1 Amazon Deforestation in CFS Edwin K. Schneider George Mason University/COLA Climate Test Bed Seminar Series April 2008

2 Abstract Many simulations of the potential effects of tropical deforestation on climate have been made using atmospheric general circulation models (AGCMs) coupled to land models and forced by specified SST. We extend these results by examining the effects of Amazon deforestation on the coupled ocean-atmosphere-land climate system using the NCEP CFS. This CGCM has a good simulation of current Amazon climate compared to many other CGCMs. Century length control and deforestation simulations are carried out with CFS. The model suggests that the impact of Amazon deforestation would be a warmer and drier Amazon, and warmer tropical Pacific and tropical North Atlantic. However, these changes are small. ENSO is not noticeably affected. Sensitivities to changes in the land surface processes are diagnosed using GFS simulations, which suggest that albedo changes are the controlling influence. An unexpected change occurs in the North Atlantic region in the deforestation simulation. We examine the Meridional Overturning Circulation in the simulations in an attempt to understand the origin of this change, which appears to be related to the treatment of sea ice.

3 Collaborators Meizhu Fan (PhD George Mason University, now at NCEP) Ben Kirtman (formerly GMU/COLA, now RSMAS, University of Miami) Paul Dirmeyer (COLA) Bohua Huang (GMU/COLA) Hua Chen (GMU PhD candidate) Ioana Colfescu (GMU PhD candidate) Thanks also to Hua-Lu Pan (NCEP) for help with the CFS land surface model

4 Scientific Motivation

5 References Schneider, E. K., M. Fan, B. P. Kirtman and P. A. Dirmeyer, 2006: Potential Effects of Amazon Deforestation on Tropical Climate. COLA Technical Report 226, 41 pp. Voldoire, A. and J.-F. Royer, 2005: Climate sensitivity to tropical land surface changes with coupled versus prescribed SSTs. Climate Dyn., DOI /s

6 Voldoire and Royer 2005 Coupled Simulation (20 years) of Deforestation of Global Tropics SST DJFSST JJA

7 Why (relative) Neglect of the Continental Heat Sources? W. Pacific heat source is by far the biggest and most important –Easily controlled by modifying SST in W. Pacific –Important component of ENSO variability, some predictability –Obvious teleconnections to CONUS Land based heat sources thought to be much less predictable, difficult to control in AGCM simulations except via SST elsewhere. CGCMs have large biases in precipitation over Amazon

8 Research Question: What is the Response of the Climate System to Land Use Change? Complete Amazon deforestation has a large local effects on rainfall and surface temperature in some AGCM simulations. What are the effects on the coupled climate system? –“New” question: what is sensitivity of coupled climate system to land influences, if any? CGCM experiment: replace Amazon rain forest with grassland type vegetation and do long simulations. –Pure sensitivity experiment, extreme scenario.

9 Societal Motivation Pressures from population and economic growth will be associated with important changes in land use and surface properties. These land surface property changes are expected to affect mean climate and perhaps climate variability and predictability.

10 Models NCEP Coupled Forecast System (CFS) –Coupled atmosphere-ocean-land CGCM –T62L64 AGCM (GFS: NCEP NWP model) –MOM3 (non-polar) –OSU land model –Specified sea ice distribution GFS stand alone –As in CFS –Specified observed SST evolution

11 Basic Experiment 100 year current climate control simulations (1st 50 years run by Kathy Pegion, we extended to 100 years). 100 year deforestation simulations

12 Topics CFS Amazon climatology Sensitivity to deforestation –CFS results –GFS diagnosis –Effects outside of tropics

13 Why CFS? Many CGCMs simulate very dry conditions over the Amazon. –COLA CGCM (Schneider et al. 2006) –ARPEGE/ISBA (Voldoire and Royer 2005) –CCSM

14 Amazon Climatology Land 80°W-40°W; 15°S-8°N COLA CFS P (mm/day) ObsDeforest P (mm/day) T(ºC) JanJunDecJan Jun Dec Control

