The University of Arizona Brazilian Partners Harvard University The University of Arizona


The influence on global climate on the Amazon region is unquestionable, and therefore, an understanding of ecological and hydrometeorological interactions is of great importance. In general circulation models the soil-vegetation-atmosphere interface is described by land surface parameterization schemes (LSPs). The models have evolved from simple tipping-bucket models to fully interactive models containing tens of parameters to be specified by the user. Calibrating LSPs against measurements available at field sites can reduce parameter estimation uncertainty. Some model parameters can be defined in controlled (i.e., temperature, humidity, CO2 concentration) conditions in leaf-size chambers or greenhouses but at a much smaller spatial scale than that of which they are used in GCMs. Fortunately, facilities such as the Biosphere 2 (B2) provide opportunity to conduct experiments at larger spatial scales that include contributions from a variety of different species. Meteorological conditions inside B2 Tropical Rain Forest (TRF) are controlled to allow ‘large chamber’ measurements of physiological and physical properties of the TRF ecosystem with practically non-existent unmeasured advection.

The research described in here takes advantage of currently available data from several eddy covariance flux towers located in the Brazilian Amazon basin to estimate parameter values of the version 3 of the Simple Biosphere model (SiB3). Background climatological data were used to assess the representativeness of the data collection period that might have affected model calibration. Variance-based sensitivity analysis is used to investigate potential structural deficiencies in SiB3 and to reduce the dimensionality of the subsequent optimization by identifying those model parameters that merit calibration. Additionally, some structural and conceptual aspects of SiB3 were tested inside B2 TRF under meteorological conditions that resemble those predicted in future climate scenarios for the Amazon basin.

Notable Accomplishments:

a. Full implementation of the SiB3 in the artificial climate in B2-TRF that resembles that predicted in the future for the Amazon region arguably suggests tropical forests may be less susceptible to the higher temperatures predicted by GCMs than hitherto thought.

b. A novel, rank-based screening method was developed and implemented in the variance-based Sobol method to account fully for the multi-output nature of land surface models. This approach is model-independent and can also be implemented with different multi-criteria strategies. For SiB3, we found the majority of the sensitive parameters are associated mainly with physiology, or soil and carbon properties; optical properties were found to have little influence on surface fluxes.

c. A multi-site calibration framework was developed and used to calibrate the SiB3 model against data from the tower sites in the LBA Experiment. This new framework not only substantially reduces the uncertainties associated with ill-defined parameters but also identifies the sources of uncertainty associated with statistical components, i.e., the uncertainty in signal mean, variability, and timing/shape.


Rosolem, R., H. V. Gupta, W. J. Shuttleworth, L. G. G. de Gonçalves, X. Zeng (Accepted) Assessing Components of Uncertainty in Parameter Estimation of the SiB3 Land Surface Model for Amazon Biomes, Hydrological Processes

Rosolem, R., H. V. Gupta, W. J. Shuttleworth, X. Zeng, and L. G. G. de Gonçalves (In Press) A fully multiple-criteria implementation of the Sobol method for parameter sensitivity analysis, Journal of Geophysical Research-Atmospheres, doi:10.1029/2011JD016355

Rosolem, R., W. J. Shuttleworth, X. Zeng, S. R. Saleska, T. Huxman (2010) Land surface modeling inside Biosphere2 Tropical Rainforest biome, Journal of Geophysical Research - Biogeosciences, 115, G04035, doi:10.1029/2010JG001443

Saad, S., H. R. da Rocha, M. A. F. Silva Dias, and R. Rosolem (2010) Can the deforestation breeze change the rainfall in Amazonia? A case study for the BR163 highway region, Earth Interactions, 14, 1-25, doi:10.1175/2010EI351.1

Rosolem, R., W. J. Shuttleworth, and L. G. G. de Gonçalves (2008) Is the data collection period of the Large-Scale Biosphere-Atmosphere (LBA) Experiment in Amazonia representative of long-term climatology?, J. Geophys. Res., 113, G00B09, doi:10.1029/2007JG000628

Project Collaborators:  R. Rosolem, Research Assistant Professor, University of Arizona; W. J. Shuttleworth, Professor, University of Arizona; H. V. Gupta, Professor, University of Arizona; X. Zeng, Professor, University of Arizona; L. G. G. de Gonçalves, CPTEC/INPE, Brazil; S. R. Saleska, Professor, University of Arizona; T. Huxman, Professor, University of Arizona; J. van Haren, Assistant Research Professor, Biosphere 2, University of Arizona; I. Baker, Colorado State University; A. C. Araújo, Embrapa, Brazil; S. C. Wofsy, Professor, Harvard University; M. L. Goulden, Professor, University of California Irvine; S. D. Miller, Professor, SUNY Albany; H. R. da Rocha, Professor, University of São Paulo, Brazil; C. von Randow, CCST/INPE, Brazil; D. R. Fitzjarrald, Professor, SUNY Albany.