Projections of harvest disruptions due to rainfall for the Australian Sugarcane industry under future high and low emission climate scenarios J. Sexton, Y.L. Everingham a and B. Timbal b Presenting Author: Justin Sexton (justin.sexton1@jcu.edu.au) a School of Engineering and Physical Sciences, James Cook University b Centre for Australian Weather and Climate Research, Bureau of Meteorology
The Sugar Industry in Australia 2/25
The Sugar Industry in Australia 3 rd Largest raw sugar exporter 3 Million tonnes of raw sugar 80% exported 4000 Cane Farming businesses $1.5 - $2.5 billion production Current value of raw sugar (sugar#11) US /lb 19.7 AUD /kg 45.19 AUD $/T 451.9 3/25
Where is Sugar Grown? Heterogeneous Landscape Spatially variable rainfall Seasonally y variable rainfall 1 2 3 4 Seven regions Considered 1. Tully 2. Herbert 3. Burdekin 4. Mackay 5. Bundaberg 6. Rocky Point 7. New SouthWales 5 6 7 Total Annual Rainfall (mm) 4/25
The Crop Cycle Grand Growing Farleigh 350 planting rainfa all (mm) 100 150 200 25 50 300 Grand Ripening harvesting Tillering 0 50 jan feb mar apr may jun jul aug sep oct nov dec 5/25
Modelling Harvest Disruptions: Wet days Rule 6/25
Wet days rule Work with industry to define how much rain causes a disruption Disruptions due to rainfall wet days (Muchow,et al. 1996) If rainfall for day i >=10mm and <20mm then day i is defined to be a wet day. If rainfall for day i >=20mm and <40mm then day i and i+1 are defined to be wet days. If rainfall for day i >=40mm then day i, i+1, i+2 are defined to be wet days. Used in climate variability impact studies (Everingham, et al. 2001). R. Muchow, A. Wood, 1996, Rainrisk Database System Information Manual. Brisbane, CSIRO Division of Tropical Crops and Pastures. Y. Everingham, R. Stone, R. Muchow, 2001, Proceedings Australian Society of Sugar Cane Technologists (ASSCT) 7/25
Data: Climate Scenarios, GCMs and Downscaling 8/25
Climate Scenarios Compare two Time Frames 1961 2000 2046 2065 Comparison between simulations using three Climate Scenarios 20C3M (1961 2000) B1 (2046 2065) A2 (2046 2065) 20C3M (1961 2000) 20 th Century forcings B1 (2046 2065) Lower emission scenario A2 (2046 2065) Higher emission scenario 9/25
General Circulation Models 200Km to 400Km 1 Spatial trends vary between and within regions 2 3 4 Maintain spatial trends Can downscale data to more appropriate size 5 0.05 x 0.05 decimal degrees 6 7 Total Annual Rainfall (mm) 1. Tully 2. Herbert 3. Burdekin 4. Mackay 5. Bundaberg 6. Rocky Point 7. NSW 10/25
General Circulation Models 200Km to 400Km 1 Spatial trends vary between and within regions 2 3 4 Maintain spatial trends Can downscale data to more appropriate size 5 0.05 x 0.05 decimal degrees 6 7 Total Annual Rainfall (mm) 1. Tully 2. Herbert 3. Burdekin 4. Mackay 5. Bundaberg 6. Rocky Point 7. NSW 11/25
Downscaling Rainfall data downscaled using Analogues Methodology Statistical downscaling methodology (SDM) Developed for Bureau of Meteorology (Timbal et.al,2011) Based on 10 climate regions in Australia B.Timbal, Y. Wang, A. Evans (2011), 19th International Congress on Modelling and Simulation, ISBN: 978 0 9872143 1 7 12/25
Methodology 13/25
Methodology Downscaled Climate Scenario Rainfall Apply Harvest Disruption Rule Analysis Identify number Bias Analysis 20C3M of Unharvestable 20C3M AWAP days 1961 2000 PDF 005 0.05 0 x0.0505 0 Identify spatial trends pixels Climate Change B1 20C3M B1 Winter and Spring A2 20C3M 2046 2065 1961 2000 A2 11 GCMs Boxplots KW test for significance Identify spatial trends 14/25
Model Bias I REJECT YOUR REALITY AND SUBSTITUTE MY OWN -Adam Savage(Mythbusters) 15/25
Model Bias Historical simulated data (20C3M) compared to baseline data (AWAP) Probability density functions compared Individual GCMs (11) Bias analysed spatially Temporal average of 11 GCMs 16/25
Results: Regional and Spatial analysis 17/25
Regional Analysis General shifts between time periods noted for each climate scenario Paired Wilcoxon test used to check for significant shifts in distribution Differences between historical and future projections analysed spatially 18/25
Change in Unharvestable days: A2 nharvestable days hange in Un Ch Sexton, Everingham, Timbal, 2012, International Journal of Climate Change Strategies and Management. in prep 19/25
Change in Unharvestable days: A2 ble days nharvestab hange in Un Ch Sexton, Everingham, Timbal, 2012, International Journal of Climate Change Strategies and Management. in prep 20/25
