Marcos Costa

In this paper, we compare three different procedures to calibrate land surface models. The procedures are illustrated using the Integrated Biosphere Simulator (IBIS) and a dataset collected at the Flona Tapajos km 83 Amazonian tropical rainforest site, located near Santarem, state of Para, Brazil. The three procedures are: (a) a single criterion procedure (minimization of RMSE H/LE); (b) a two criteria procedure (minimization of RMSEH + RMSELE); (c) a two-step, six-criteria procedure (our proposal). The two-step procedure first filters only the non-biased estimates and then minimizes a four parameter function. The two-step procedure guarantees a consistent non-biased estimation for the evaluated output variables, sensible heat flux (H) and latent heat flux (LE), and seems to be more robust than the other two methods, for the skill functions evaluated.

The El Niño-Southern Oscillation (ENSO) phenomenon is one of the dominant drivers of environmental variability in the tropics. In this study, we examine the connections between ENSO and the climate, ecosystem carbon balance, surface water balance, and river hydrology of the Amazon and Tocantins river basins in South America. First we examine the climatic variability associated with ENSO. We analyze long-term historical climate records to document the "average" climatic signature of the El Niño and La Niña phases of the ENSO cycle. Generally speaking, the "average El Niño" is drier and warmer than normal in Amazonia, while the "average La Niña" is wetter and cooler. While temperature changes are mostly uniform through the whole year and are spatially homogeneous, precipitation changes are stronger during the wet season (January-February-March) and are concentrated in the northern and southeastern portions of the basin. Next we use a land surface/ecosystem model (IBIS), coupled to a hydrological routing algorithm (HYDRA), to examine how ENSO affects land surface water and carbon fluxes, as well as changes in river discharge and flooding. The model results suggest several responses to ENSO: (1) During the average El Niño, there is an anomalous source of CO2 from terrestrial ecosystems, mainly due to a decreased net primary production (NPP) in the north of the basin. There is also a decrease in river discharge along many of the rivers in the basin, which causes a decrease in flooded area along the main stem of the Amazon. (2) During the average La Niña, there is an anomalous sink of CO2 into terrestrial ecosystems, largely due to an increase in NPP in the northern portion of the basin. In addition, there is a large increase in river discharge in the Amazon basin, especially from the northern and western tributaries. There is a corresponding increase in flooded area, largely in the northern rivers. These results illustrate that changes in water and carbon balance associated with ENSO have complex, spatially heterogeneous features across the basin. This underscores the need for comprehensive analyses, using long-term observational data and model simulations, of regional environmental systems and their response to climatic variability.

At the small scale (< 10 km2), studies that relate changes in land cover with changes in river discharge are abundant. Usually, these studies indicate that deforestation causes an increase in the annual mean discharge, and an increase in the wet season discharge and a decrease in the dry season discharge. However, previous studies that evaluated the effects of changes in land cover in large river basins (> 100 km2) could not find similar relationships. Here we analyze a 50-year long time series of discharge of the Tocantins River at Porto Nacional (175,360 km2), as well as precipitation over this drainage area, during a period where substantial changes in land cover happen in the basin (1949-1998). Initially, we compare one period with little changes in land cover (period 1 - 1949-1968) with another with more intense changes in land cover (period 2 - 1979-1998). Based on agricultural census data, we estimate that, in the middle of period 1 (1960), about 30% of the basin was used as natural pasture, while in the end of period 2 (1995), about 60% of the basin land was used as cropland and pastures. Our analysis indicates that: (a) annual mean precipitation over the basin is not different between period 1 and period 2, at the 95% level of significance (l.s.); (b) annual mean discharge in period 2 is greater than in period 1, at the 97% l.s.; (c) wet season precipitation in period 1 is not significantly different from the wet season precipitation in period 2, at the 95% l.s.; (d) wet season discharge in period 2 is greater than in period 1, at the 99.9% l.s. Our study indicates that, in this basin and for the period considered, the large-scale effects of land cover change are consistent with the effects observed at the small scale. Finally, we compare the effects of climate variability with the effects of changes in land use, by selecting the five wettest and five driest years in each 20-year period. We find that the variability of discharge of the Tocantins River is more affected by climate variability than by land cover change.

Japanese Earth Resources Satellite 1 (JERS-1) imagery acquired over the Amazon basin during low- and high-water periods makes it possible to map seasonal inundation and vegetation of wetlands for most of the basin. Dual-season mapping has now been completed for a central Amazon quadrat extending from 72\deg W,0\deg S to 54\deg W,8\deg S. Imagery was acquired by the JERS-1 L-band, HH-polarized SAR during Sept.-Oct. 1995 and May-June 1996, and mosaicked at the Jet Propulsion Laboratory into low- and high-water mosaics with pixel dimensions of approx. 100 m. Image segmentation software developed at INPE was used to carry out a polygon-based classification of the co-registered mosaics into wetland and non-wetland classes. Wetland areas were classified by inundation state (flooded vs. non-flooded) and vegetation type (non-vegetated, woody, or herbaceous), and classification accuracy was assessed using geo-coded digital videography acquired during aerial surveys of the Brazilian Amazon. Seventeen percent of the study quadrat is occupied by wetlands, which are 96% inundated at high water and 26% inundated at low water (including river and stream channels). Flooded forest constitutes nearly 70% of the wetland area at high water. This mapping methodology is being applied to the entire lowland portion of the basin. In order to map inundation extent at intermediate water stages, and to increase classification accuracy in savanna regions, we are using time series of high-resolution JERS-1 and Radarsat data, and will make extensive use of planned acquisitions from the ENVISAT ASAR and ALOS PALSAR sensors.

Tropical forests harbor a significant portion of global biodiversity and are a critical component of the climate system. Reducing deforestation and forest degradation contributes to global climate-change mitigation efforts, yet emissions and removals from forest dynamics are still poorly quantified. We reviewed the main challenges to estimate changes in carbon stocks and biodiversity due to degradation and recovery of tropical forests, focusing on three main areas: (1) the combination of field surveys and remote sensing; (2) evaluation of biodiversity and carbon values under a unified strategy; and (3) research efforts needed to understand and quantify forest degradation and recovery. The improvement of models and estimates of changes of forest carbon can foster process-oriented monitoring of forest dynamics, including different variables and using spatially explicit algorithms that account for regional and local differences, such as variation in climate, soil, nutrient content, top...

ABSTRACT STATE-OF-THE-ART OF THE SIMULATION OF CARBON FIXATION RATES BY TROPICAL ECOSYSTEMS This work evaluates the state-of-the-art of the simulation of carbon fixation rates, or net primary production (NPP) by tropical ecosystems in two diagnostic configurations (simulation at micrometeorological sites and simulation forced by climate datasets and remote sensing products) and one prognostic (simulation by a coupled climate-biosphere model). The results indicate that the NPP simulation models are capable of unbiased estimates of average regional values of NPP, both in diagnostic and in prognostic mode, with errors smaller than 5%, while the expected value of the error in each site is smaller than 10% in the diagnostic mode and smaller than 20% in the prognostic mode. State-of-the-art models are capable of routinely monitoring the net primary production of tropical ecosystems, with low errors expected for these estimates.

The construction of geographically precise, high-resolution time series of historical changes in land use is an important step in constructing more realistic models of bio-geophysical and bio-geochemical interactions with the atmosphere. However, such data are rarely ...