15 CFS Precipitation Climatology DJFMAMJJASON

16 Model Changes Only surface data over Amazon is changed –sfcsub.f: original data is read in and overwritten with new values (need to be careful with lat-lon indexing, which is different for different fields). –Vegetation type: tropical rain forest replaced by grassland over whole “Amazon.” –Choices Change surface albedo Surface roughness consistent with vegetation change No change made to percentage vegetation coverage No change made to soil type. All else unchanged

17 Vegetation Type In Amazon region, replace vegetation type 1 (tropical rainforest) with type 7 (ground cover perennial) Effect: reduces vegetation resistance r to evaporation (I=V/r)

18 Surface Albedo ControlDeforestDeforest minus Control Take values appropriate to nearby vegetation type 7: Albedo 1 (diffuse near ir): unchanged Albedo 2 (beam near ir): 9% (small increase) Albedo 3 (diffuse sw): unchanged Albedo 4 (beam sw): 30% (increased from about 23%)

19 Surface Roughness Reduced to 0.1m from rainforest value >2m

20 Low Latitude Effects

21 CFS Deforest Minus Control 2m Air Temperature Difference (C) Annual DJFMAMJJASON

22 CFS Deforest Minus Control Precipitation Difference (mm/day) Annual DJFMAMJJASON

23 CFS Surface Wind Stress on Ocean Deforest minus Control (dynes/cm2)

24 Standard Deviation of SST Anomalies (CFS Control Run)

25 CFS Difference in Tropical SST Variability NINO3.4 SSTA SSTA Standard Deviation (C)

26 GFS Simulations 5 year simulations with observed SST and sea ice distribution Simulations –Control –Deforestation –Vegetation change only –Albedo and roughness changes only Annual means

27 T2m DEF-CTLVEG-CTL ALBRGH-CTL (VEG-CTL)+(ALBRGH-CTL)

28 Precipitation DEF-CTLVEG-CTL ALBRGH-CTL (VEG-CTL)+(ALBRGH-CTL)

29 Cloud Cover DEF-CTLVEG-CTL ALBRGH-CTL (VEG-CTL)+(ALBRGH-CTL)

30 Global Scale Effects

31 CFS Global and Annual Mean 2m Air Temperature Red: Control Black: Deforest

32 CFS Annual Mean 2m Air Temperature Difference

33 200mb ‘Geostrophic Streamfunction’: Rossby Wave Propagation Connects Deforestation Region with Warm North Atlantic

34 What is Happening in North Atlantic? Wave train from Amazon (x) MOC? Something else?

35 MOC Diagnosis in CFS (courtesy of Bohua Huang) MOC in CFS Control Comparison to GODAS MOC in CFS Deforestation compared to Control

36

37

38

39

40

41

42 Results from Analysis of Ocean Model Results AMOC in CFS is similar to that in GODAS. Huge “deforestation” temperature response in N. Atlantic appears to actually be a problem with lack of coupling between ocean and sea ice. –Did something happen to the model (e.g. sponge layer) in setting up the deforestation experiment? Also, sea ice data are not identical. AMOC is insensitive to very large thermal anomaly in deforestation experiment. This is consistent with it being primarily wind driven, as deduced from other considerations by Dr. Huang.

43 Summary of Results Simulation of current Amazon climate by CFS is realistic compared to many CGCMs. Local response to deforestation is weak warming and drying. This response appears to be explained by the Charney albedo mechanism, which is mitigated by the decrease in vegetation resistance.

44 Summary of Results II Amazon deforestation in CFS leads to warming and increased precipitation in the equatorial eastern Pacific (El Niño-like response), warming in the tropical North Atlantic, and increased precipitation in central South America. CFS has a 0.5ºC commitment to global warming over about 70 years built in.

45 Conclusions Results are not yet ready for publication because of the North Atlantic high latitude problem. Do over until we get it right.