Modeling and sensing growth rates of tropical rainforests Methodologies and needs Marcos Heil Costa Universidade Federal de Viçosa Workshop on Modeling and Sensing Environmental Systems LNCC – Petrópolis – August 2008 Outline 1. Introduction 2. Growth rates of tropical rainforests a. Field measurements b. Regional monitoring (diagnostic) • Models • Merging models and remote sensing c. Prognostic d. Intercomparison of methodologies 3. Discussion, conclusion and future needs Introduction The most likely scenario for the next decades presents strong modifications in the global environment, including increase of the atmospheric concentration of CO2 and other trace gases, climate change and intensification of the impacts caused by the man's action. These changes may cause important modifications in the composition, structure and distribution of the ecosystems on the planet. A precise monitoring of the changes in the terrestrial biosphere is extremely important. Introduction The detection of the interannual variability and long-term trends in the ecosystems structure and dynamics will provide indications of change that would otherwise be unnoticed until the beginning of the transformation of the biome. The structure and composition of an ecosystem depend on its growth rate, or Net Primary Production (NPP) and mortality rate. Monitoring strategies of NPP: Field measurements Modeling Remote sensing + modeling Field measurements Atlantic Rainforest Conservation Reserve Aracruz Celulose SA – Teixeira de Freitas – BA Field measurements Measurements of yearly biomass increments: • Thousands of trees are measured per site • One measurement costs about US$ 50k Field measurements Diagnostic Models • 0-D models – Forced by local meteorological data and vegetation structure parameters – Expensive, given the need for local measurements – Results validated against local NPP field measurements – Monitoring ability limited by availability of local data Diagnostic Models • 2-D models – Forced by climate data and regional vegetation structure measurements taken from remote sensing – Monitoring ability limited by availability of: • climate data: near real time • Satellite data: continuity of sensors – New remote sensing data may demand algorithm upgrades Framework of the RATE algorithm FAPAR MODIS assimilated Land cover MODIS LAI MODIS assimilated Climate variables NCEP RATE algorithm 0,4º resolution NPP Algorithm Validation NPP field data Prognostic Models • Must predict their own future climate and also NPP and vegetation structure • This means a coupled climate-dynamic vegetation model • Careful validation against present time field measurements essential • Reconstruction of known past conditions and trends desirable Some data and results Assimilation of LAI and FAPAR LAI - KM67 - 2004 LAI - KM67 - 2004 2 2 7,00 7,00 6,00 6,00 5,00 4,00 LAI LAI 5,00 3,00 4,00 3,00 2,00 2,00 1,00 1,00 0,00 0,00 1 25 49 73 97 121 145 169 193 217 241 265 289 313 337 1 361 2 3 4 5 6 7 8 9 10 11 10 11 12 Mês do Ano Dia do Ano LAI MODIS LAI observado LAI assimilado FAPAR - KM67 - 2002 LAI assimilado FAPAR - KM67 - 2002 1,000 1,000 0,900 0,900 0,800 0,800 0,700 0,700 0,600 0,600 FAPAR FAPAR 2 (m /m ) 2 (m /m ) 0,500 0,400 0,300 0,500 0,400 0,300 0,200 0,200 0,100 0,100 0,000 1 25 49 73 97 121 145 169 193 217 241 265 Dia do Ano 289 313 337 361 0,000 1 2 3 4 5 6 7 8 9 Mês do Ano FAPAR MODIS FAPAR assimilado FAPAR observado FAPAR assimilado 12 Simulated NPP Simulated NPP Simulated NPP Intercomparison of methodologies • Two 0-D