Change in Unharvestable days: A2 ble days nharvestab hange in Un Ch Sexton, Everingham, Timbal, 2012, International Journal of Climate Change Strategies and Management. in prep 21/25
Sexton, Everingham, Timbal, 2012, International Journal of Climate Change Strategies and Management. in prep 22/25
How can industry adapt? Start harvest early Modifying harvest window can improve farmer profits (Everingham et.al. 2011) Expand harvest regions 23/25 Y. Everingham, N. Stoeckl, J. Cusack, J. Osbourne 2011, International Journal of Climatology
How can industry adapt? Upgrade harvesters Wet Harvesters used in 2010 Season http://www.dailymercury.com.au/news/hi tech tyres assistenvironment/1568469/ Track harvester Ingham, tracks offer advantage of less soil compaction under wet conditions. ComsinHarvesting i owner Stephen Comelli with one of the new [Floatation] tyres he is using on some of his haul out machinery. The tyres tread more lightly on farmers paddocks and are better for the environment. Peter Holt 24/25
Key Points 1. Some models perform better than others historically Good models may not be best in future Use multiple models 2. Downscaling allows precision results Reproduce spatial patterns 3. Higher emissions = more down time Later in the harvest season 25/25
Thank You. 26/25
Rainfall variability planting Grand Ripening harvesting Tillering 27/25
The Crop Cycle Ratoon crop Planting Growing Harvest 28/25
Summary of Regional Analysis Simulated seasonal trends Increases in spring Decreases in winter Burdekin decreased in spring Simulated significant trends Tll Tully : Increase (A2) Herbert: Increase(B1, A2) Burdekin: Decrease(A2) 29/25
Global Climate Scenarios Use Global Climate Models developed as part of World Climate Projects CMIP3 database 11 GCMs Developed around the world Models used in Australian research previously Daily rainfall data available for both time periods Group Country Acronym Canadian Climate Centre Canada CCM Metro-France France CNRM CCM: Canadi an CSIRO Australia CSIRO CSIRO Australia CSIRO2 Geophysical Fluid Dynamics Lab U.S.A GFDL1 Geophysical Fluid Dynamics Lab U.S.A GFDL2 NASA/Goddard Institute for Space Studies U.S.A GISSR Institut Pierre Simon Laplace France ISPL Centre for Climate Research Japan MIROC Max Planck Institute for meteorology DKRZ Germany MPI Meteorological Research Institute Japan MRI 30/25
AWAP predictands NRR reanalysis predictors Validation 11 GCMs 3 Climate Scenarios Downscaled Climate Scenarios 20C3M 20C3M A2 B1 ANALOGUE STATISTICAL DOWNSCALE MODEL A2 B1
The Harvest Period Harvest during drier months June November High rainfall during Harvest period can reduce sugar content of cane Disruptions can push harvest into wet season December February 32/25
Sexton, Everingham, Timbal, 2012, International Journal of Climate Change Strategies and Management. in prep 34
Overview 1. The sugar industry in Australia 2. Modelling harvest disruptions: wet day rule 3. Data Climate scenarios Global Circulation Models Downscaling 4. Methodology Bias analysis Change over time 5. Results Regional change over time Spatial patterns 6. How can industry adapt? 7. Key points 35/25
Summary of Bias Some GCMs reproduced PDFs better Differedregionally regionally andseasonally Spatial trends between and within regions North/South trends between regions East/West trends within regions Evidence for over estimates in Tully Evidence for underestimates in Herbert / Burdekin 36
PDF Reproducing GCMs Region Winter Global Climate Models 1. Tully/ Far North MRI, GFDL1 MPI Spring 2. Herbert GFDL1, CCM CNRM, MIROC 3. Burdekin CSIRO, CNRM GISSR, MIROC 4. Mackay ISPL CSRIO2, CCM 5. Bundaberg CNRM, MRI GISSR, GSS CNRM 6. Rocky Point GFDL1, CNRM GISSR, MIROC 7. New SouthWales MRI, GISSR ISPL 37
Summary of Bias Sexton, Everingham, Timbal, 2012, International Journal of Climate Change Strategies and Management. in prep 38
Direction of change in Number of Unharvestable days across 11 GCMs Region B1 A2 Winter Spring Winter Spring 1. Tully 2. Herbert 3. Burdekin 4. Mackay 5. Bundaberg 6. Rocky Point 7. New South Wales Sexton, Everingham, Timbal, Unpub (2012) Increase Increase Increase Increase Decrease Increase Decrease Decrease Decrease Increase Decrease Decrease Increase Increase 39
Summary of Spatial Analysis Shifts in Number of unharvestable days varied Bt Between regions Within regions Northern and Southern regions were more variable than central regions Central area of Tully region showed higher increases than edges Burdekin region experienced least geographic spread 40