How can we best constrain terrestrial biosphere model carbon cycle predictions using satellite remote sensing data?
Abstract
Predicting the fate of the terrestrial carbon, C, sink under future global change strongly relies on our ability to accurately model global scale vegetation dynamics and surface CO2fluxes. However, terrestrial biosphere model (TBM) leaf phenology and C cycle processes remain subject to large uncertainties, partly because of unknown or poorly calibrated parameters.
Satellite remote sensing (RS) data offer us the possibility to optimize these model parameters for the dominant plant functional types over large spatial scales and in regions that have limited field-based data. Here, we present several past studies in which we have used RS data to optimize ORCHIDEE TBM simulations of vegetation phenology and gross C uptake (GPP). We have used our state-of-the art C cycle data assimilation (DA) system to investigate the constraint on ecosystem to global seasonal and annual GPP brought by assimilating various different RS products, including the normalized difference vegetation index (NDVI) and solar induced chlorophyll fluorescence (SIF). Via in situ DA experiments, we also explore the potential of assimilating upcoming satellite biomass estimates. We review our DA studies in terms of the impact on key characteristics of the global C cycle - such as long-term greening trends, inter-annual variability (IAV) in land gross C uptake, and the distribution of global GPP between the northern extratropical versus tropical regions. Throughout, we discuss the DA challenges we have encountered - particularly associated to simultaneously assimilating multiple datasets in an optimal framework to constrain different model processes. We examine our work in the context of these challenges, and propose solutions for the community going forward, including: i) how we can address biases related to inconsistencies between what is observed and how the relevant processes are represented in a model; ii) how we can use new RS technologies to leverage information across spatial scales; iii) how we can better utilize high resolution RS data for improving modeled vegetation fractional cover and leaf phenology, especially in spatially heterogeneous dryland ecosystems that dominate C cycle IAV; and iv) how we may fuse a range of RS data within C cycle DA systems to test and constrain the latest generation of dynamic vegetation demographic models. Predicting the fate of the terrestrial carbon, C, sink under future global change strongly relies on our ability to accurately model global scale vegetation dynamics and surface CO2fluxes. However, terrestrial biosphere model (TBM) leaf phenology and C cycle processes remain subject to large uncertainties, partly because of unknown or poorly calibrated parameters. Satellite remote sensing (RS) data offer us the possibility to optimize these model parameters for the dominant plant functional types over large spatial scales and in regions that have limited field-based data. Here, we present several past studies in which we have used RS data to optimize ORCHIDEE TBM simulations of vegetation phenology and gross C uptake (GPP). We have used our state-of-the art C cycle data assimilation (DA) system to investigate the constraint on ecosystem to global seasonal and annual GPP brought by assimilating various different RS products, including the normalized difference vegetation index (NDVI) and solar induced chlorophyll fluorescence (SIF). Via in situ DA experiments, we also explore the potential of assimilating upcoming satellite biomass estimates. We review our DA studies in terms of the impact on key characteristics of the global C cycle - such as long-term greening trends, inter-annual variability (IAV) in land gross C uptake, and the distribution of global GPP between the northern extratropical versus tropical regions. Throughout, we discuss the DA challenges we have encountered - particularly associated to simultaneously assimilating multiple datasets in an optimal framework to constrain different model processes. We examine our work in the context of these challenges, and propose solutions for the community going forward, including: i) how we can address biases related to inconsistencies between what is observed and how the relevant processes are represented in a model; ii) how we can use new RS technologies to leverage information across spatial scales; iii) how we can better utilize high resolution RS data for improving modeled vegetation fractional cover and leaf phenology, especially in spatially heterogeneous dryland ecosystems that dominate C cycle IAV; and iv) how we may fuse a range of RS data within C cycle DA systems to test and constrain the latest generation of dynamic vegetation demographic models.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.B11G2335M
- Keywords:
-
- 0428 Carbon cycling;
- BIOGEOSCIENCES;
- 0439 Ecosystems;
- structure and dynamics;
- BIOGEOSCIENCES;
- 0480 Remote sensing;
- BIOGEOSCIENCES;
- 1615 Biogeochemical cycles;
- processes;
- and modeling;
- GLOBAL CHANGE