection and analysis of eddy covariance data. He will co-supervise one Ph.D. student working on observational aspects of this project.

Chief Investigator (Steven Siems - Monash University): Has considerable experience with numerical modelling techniques. Steven will liaise with Professor Sturman in implementing the modelling aspects of this study and will be responsible for developing the MM5 capabilities at Monash University. He will provide primary supervision of one Ph.D. student working on modelling aspects of this project.

Partner Investigator (Lindsay Hutley - Northern Territory University): Lindsay is familiar with the local area and with the flux measurement techniques and will be crucial to the selection of appropriate research sites. He will also be responsible for liaison with appropriate agencies regarding the proposed burning. Finally he will supervise a Research Assistant who will assist in the establishment, maintenance and collection of data from the stations. He will associate-supervise one Ph.D. student working on observational aspects of this project.

Partner Investigator (Andrew Sturman - University of Canterbury): Leads a research group with extensive expertise in local to regional scale atmospheric processes and phenomena, and would be responsible for undertaking the MM5 modelling. He would also be able to provide input to the measurement program, particularly aspects relevant to running the regional simulations. He will provide associate supervision of one Ph.D. student working on modelling aspects of this project.

Partner Investigator (Scott Rupp - University of Alaska Fairbanks): Has extensive experience in frame-based modelling and would be responsible for the application of the frame-based model to Australia's tropical savanna. He would develop scenarios of future vegetation and fire regimes under varying climate as well as addressing land management practices.

All investigators will be involved in the publication and dissemination of results from this project.


Anthes, R.A. and T.T. Warner. 1978. Development of hydrodynamic models suitable for air pollution and other mesometeorological studies. Monthly Weather Review, 106: 1045-1078.

Archer, S., D.S. Schimel, and E.A. Holland. 1995. Mechanisms of shrubland expansion: land use, climate or CO2? Climatic Change, 29: 91-99.

Baker, W.L. 1994. Restoration of landscape structure altered by fire suppression. Conservation Biology 8: 763-769.

Beringer, J., F.S. Chapin III, I. McHugh, N.J. Tapper, A. Lynch and M. Serreze. 2001a. Impact of Arctic treeline on synoptic climate. Geophysical Research Letters (submitted).

Beringer, J., N.J. Tapper and T.D. Keenan. 2001b. Evolution of Maritime Continent thunderstorms under varying meteorological conditions over the Tiwi Islands. International Journal of Climatology (in press)

Beringer, J., D. Packham and N.J. Tapper. 1995 Biomass burning and resulting emissions in the Northern Territory, Australia. International Journal of Wildland Fire 5(2): 229-235.

Bonan, G. 1995 Land-atmospheric CO2 exchange simulated by a land surface process model coupled to an atmospheric general circulation model. Journal of Geophysical Research 100: 2817-2831.

Braithwaite, R.W. 1991. Aboriginal fire regimes of monsoonal Australia in the 19th century.Search 22: 247-249.

Bremer, D.J. and J.M. Ham. 1999. Effect of spring burning on the surface energy balance in a tallgrass prairie. Agricultural & Forest Meteorology 97(1): 43-54

Cachier, H., and J. Ducret. 1991. Influence of biomass burning on equatorial African rains. Nature 352: 228-30.

Egan J.L., R.J. Williams. 1996. Life form distribution of woodland plant species along a moisture availability gradient in Australia's monsoonal tropics. Aust. Sys. Bot. 9: 205-217.

Eugster, W., J.P. McFadden and F.S. Chapin III. 1997. A comparative approach to regional variation in surface fluxes using mobile eddy correlation towers. Boundary-Layer Meteorology 85:293-307.

Hahn, B.D., F.D. Richardson and A.M. Starfield. 1999. Frame-based modeling as a method of simulating rangeland production systems in the long term. Agricultural Systems 62: 29-49.

Hao, W. M., and D. E. Ward. 1993. Methane production from global biomass burning.Journal of Geophysical Research 98: 20657-61.

Hao, W. M., M.-H. Liu, and P. J. Crutzen. 1990. Estimates of annual and regional releases of CO2, and other trace gases to the atmosphere from fires in the tropics, based on FAO statistics for the period 1975-1980. In. Fire in the tropical biota, Ecological studies 84. Ed. J. G. Goldammer, 440-462. New York: Springer-Verlag.

Hutley L.B., A.P. O'Grady and D. Eamus. 2000. Evapotranspiration from Eucalypt open-forest savanna of Northern Australia.Functional Ecology 14: 183-194.