models forced by local met station data – IBIS – SITE • One 2-D model forced by climate and remote sensing data – RATE • One coupled climate-vegetation model – CCM3-IBIS Data availability Site City Coordinates Period Vegetation Data available Source Model evaluated Flona Tapajós K67 Belterra – PA 55.04º W; 2.86º 2004 Floresta tropical amazônica EM; NPP Turner et al. (2006) SITE-EM, SITE-RA RATE, IBIS-EM, IBIS-RA BA712 Teixeira de Freitas – BA 39.67º W; 17.29º S 2006 Mata atlântica EM; NPP Nunes (2008) SITE-EM, SITE-RA, RATE, IBIS-EM, IBIS-RA Flona Tapajós K67 Belterra – PA 55.02º W; 2.86º S 2001 Floresta tropical amazônica NPP Vieira et al. (2004) RATE UFAC Rio Branco – AC 68.03º W; 10.12º S 2001, 2002 Floresta tropical amazônica NPP Vieira et al. (2004) RATE ZF-2 Manaus – AM 60.18º W; 2.97º S 2001, 2002 Floresta tropical amazônica NPP Vieira et al. (2004) RATE Caxiuanã Melgaço – PA Várias parcelas entre 1°41’S e 1°46’S e entre 51°30’W e 51°23’W Longterm Floresta tropical amazônica NPP Malhi et al. (in press) CCM3-IBIS Flona Tapajós K67 Belterra – PA Várias parcelas entre 2°50’S e 3°18’S; entre 55°05’W e 54°55’W Longterm Floresta tropical amazônica NPP Malhi et al. (in press) CCM3-IBIS Manaus Manaus – AM Várias parcelas entre 2°20’S e 2°38’S; entre 60°15’W e 59°58’W Longterm Floresta tropical amazônica NPP Malhi et al. (in press) CCM3-IBIS 0-D models forced by met data NPP (kg -C m-2 yr -1 ) Site Year Model Observed Simulated Relative value value error Flona Tapajós K67 2004 SITE -EM 1.055 1.049 -0.6% Flon a Tapajós K67 2004 IBIS -EM 1.055 0.971 -8.0% BA712 2006 SITE -EM 1.366 1.214 -11.1% BA712 2006 IBIS -EM 1.366 1.458 6.7% Average relative error – 1.211 1.198 -3.3% Average | relative error | – – – 6.6% 2-D model forced by climate and remote sensing data NPP (kg-C m-2 year -1 ) Site Year Model Observed Simulated value value Error Flona Tapajós K 67 2001 RATE 1.230 1.338 8.8% Flona Tapajós K 67 2004 RATE 1.055 1.255 19.0% ZF-2 2001 RATE 1.063 1.226 15.3% ZF-2 2002 RATE 1.356 1.237 -8.8% UFAC 2001 RATE 1.343 1.307 -2.7% UFAC 2002 RATE 1.299 1.255 -3.4% BA712 2006 RATE 1.366 1.305 4.5% Average of the relative error – – 1.245 1.275 4.7% Average of | relative error | – – – – 8.9% Coupled climate-vegetation model NPP (kg -C m-2 year -1 ) Site Model Observed Simulated value value Erro r Manaus (long -term) CCM3 -IBIS 1.01 1.23 21.8% Flona Tapajós ( long -term) CCM3 -IBIS 1.44 1.16 -19.4% Caxiuanã ( long -term) CCM3 -IBIS 1.00 0.91 -9.0% Average of relative errors – 1.15 1.10 -4.3% Average of | relative error | – – – 16.7% Synthesis Group of models 0-D model s for ced by met station data 2-D m odels for ced by climate and Average of Maximum | error | | error | (%) (%) (%) 3.3 6.6 11.1 4.7 8.9 19.0 4.3 16. 7 21.8 | error average | remote sensing data Coupled climate -biosphere model Discussion • Results indicate that current 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%. • 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. Main Conclusion: State-of-the-art models are capable of routinely monitoring the net primary production of tropical ecosystems, with low errors expected for these estimates. Future needs/directions: • More validation points, particularly in other ecosystems • Test of the LAI and FAPAR assimilation algorithm for other ecosystems field data • Assimilation of other types of data – CERES (flew 2002): skin temperature – OCO/GOSAT (to fly 2008/2009): surface carbon balance – DESDynI (to fly 2010-2013): vegetation structure and biomass • Multi-model ensemble Thank you!