Otimização multi-sítio de modelo de assimilação de carbono e aplicações em ecossistemas de cerrado e pastagem

• Autor(a) principal: Fischer, Graciela Redies
• Data de Publicação: 2011
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: LOCUS Repositório Institucional da UFV
• Texto Completo: http://locus.ufv.br/handle/123456789/1505
• topic: Carbono; Micrometeorologia; Pastagem; Cerrado; Carbon; Micrometeorology; Grassland; Cerrado; CNPQ::CIENCIAS EXATAS E DA TERRA::GEOCIENCIAS::METEOROLOGIA::MICROMETEOROLOGIA

Front of global climate change scenarios, studies using land surface models are needed to assess the impact of increased CO2 and the carbon balance in terrestrial ecosystems. However, to be effective these models must be carefully calibrated to accurately represent ecosystem processes. Generally, the model is calibrated for one site and then used with the same set of calibrated parameters, either in other sites or in a whole region with the same plant functional type. In view of the exposed, this research has two main objectives: first, make a multi-site calibration, using two sites simultaneously, and evaluate their performance; and then, make application of the results of multi-site calibration to generate estimates of net primary production (NPP) in cerrado and grassland ecosystems in South America. To investigate the multi-site calibration, the calibration was performed to net ecosystem exchange (NEE) for two pasture sites in the Amazon. Twenty different objective functions, being five adjustment measures subject to four optimization methods (two individual optimizations - one for each site - and two multi-site optimizations - one with the same weight to each site and the other with a weight proportional to the size of the dataset) are evaluated to investigate the consistency of the results in a multi-site model optimization. The five adjustment measures were used: correlation coefficient (r), root mean squared error (RMSE), mean absolute error (MAE), mean bias error (MBE) and maximum bias error (Emax). Our results indicate that, with some restrictions regarding the choice of objective function, multi-site calibration is possible and produces consistent results across sites. We recommend that the choice of objective function should be based on the intended use of the model. On a short time scale we recommend the site-weighted method using mean absolute error as adjustment measure. For longer times scales we recommend using the site-weighted maximum bias error. For this research the priority is long-term simulations, so we applied the site-weighted method using maximum bias error to generate estimates of NPP in cerrado and grassland in South America. The calibrated parameters were applied in Algorithm RATE (based on the SITE vegetation model), which is an algorithm for regional monitoring of the carbon fixation rate using land cover, LAI and FAPAR datasets derived from a remote sensing satellite (MODIS) product. The results indicate a pronounced seasonal variability in monthly NPP estimates for cerrado and grassland in 2010 linked to the climate variability of the region - periods of high and low precipitation and temperature driving seasonal differences in carbon fixation. Interannual NPP variability was higher for grassland than cerrado, perhaps indicating that grassland is more susceptible to climate variability. There were also clear differences between MODIS NPP estimates and those of RATE, indicating that further studies are urgently needed to rigorously validate these alternative strategies to estimate NPP.