An Approach to Deal with Sources of Data Uncertainty in Dissolved Organic Carbon Dynamic
Virginia
Brilhante, Computing Science Departament, Federal University of Amazonas – UFAM, Manaus-AM, Brazil, virginia@dcc.fua.edu.br
J.
L.
Campos dos Santos, The National Institute for Amazon Research – INPA, Manaus-AM, Brazil, lcampos@inpa.gov.br
(Presenting)
Uncertainty is an intrinsic feature of complex simulation model. Given that it is not possible to rid the models from uncertainty we are left with taking notice of it for consideration in model-based decision making. Traditional ecological modeling methods and tools do not support explicit account of model uncertainty. LBA experiments that investigate dissolved organic carbon dynamic for different forest compartments (rainfall, throughfall, groundwater, freshwater) also have to deal with data uncertainty before any simulation methods take place, since it utilize the same calibration of one compartment to all others compartments. This work gives a contribution towards making known, or bringing to the surface, sources of uncertainty that are embedded in ecological models. The sources of uncertainty are related to the model’s supporting data and equations. A metadata standard is integrated to specify data-related sources of uncertainty, such as publisher, coverage, date, place, etc. In the technique implemented, models are described and simulated using logic, which allows the sources of uncertainty to be represented, propagated and combined during a simulation process. The combined sources of uncertainty can then be presented to data brokers to assess their impact on models outputs and tune up his confidence in the model for decision making.
Submetido por Jose Campos Dos Santos em 25-MAR-2004
Tema Científico do LBA: CD (Armazenamento e Trocas de Carbono)