Johnson, B., G. Miller, M. Fogel, J. Magee, M. Gagan and A. Chivas. 1999. 65,000 Years of Vegetation Change in Central Australia and the Australian Summer Monsoon. Science 284: 1150-1152.

Keenan, T., S. Rutledge, R. Carbone, J. Wilson, T. Takahashi, P. May, N. Tapper, M. Platt, J. Hacker, S. Sekelsky, M. Moncrieff, K. Saito, G. Holland, A. Crook and K. Gage. 2000. The Maritime Continent Thunderstorm Experiment (MCTEX): Overview and some results. Bulletin of the American Meteorological Society 81: 2433-2455.

Knowles, J. 1993 The influence of forest-fire-induced albedo differences on the generation of mesoscale circulations. M.Sc., Colorado State University.

Leuning, R. et al. 2001. OZFLUX a cooperative flux network in Australia (unpublished report of meeting held at UNSW, 1 December, 2000).

Parr, C.L. and B.H. Brockett. 1999. Patch-mosaic burning: a new paradigm for savanna fire management in protected areas? Koedoe 42: 117-130.

Physick, W. and N.J. Tapper. 1990 A numerical study of circulations induced by a dry salt lake. Monthly Weather Review 118: 1029-1042.

Russell-Smith, J., G. Allan, R. Thackway, T. Rosling and R. Smith. 2000 Fire Management and Savanna Landscapes in Northern Australia. P.95-101, in Russell-Smith, J., Hill, G., Djoeroemana, S. and Myers, B.(eds.) Fire and Sustainable Agricultural and Forestry Development in Eastern Indonesia and Northern Australia. ACIAR Proceedings No.91, Australian Centre for International Agricultural Research, Canberra, 163pp.

Russell-Smith, J., P.G. Ryan and R. Durieu. 1997. A LANDSAT MSS-derived fire history of Kakadu National Park, monsoonal northern Australia, 1980-94: seasonal extent, frequency and patchiness. J. App. Ecol. 34: 748-766.

Russell-Smith, J., P.G. Ryan, D. Klessa, G. Waight and R. Harwood. 1998. Fire regimes, fire-sensitive vegetation and fire management of the sandstone Arnhem Plateau, monsoonal northern Australia. J. App. Ecol. 35: 829-846.

Scholes, R.J. and B.H. Walker. 1993. An African Savanna: Synthesis of the Nylsvley Study. Cambridge University Press, Cambridge, U.K.

Scholes, R.J. and D.O. Hall 1996. The carbon budget of tropical savannas, woodlands and grasslands. In Global Change: Effects on Coniferous Forests and Gransslands. Ed. A.I. Breymeyer, D.O. Hall, J.M. Melillo and G.I. Ågren John Wiley and Sons, New York..

Starfield, A.M., D.H.M. Cumming, R.D. Taylor and M.S. Quadling. 1993. A frame-based paradigm for dynamic ecosystem models. A I Applications 7: 1-13.

Tapper, N. 2000 Atmospheric Issues for Fire Management in Eastern Indonesia and Northern Australia. P.21-30, in Russell-Smith, J., Hill, G., Djoeroemana, S. and Myers, B. (eds.) Fire and Sustainable Agricultural and Forestry Development in Eastern Indonesia and Northern Australia. ACIAR Proceedings No.91, Australian Centre for International Agricultural Research, Canberra, 163pp.

Tapper, N. 1991 Evidence for a mesoscale thermal circulation over dry salt lakes. Palaeogeography, Palaeoclimatology, Palaeoiecology 84: 259-269.

Tapper, N., G. Garden, J. Gill and J. Fernon. 1993. The climatology and meteorology of high fire danger in the Northern Territory. The Rangeland Journal, 15(2): 339-351.

Williams, A., Karoly, D. and Tapper, N. 2001 The sensitivity of Australian fire danger to climate change. Climatic Change (in press).

Williams, R.J., G.D. Cook, J.A. Ludwig and D. Tongway. 1997. Torch, trees, teeth and tussocks: disturbance in the tropical savannas of the Northern Territory, in N. Klomp and I.D. Lunt (eds) Frontiers in Ecology: Building the Links, Elsevier Science, Oxford, 334 pp.

Zeng N. J.D. Neelin, C. and Chou.2000. A quasi-equilibrium tropical circulation model - Implementation and simulation, Journal of the Atmospheric Sciences. 57(11): 1767-1796.