Reconstructing late Pliocene global vegetation with a new BIOME4- based Data/Model Hybrid

Reconstructing late Pliocene global vegetation with a new BIOME4- based Data/Model Hybrid Ulrich Salzmann 1,2, Alan Haywood 3 & Dan Lunt 1,4, Dan Hill 1,3 & Harry Dowsett 5 1 British Antarctic Survey, Cambridge, UK 2 School of Applied Sciences, Northumbria University, Newcastle, UK 3 School of Earth & Environment, University it of Leeds, UK 4 School of Geographical Sciences, University of Bristol, UK 5 Eastern Earth Surface Processes, U.S. Geological Survey, Reston, USA

The Late Pliocene / early Pleistocene a key period to the understanding of past human evolution

The Pliocene understanding future climate change the mid Pliocene ( 3 Ma) represents an accessible example of a world that is similar in many respects to what models estimate could be the Earth of the late 21st Century (IPCC, 2007) Multi-model means of future surface warming F UTURE after Zachos et al (2008), Nature 2300 2100 year 2.6-5.3 Ma

The Late Pliocene an ideal Palaeo-Laboratory Continents Orography Flora & Fauna = 400 ppmv atm. CO 2 concentration (present 380 ppmv / pre-indust. 280 ppmv) warmer and wetter (+3 C global) reduced d ice sheets Lisiecki & Raymo (2005) 25 m increase in sea level =

The Pliocene big questions What does/did a warmer world look like? environment for human evolution Is the Pliocene a potential analogue for future climate warming? H lid t d i di t How can we validate and improve predictions of General Circulation Models (GCMs)?

The Pliocene big questions create a robust global vegetation reconstruction for the Pliocene which can communicate communicate with GCMs improved land-cover boundary condition for Pliocene modelling experiments new data set for testing ability of GCMs to successfully retrodict past climate change

Pollen and plant remains a key to understand past continental climates Vegetation is the crystallized, visible climate (Vladimir Köppen, 1936).. bioclimatic envelope, interspecific competition, feedback-mechanisms

Structure of Talk a focus on methods 1. HOW TO. create a Global Biome Reconstruction 2. WHAT s NEW?..the new PRISM3D dataset 3. WHAT FOR? testing and validating GCMs

HOW TO create a Global Biome Reconstruction Data-Model Coupling Strategy MO DEL Mid PRISM Pliocene boundary conditions Incorporation into into HadAM3 GCM Vegetation BIOME4 vegetation model model Data & model use same.. language BIOME4 Dat ta- Model Comparison Advanced Global Biome Reconstruction DATA A Palaeobotanical GIS-database TEVIS Data-synthesis Data-synthesis & translation BIOME4scheme Biome Vegetation mapping reconstruction

HOW TO create a Global Biome Reconstruction THE MODELS Hadley Center Atmospheric Model Vs. 3 BIOME4 Vegetation Model Classification in 27 biomes derived from GCM simulation outputs growing degree days (GDD) mean temperature of coldest & warmest month - 2.5 x 3.75 resolution - 19 levels l in the vertical soil moisture index

HOW TO create a Global Biome Reconstruction THE DATA GIS Database TEVIS Tertiary Environment & Vegetation Information System GCMs layers spatial & temporal queries statistical tools variable boundary conditions

HOW TO create a Global Biome Reconstruction GIS Database TEVIS Palaeobotanical Proxies Data sources: papers, books, conference abstracts, existing compilations 27 biomes / 8 mega-biomes Pollen Translation & Biomisation: using the authors interpretation of palaeobotanical data, taken from the original research papers and/or pers. comm. Rainforest Grassland Fossil Leaves Grouping into mega-biomes to increase statistical robustness Savanna Fossil Wood

HOW TO create a Global Biome Reconstruction Late Pliocene (3.6-2.6 26Ma) 202 palaeo-sites Poor data coverage: central & northern North America northern Eurasia tropical Africa South America Antarctica Salzmann et al (2008), Global Ecol. Biogeography

HOW TO create a Global Biome Reconstruction Selecting best GCM: Coupled AOGCM DGVM TRIFFID PRISM boundary conditions HadAM3 / BIOME4 - fixed SSTs (PRISM) - DGVM TRIFFID - less sea ice Variable sea ice - 400 ppmv atmosph. CO 2

Comparing numerical climate parameters deduced from TEVIS: palaeobotanical A tools for data testing and GCMs and validating GCMs l E 4 Model BIOME Surface Air Temperature KAPPA PalaeoData Vs - Model Degree of Agreement R 2 = 0.754 40 Palaeobot tanical Da ata M O D E L 30 20 10 0 Polar Temperate Tropical Global (> 60) (60-23.5) < 23.5? BIOME 0.32 0.34 0.23 0.31-10 MEGA- -20 BIOME -30 0.46 0.48 Y= -0.313+1.032*x 0.32 0.52-40 -15-10 -5 0 5 10 15 Palaeo-Data 20 25 30 Salzmann et al (2008), Global Ecol. Biogeogr.

HOW TO create a Global Biome Reconstruction Degree of consistency and inconsistency = Global Middle Pliocene Biome Reconstruction Level of Conficence excellent fit moderate fit, (2-3 grids) High confidence change Moderate confidence change No data available, model adapted Salzmann et al (2008), Global Ecol. Biogeogr.

