Analysis of long-term fire dynamics and impacts in the Amazon using integrated multi-source fire observations

Principal investigator:

Ivan A. Csiszar

University of Maryland, Department of Geography

Co-principal investigators:

Alberto Setzer

CPTEC/INPE, Brazil

Karla Maria Longo de Freitas, Modeling Division

CPTEC/INPE, Brazil

Co-investigators:

Christopher Schmidt

Cooperative Institute for Meteorological Satellite Studies

Space Science Engineering Center

University of Wisconsin-Madison

Jeffrey T. Morisette

NASA Goddard Space Flight Center

Elaine Prins

Consultant in environmental remote sensing applications

Wilfrid Schroeder

University of Maryland, Department of Geography

Saulo Ribeiro de Freitas, Modeling Division

CPTEC/INPE, Brazil

Important results from the LBA-Eco Phases I and II provided improved understanding of vegetation fires as observed from satellite sensors and this knowledge can now be used to reassess conventional methods in order to create more robust fire data sets for the region and to more precisely quantify fire implications in terms of land use management and carbon emissions. This research is based on the creation of a standardized fire climate data record for the Amazon as derived from multiple satellite sensors and on the application of the resulting data set for the quantification of fire impacts in the region.

In the initial phase of this project GOES East data will be reprocessed using an updated active fire detection and characterization algorithm (version 6.0 of the WF_ABBA) that generates reduced detection errors, along with the production of an improved cloud product that will be used in combination with the fire data to help quantify fire regimes in the region. This will extend the GOES time series from the actual 2002-present (version 6.0) to 1994-present (more than 11 years) with which improved trend analysis can be achieved. Access to the extended GOES fire data records will be made open to the scientific community (LBA and others) via a data bank architecture that will be accessible online.

An extensive data inter-comparison analysis is done to study the relationship between different satellite fire products. This component of the project will include the adaptation of the MODIS active fire validation methodology developed within the LBA-Eco project LC-23 “Quantifying the Accuracy of MODIS Fire Products and Establishing Their Relationship with Land Cover Dynamics”, based on coincident ASTER imagery. We are developing a more generic multi-platform technique that will allow the evaluation of any fire product from moderate resolution sensors (primarily GOES in this study) based on the availability of near-coincident high resolution imagery (primarily Landsat and ASTER); and the comparison of thus validated fire products with those from sensors where coincident high resolution imagery is not available (i.e. AVHRR). The linkages between moderate and coarse resolution products are established using data fusion methods towards the creation of a unified fire diurnal cycle map as derived from multiple sensors that can be used to model time dependent variables (e.g.: emissions). Fire area and temperature and fire radiative power, important parameters that are related to biomass burning efficiency and emission values, will be implemented using the suite of data sets selected here.

The final stage of the project will address the impacts caused by vegetation fires in the Amazon region. The extended and improved fire data products will be used to produce enhanced fire spatial and temporal distribution maps which, in combination with other parameters (e.g.: deforestation data) will be used to explain land management practices and their consequences to the environment. Emission modeling studies will also be addressed at this final phase of the proposal by feeding models with the results produced in the previous sections.

The methods designed for this project should provide the basis for the development of an integrated fire product that can be applied to climate studies at the regional and global scales as new sensors with improved spatial coverage become available.