Close Window

CD-18 Abstract

Modeling Amazonian Carbon Release with Calibrated SVAT Models

John H. Gash — Centre for Ecology & Hydrology (EU-PI)
Carlos Afonso Nobre — INPE - Instituto Nacional de Pesquisas Espaciais (SA-PI)
William James Shuttleworth — University of Arizona (US-PI)

This proposal is motivated

by the question “How do ecosystems respond to and affect global

environmental change and the carbon cycle?” for the Amazônian forest

ecosystem. Over the last decade, description of carbon exchange processes has

been introduced into some of the more realistic and important

Soil-Vegetation-Atmosphere Transfer (SVAT) models used in General Circulation

Models (GCMs). Over the last decade, important progress has also been made in

developing state-of-the-art multi-parameter estimation techniques that can

provide values of preferred sets of the (often many) model parameters used in

complex SVAT models by optimizing against the (often multiple) measurements

collected in present-day field experiments. The LBA Experiment is providing a

uniquely rich source of long-term field data for Amazon forest sites in

different climatic conditions and for a range of soils and disturbance

regimes. We propose to use these data to calibrate the description of the

carbon and energy-water exchange processes represented in two advanced SVAT

models (SiB2C and MOSES-TRIFFID) using advanced multi-parameter estimation

techniques. Further, we will investigate whether and how the preferred sets of

model parameters in the two SVAT models change with season, the nature of the

underlying soil, and/or disturbance regimes, and explore relationships between

the preferred parameters found at individual LBA sites and relevant remotely

sensed (e.g., TERRA) data products. The National Centers for Environmental

prediction (NCEP) and the European Centre for Medium-Range Weather Forecasting

(ECMWF) are both currently preparing long (50-year and 40-year, respectively)

time series of atmospheric variables (including near-surface atmospheric

forcing variables) from historical atmospheric and remotely sensed

observations using data assimilation techniques. These time series will become

available for scientific use within the lifetime of this proposal. We propose

to compare these new model-calculated re-analysis data with Amazônian climate

records and with field data obtained during LBA and earlier studies to

investigate their reliability within Amazônia. Assuming these data are

reasonably realistic (or that simple corrections can be made to make them so),

we then propose to use these historical data to force two-dimensional arrays

of calibrated versions of SiB2C and MOSES-TRIFFID to investigate

model-to-model differences and the spatial and temporal variability in carbon

exchange of and within the Amazônian region. When doing so, we will exploit

any relationships we have previously found between calibrated parameters and

seasonal climate, forest disturbance, underlying soil type, and remotely

sensed variables, as appropriate.

Close Window