Boston University
Strengthening feedbacks between models and data to improve predictions of the terrestrial carbon sink
Abstract
dc:description.abstractCurrent terrestrial ecosystem models represent a sophisticated understanding of the physical, chemical and biological processes that govern the world’s terrestrial ecosystems. However, model predictions of terrestrial carbon uptake in the next century disagree widely, ranging from a net sink to a net source. To improve predictions, models and observational data must be better integrated (1) to test the assumptions and hypotheses on which models are based and (2) to quantify the uncertainty around model predictions. Iteratively using data to constrain models and models to inform which data to collect ensures that model-data synthesis projects can maximize efficiency and efficacy. This model-data `feedback loop' approach provides a roadmap for this dissertation: each chapter addresses bottlenecks faced by the scientific community in performing model-data synthesis. Chapter One outlines the current challenges in data ingest reproducibility, generalizability and scalability and provides solutions by introducing a suite of data ingest tools based on data standards, as well as a workflow for data processing and provenance. As a test case, the data ingest workflow is used in preparing a host of meteorological data products as drivers for ecological models. Chapter Two focuses on model validation and benchmarking. To facilitate efficient, robust and repeatable model intercomparison projects, a benchmarking system is developed that ingests observational data sets and compares them against ensemble model outputs using a diverse suite of metrics. These tools are used to revisit the work of the FACE Model-Data Synthesis Project, incorporating quantifications of model uncertainty into benchmarking assessments. Chapter Three demonstrates the use of iterative uncertainty analysis to inform model-data synthesis projects. This approach proves useful when assessing the effects of adding mechanistic hydraulic process complexity to the Ecosystem Demography model (ED2.2) on predictive uncertainty. Prioritizing the acquisition of data to constrain parameters dominating model uncertainty optimizes time and computational energy. As a whole, this dissertation serves a guide for designing and executing model-data synthesis projects. Making model-data synthesis accessible to both empiricists and modelers will advance the development of terrestrial ecosystem models and improve predictions of future terrestrial carbon uptake.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Earth & Environment
- Grantor
- Boston University
- Year dc:date.issued
- 2021
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Cowdery, Elizabeth
- Advisor dc:contributor.advisor
-
- Dietze, Michael
- ORCID dc:identifier.orcid
Subjects
dc:subject × 4Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/2144/53526