1. Simulated Interactions of Soil Moisture, Drought Stress, and Regional Climate in the Amazon Basin Scott Denning 1, Jun Liu 1, Ian Baker 1, Maria Assun ç ão F. Silva Dias 2, Pedro L. Silva Dias 2, John Kleist 1 1 Colorado State University 2 University of São Paulo

2. Tropical Drought Stress Variations in photosynthesis and respiration in tropical forests are strongly influenced by soil moisture Variations in evapotranspiration due to ecosystem drought stress may interact with regional precipitation and climate Ecosystem response to seasonal and interannual drought can help us to test the ability of models to represent physiological stress in tropical forests

3. IGPB News 2001 Niño period

4. Model Output from: TEM (Tian et al, 1998, 2000) & IBIS (Botta & Foley, 2002) models (8 years, colored lines) Data from Tapajos Forest eddy flux sites (black line is moving average monthly NEE  SD, points are average weekly NEE) Mean seasonal NEE and precipitation (  SD of interannual variation) in the Flona Tapajós, Para Slide and analysis provided by Scott Saleska, Harvard University What’s missing in the models? Vertical resolution of respiration- relevant moisture? Diagnosis of moisture levels of litter and CWD? Moisture dependence of surface soil respiration?

5. SiB2 Simulation: Tapajos 1993 Point simulation driven by ECMWF reanalysis Photosynthesis is almost constant through year Respiration decreases during dry season due to dry surface soil layer Simulated net release of CO 2 during wet season, net uptake during dry season, in agreement with flux tower data

6. SiB2 Simulation: Reserva Jaru 1993 11º S: much longer and more pronounced dry season Both photosynthesis and respiration are impacted by drought stress NEE experiences complicated seasonal cycle Is this realistic? Month

7. Precipitation and Ecosystem Stress: A Positive Feedback Loop Less Rainfall Lots of Sensible Heat Hot, dry PBL Low Soil Moisture Severe drought stress No transpiration +

8. Change in Global Climate in HadCM3LC (expt:abuig) Lat: 15 o S - 0 o N Lon: 70 o W - 50 o W Interactive CO 2 and Dynamic Vegetation 2090s - 1990s Results from the Haddley Center - England

9. Haddley Center Climate-driven Amazon dieback 185020002100

10. Change in Amazon Climate and Hydrology in HadCM3LC (expt:abuig) Lat: 15 o S - 0 o N Lon: 70 o W - 50 o W

11. 3-year Coupled Simulation: SiB2-CSU/GCM

12. Exaggerated drought stress feedback in coupled model Photosynthesis collapses, respiration increases Simulated forest is dying! Elapsed months 3-year Coupled Simulation: SiB2-CSU/GCM

13. First Step: Explore the Impact of Parameterization of Drought Stress Soil water potential calculated from soil texture and volumetric water content Exponential dependence Stress is exponential with water potential This makes a step function w.r.t. soil water content! Does this make sense at grid scale? We tried a new parameterization that is less steep in the middle (based on the idea that soil heterogeneity is important) RSTFAC=1/(1+exp(-SLOPE*(www2-wHalfStress))) Control new

14. Result: Kinder, Gentler Droughts Revised parameterization produces moderate reduction in drought stress Allows soils to dry further, so stress still develops 3 year coupled simulation still produces severe sustained drought Reserva Jaru, 1993

15. Next: Effect of Rooting Depth Experiment increased root zone to 10 m thickness, underlain by 5 m deep recharge zone Drought stress continues all year in this scenario, even after 10-year spinup from saturated initial condition Seasonal cycle of simulated NEE reverses sign in this experiment!

16. Impact of model vertical resolution in the water balance in the soil: 15m layer1 2 3 4 5 6 7 8 9 10 root zone 2cm 10m 15m ~45% Porosity ~33% of saturation < wilting point more water in lower layers drier upper layers ~ 2.5m precipitation ~1 m runoff ~ 2.5m precipitation ~1 m runoff SiB2 SiB3

17. SiB3 Soil Structure 10 layers with exponentially increasing thickness Transpiration limited by total plant available water Transpiration distributed through root profile, weighted by water potential and root density Layers Roots

18. Effect of 10-layer Soil Level of saturation shallows during wet season, deepens during dry season Roots can access deeper layers during dry season Only minor stress develops at end of dry season Reserva Jaru 1993

19. Conclusions Seasonal water stress is a major determinant of seasonal carbon exchange in tropical forests in the Amazon region Seasonal cycles of equatorial forest physiology are well captured by SiB2 Seasonal drought stress at higher latitudes is exaggerated by the model Overestimation of drought stress may lead to severe climate simulation errors in coupled models Correct coupled climate simulation requires resolution of vertical gradients of soil moisture, and probably also of soil respiration

20. Ongoing Work Currently running coupled CSU/Coupled model; RAMS experiments with SIB2/3: –Improving numerical efficiency and parallelism of SiB3 –Impact on the downscaling of the seasonal forecasting –Impact on regional forecasting (up to 3-5 days)

21. How realistic are these patterns? Old New Impact on downscaling of the seasonal forecasting