Simulating carbon stocks and fluxes of the Amazon rainforest: a journey across temporal and spatial scales

Bitte benutzen Sie diese Kennung, um auf die Ressource zu verweisen:
https://osnadocs.ub.uni-osnabrueck.de/handle/urn:nbn:de:gbv:700-2018011916561
Open Access logo originally created by the Public Library of Science (PLoS)
Titel: Simulating carbon stocks and fluxes of the Amazon rainforest: a journey across temporal and spatial scales
Autor(en): Rödig, Edna
Erstgutachter: Prof. Dr. Andreas Huth
Zweitgutachter: Prof. Dr. Horst Malchow
Zusammenfassung: Global forests cover approximately 30% of land’s surface storing around 45% of above-ground terrestrial carbon. This carbon storage is constantly endangered by anthropogenic activities. Especially, tropical regions like the Amazon rainforest suffer from deforestation taking a great share in global CO2 emissions. In addition, forest dynamics are affected by climatic change like more frequent drought events. Quantifying the impact and feedback mechanisms of such climatic and anthropogenic changes on the global carbon cycle is still a great challenge. In this thesis, we developed a regionalization scheme to apply a forest gap model on the entire Amazon rainforest. Such a forest model has the advantage that it calculates forest growth at the individual tree level. It considers different successional states, that evolve form natural forest dynamics and disturbances, including information on tree height and species. The regionalized forest model thereby allows for integrating forest structure and species compositions into large-scale carbon analyses. The approach is independent of spatial scale and the simulation results can be linked to measurements from field inventory, eddy covariance, and remote sensing at local to continental scales. In a first study (chapter 2), we tested the capability of the forest model FORMIND to simulate gross primary production (GPP), respiration, and net ecosystem exchange (NEE) at daily and yearly time scales. The forest model was applied to spruce forests in Germany in order to analyze how the variability in environmental factors affects simulated carbon fluxes. Simulation results were compared to 6 years of eddy covariance (EC) data at a daily scale. The analysis shows that the forest model described the seasonal cycle of the carbon fluxes correctly, but estimated GPP differed from the observed data on days with extreme climatic conditions. Based on these findings, we developed two new parameterizations. One resulted from a numerical calibration against EC data. The other parameterization resulted from a method where EC data is filtered to extract the limiting factors for productivity. Thereby, new parameter values and even a new function for the temperature limitation of photosynthesis were found. The adopted forest model was then tested successfully at another spruce forest for cross validation. In general, the forest model reproduced the observed carbon fluxes of a forest ecosystem quite well. Although the overall performance of the calibrated model version was best, the filtering approach showed that calibrated parameter values did not necessarily correctly display the individual functional relations. The study has shown that the concept of simulating forest dynamics at the individual tree level is a valuable approach for simulating the NEE, GPP, and respiration of forest ecosystems. The focus of the second study (chapter 3) lied on the simulation of forest structure and above-ground biomass in the Amazon region with the forest model FORMIND. Estimating the spatial variation of biomass in the Amazon rainforest is challenging and, hence, a source of substantial uncertainty in the assessment of the global carbon cycle. On the one hand, estimates need to consider small-scale variations of forest structures due to natural tree mortality. On the other hand, it requires large-scale information on the state of the forest that can be detected by remote sensing. We, here, introduced a novel method that considered both aspects by linking the forest model and a wall-to-wall canopy height map derived from LIDAR remote sensing. The forest model was applied to estimate above-ground biomass stocks across the Amazon rainforest. This allowed for the direct comparison of simulated and observed canopy heights from remote sensing. The comparison enabled the detection of disturbed forest states from which we derived a biomass map of the Amazon rainforest at 0.16 ha resolution. Simulated biomass varied between 20 and 490 t(dry mass) ha-1 across 7.8 Mio km² of the Amazon rainforest (elevation < 1000 m). That equals a total above-ground biomass stock of 76 GtC with a strong spatial variation (coefficient of variation = 63%). The estimated biomass values fit estimates, that had been observed in 114 field inventories, well (deviation of only 15%). Beside biomass, the forest model allowed for estimating additional forest attributes such as basal area and stem density. The linkage of a forest model with a canopy height map allows for capturing forest structures at the individual to large scale. The approach is flexible and can also be combined with measurements of future satellite missions like ESA Biomass or GEDI. Hence, the study sets a basis for large-scale analyses of the heterogeneous structure of tropical forests and their carbon cycle. In a third study (chapter 4), we analyzed the interactions of productivity, biomass, and forest structure that are essential for understanding ecosystem’s response to climatic and anthropogenic changes. We here applied the forest model on the Amazon rainforest, combined simulation results with remotely-sensed data as in chapter 3, and additionally simulated ecosystem carbon fluxes. We found that the successional state of a forest has a strong influence on mean annual net ecosystem productivity (NEP), woody above-ground net primary production (wANPP), and net ecosystem productivity (NEP). These relations were used to derive maps of carbon fluxes at 0.16 ha resolutions (current state of the Amazon rainforest under spatial heterogenic environmental conditions). The Amazon was estimated to be a sink of atmospheric carbon with a mean NEP of 0.73 tC ha-1 a-1. Mean wANPP equals 4.16 tC ha-1 a-1 and GPP 25.2 tC ha-1 a-1. We found that forests in intermediate successional states are the most productive. Under current conditions, the Amazon rainforest takes up 0.59 PgC per year. This third study shows that forest structure and species compositions substantially influence productivity and biomass, and should not be neglected when estimating current carbon budgets or climate change scenarios for the Amazon rainforest. The findings of this thesis set a fundament for future analyses on carbon storage and fluxes of forests. Simulating at the tree level has the potential to investigate carbon dynamics from individual to continental scales. The regionalized forest model allows for the integration of different types of remotely sensed data in order to improve the spatial accuracy of estimates. The insights, we have gained from the eddy covariance study (chapter 2), help to investigate carbon dynamics of forests at continental scale also under changing climate. In combination with the regionalization approach (chapter 3 and 4), the findings of this thesis may be used to complement studies on drought events in forests and to understand feedback mechanisms caused by anthropogenic disturbances.
URL: https://osnadocs.ub.uni-osnabrueck.de/handle/urn:nbn:de:gbv:700-2018011916561
Schlagworte: forest modelling; forest biomass; carbon fluxes; remote sensing
Erscheinungsdatum: 19-Jan-2018
Lizenzbezeichnung: Namensnennung 3.0 Unported
URL der Lizenz: http://creativecommons.org/licenses/by/3.0/
Publikationstyp: Dissertation oder Habilitation [doctoralThesis]
Enthalten in den Sammlungen:FB06 - E-Dissertationen

Dateien zu dieser Ressource:
Datei Beschreibung GrößeFormat 
thesis_roedig.pdfPräsentationsformat52,13 MBAdobe PDF
thesis_roedig.pdf
Miniaturbild
Öffnen/Anzeigen


Diese Ressource wurde unter folgender Copyright-Bestimmung veröffentlicht: Lizenz von Creative Commons Creative Commons