Pliocene vs. Present: Vegetation of a warmer world northward th d shift hift off evergreen taiga t i with much reduced tundra northward shift of temperate forests Middle Pliocene and grasslands in Russia and North America? spread of warm-temperate forests (with subtropical taxa) in middle E Europe and da Asia i expansion of tropical savannas and woodland in Africa and Australia at the expense of deserts Tundra in Antarctica (?) Present-Day

Africa during the Pliocene desert tropical savanna 3 Ma tropical xerophytic shrubland tropical grassland temperate grassland tropical deciduous forest/ woodland 4 tropical semi-deciduous forest 0 BP tropical evergreen forest

TEVIS TEVIS: Relevance and Network Haywood AM, Dowsett H, Otto- Bliesner, B, Chandler M, Dolan A, Hill DJ, Lunt DJ, Robinson M, Nan Rosenbloom N, Salzmann U, Sohl L. (under review.) Geoscientific Model Development (GMD) Lunt D.J., Haywood, A. Schmidt, G.., Salzmann, U., Valdes, P..& Dowsett H.J. (in press), Nature Geoscience Haywood, A.M., Chandler, M.A., Valdes, P.J., Salzmann, U., Dowsett H. J. & Lunt D. J. (2009). Global l and Planetary Change Salzmann, U., Haywood, A. & Lunt, D.J. (2009) Phil Trans Roy Soc A Bonham, S.G., Haywood, A.M., Lunt, D.J., Collins, M. & Salzmann, U. (2009) Phil Trans Roy Soc A, 367 Salzmann, U., Haywood, A., Lunt D. J., Valdes, P. J. & Hill, D. J. (2008) Global Ecology & Biogeography Lunt D.J., Flecker, R., Valdes, P.J., Salzmann, U., Gladstone, R. & Haywood, A. (2008) Earth and Planetary Science Letters http://geology.er.usgs.gov/espteam/prism/prism3main.html..to develop a series of global scale, quantitative data sets for use in mid Pliocene modeling experiments PlioMIP - Pliocene Model Intercomparison Project new land surface boundary condition http://pmip2.lsce.ipsl.fr/

WHAT FOR? A tool for testing and validating GCMs Initial PlioMIP comparison of two GCM s KAPPA degree of agreement HadAM3 Hadley Centre for Climate Prediction and Research GCMAM3 Goddard Institute of Space Studies (GISS) Haywood et al (2009), Glob Planet Change

TEVIS - A tool for testing and validating GCMs Comparing numerical climate parameters deduced from palaeobotanical data and GCMs Surface Air Temperature Precipitation PalaeoData Vs Model R 2 = 0.754 PalaeoData Vs Model R 2 = 0.247 M O D E L 40 30 20 10-10 -20-30 -40 0-15 -10-5 0 5 10 15 Palaeo-Data M 3000 2500 O 2000 D E L 1500 1000 500 0 20 25 30 0 500 1000 1500 2000 2500 3000 3500 Palaeo-Data Comparison of temperature estimates for high latitudes: Region n Unit Present Model Palaeo- Level of Data significance Alaska, E-Siberia 6 MAT -9.1 ± 4.1-3.5 ± 5.0 2.3 ± 0.6 p < 0.05 West Siberia 1 MAT -0.7 7.8 16 - Labrador/Quebec1 MAT 2.2 6 5.7 - Iceland 2 MAT 0.8 ± 1.2 2.1 ± 0.3 3.7 ± 0.4 - Antarctica 1 MAT -47-33 -12 - Uncertainties in model physics and/or boundary conditions

TEVIS - A tool for testing and validating GCMs

TEVIS - A tool for testing and validating GCMs Middle Pliocene AOGCM (HadCM3) Future Scenario A (pre-indust. 400 ppmv CO 2 ) Future Scenario B (pre-indust. 560 ppmv CO 2 ) Validation of simulations with TEVIS Middle Pliocene Dataset

TEVIS - A tool for testing and validating GCMs Middle Pliocene AOGCM Results: good overall agreement same regional changes in vegetation distribution (e.g N- shift of boreal forests) Future Scenario B: 560 ppmv CO 2 best fit with Future Scenario 560 ppmv (k= 0.35 for data-model & k= 0.54 for model-model) Salzmann et al (2009), Phil Trans Roy Soc A

TEVIS - A tool for testing and validating GCMs Middle Pliocene AOGCM Pliocene model (400ppmv) at high h latitudes still warmer than Future Scenario with 550 ppmv!? slow response of polar ice sheets to global warming prevents rapid vegetation change ( buffer ) Future Scenario B: 560 ppmv CO 2 Biomes not in equilibrium state Future model predictions must consider long-term climate sensitivities; e.g. ice-sheet albedo feedbacks Salzmann et al (2009), Phil Trans Roy Soc A

Outlook 1. extending global TEVIS database into other time periods e.g. Miocene & Eocene/Oligocene Early Pleistocene 2. implementing taxa counts (biomisation, palaeo-biodiversity) 3. deducing global numerical climate estimates e.g. coexistence approach (Utescher & Mosbrugger, 1997)

Outlook acquire and explore new palaeo-data to fill existing gaps! Thank you!