Presentation on theme: "Modelling Australian Tropical Savanna Peter Isaac 1, Jason Beringer 1, Lindsay Hutley 2 and Stephen Wood 1 1 School of Geography and Environmental Science,"— Presentation transcript:
Modelling Australian Tropical Savanna Peter Isaac 1, Jason Beringer 1, Lindsay Hutley 2 and Stephen Wood 1 1 School of Geography and Environmental Science, Monash University, Melbourne 2 School of Science and Primary Industry, Charles Darwin University, Darwin
Introduction Savanna occupies ~20% of the Earth’s surface and ~25% of Australia Undisturbed Australian savanna is a sink of CO 2, -3 to -4 tCha -1 yr -1 Tropical savanna is a highly dynamic ecosystem –mix of C3 and C4 plant species –large annual variation in F e and F c in response to wet season/dry season climate –large inter-annual variation due to fire
ARC Discovery Project Title –“Patterns and Processes of Carbon, Water and Energy Cycles Across Northern Australian Landscapes: From Point to Region” People –Beringer (Monash), Hacker (ARA), Paw U (UCD), Neininger (MetAir AG), Hutley (CDU) Methods
Sites Howard Springs –open forest savanna Fogg Dam –wetland Adelaide River –woody savanna Daly River Uncleared –open forest savanna Daly River 5 year –regrowth Daly River 25 year –pasture
Howard Springs Fluxes –F sd, F su, F ld, F lu, F n, PAR –F m, F e, F h, F c, F g Meteorology –T a, RH, WS, WD Concentrations –CO 2, H 2 O Precipitation –Rainfall Soil –moisture (10 & 40 cm) –temperature 12º 29.655S 131º 09.143E Open forest savanna
Questions To accurately model the seasonal variation in F e and F c over tropical savanna, is it necessary: 1) for the data input to the model to resolve the seasonal change in L ai ? 2) for the data input to the model to resolve the seasonal change in C4 ? 3) to use a multi-layer model to resolve changes in the canopy ?
CABLE CSIRO Atmosphere Biosphere Land Exchange model Big leaf model –Kowalczyk et al. (2006), CMAR Paper 013 –coupled assimilation/transpiration –one sunlit leaf, one shaded –mixed C3/C4 canopy by specifying C4 fraction –seasonally varying L ai and C4 fraction –radiation in IR, near IR and visible –13 vegetation types, 9 soil types, 6 soil layers –destined to be the LSM in ACCESS
ACASA Advanced Canopy Atmosphere Soil Algorithm Multi-layer model –University of California, Davis –Pyles et al., 2000, QJRMS, 126, 2951-2980 –coupled assimilation/transpiration –third-order closure turbulence sub-model –100 canopy layers for radiation –20 canopy layers for turbulence/fluxes –15 soil layers –no C4 pathway
Savanna canopy C3C4 Sunlit Shaded CABLE C3 ACASA C3 overstorey L ai 0.6 - 1.0 h c 16 m C4 understorey L ai 0.08 - 1.4 h c 0.1 - 2 m C4 roots shallow C3 roots deep Reality vs Model
Results 1) Out-of-the-box –all defaults except LAI 2) Basic Tuning –“educated guess” 3) Constant L ai –as for 2), L ai = 1.4 4) Constant C4 fraction –as for 2), C4 fraction = 0.39
Basic Tuning Soil moisture at wilting point reduced from 0.135 to 0.08 m 3 m -3 based on observations Root fraction for E. tetradonta according to Eamus et al. (2002) CABLE –v cmax increased from 10 to 30 molm -2 s -1 ACASA –set soil microbial respiration to 0
Summary Tropical savanna is a dynamic system –mix of C3 overstorey and C4 understorey –L ai and C4 fraction respond mainly to soil moisture –soil moisture driven by bi-modal rainfall Questions –do we need a multi-layer model ? –do we need seasonally varying L ai ? –do we need seasonally varying C4 fraction ?
Conclusions (ACASA) The multi-layer model (ACASA) did not perform better than the single layer model (CABLE) for this study –Raupach and Finnigan (1988) The multi-layer model did not perform well enough to make conclusions about the necessity of resolving seasonal changes in L ai and C4 fraction in the input data.
Conclusions (CABLE) Basic tuning significantly improves model performance. When tuned, CABLE over-predicts F c in the wet season and under-predicts F e in the dry season. CABLE is sensitive to both L ai and C4 fraction when predicting F c but is not sensitive to either when predicting F e. C4 fraction must vary by season to correctly predict seasonal changes in F c.