Multiple sources of uncertainty in high-latitude terrestrial ecosystem modeling

Author

• Hannah Mevenkamp

Date of Award

12-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Statement

Doctor of Philosophy in Biological Sciences

Department

Department of Biology and Wildlife

Granting Institution

University of Alaska Fairbanks

Committee Chair

Euskirchen, Eugenie

Committee Member

Genet, Hélène

Committee Member

Bhatt, Uma

Committee Member

Serbin, Shawn

Committee Member

Bret-Harte, Syndonia

Table of Contents

Chapter 1: General introduction.

1.1 Approaches to certainty and uncertainty in science history

1.1.1 Classical worldviews - uncovering the certain, objective truth

1.1.2 The introduction of essential uncertainty

1.2 Classifications of uncertainty in science

1.2.1 Aleatory, epistemic, and ontological uncertainty

1.2.2 Scientific, deficient, consensus, and technical uncertainty

1.3 (Uncertainty in) terrestrial arctic ecosystem modeling, or: from certainty to uncertainty to adequacy

1.3.1 Uncertainties in modeling

1.3.2 (Modeling) the Arctic (with DVM-DOS-TEM)

1.4 The human dimension of uncertainty

1.4.1 Communication

1.5 Overview of manuscript chapters

1.5.1 Chapter 2: Reducing uncertainty of high-latitude ecosystem models through identification of key parameters

1.5.2 Chapter 3: How initial conditions-, structural-, and parameter-based model uncertainty interact and influence predictions in permafrost ecosystems

1.5.3 Chapter 4: Modeling the post-fire recovery dynamics of the Anaktuvuk River fire with the model DVM-DOS-TEM

1.6 References. Chapter 2: Reducing uncertainty of high-latitude ecosystem models through identification of key parameters.

2.1 Abstract

2.2 Introduction

2.3 Methods

2.3.1 Building a causal loop diagram

2.3.2 Models and databases

2.3.3 Causal loop diagram evaluation of model parameters

2.4 Results

2.4.1 CLD analysis

2.4.2 Database analysis

2.4.3 Combined analysis

2.5 Discussion

2.5.1 Overview

2.5.2 Model process and structural uncertainty: implications

2.5.3 Parameter uncertainty and data informed parameters: implications

2.5.4 Pa-factor and parameter prioritization

2.6 Conclusion

2.7 Acknowledgments

2.8 References. Chapter 3: How initial conditions-, structural-, and parameter-based model uncertainty interact and influence in permafrost ecosystems.

3.1 Abstract

3.2 Introduction

3.3 Methods

3.3.1 Model description

3.3.2 Study sites

3.3.3 Initial conditions-based uncertainty

3.3.3.1 Paleo temperature scenarios

3.3.3.2 Other climatic factors

3.3.4 Structural-based uncertainty

3.3.4.1 Soil depth

3.3.4.2 Latent heat

3.3.4.3 Geothermal gradient

3.3.5 Parameter uncertainty

3.3.5.1 Sensitivity and uncertainty analysis

3.3.5.2 Parameter selection

3.4 Results

3.4.1 Initial conditions- and structural-based uncertainty/ecosystem measures

3.4.1.1 Vegetation and ecosystem respiration dynamics

3.4.1.2 Carbon storage and carbon balance

3.4.1.3 Soil temperatures

3.4.2 Parameter-based uncertainty/model performance measures

3.4.2.1 Multi-variation variance

3.4.2.2 Uncertainty and sensitivity

3.4.2.3 Covariance term

3.5 Discussion and conclusion

3.5.1 Initial conditions- and structural-based uncertainty

3.5.2 Parameter-based uncertainty

3.5.3 Considerations on the community types and soil processes modeled in this study

3.5.4 Implications and conclusions

3.6 Acknowledgements

3.7 References. Chapter 4: Modeling the post-fire recovery dynamics of the Anaktuvuk River fire with the model DVM-DOS-TEM.

4.1 Abstract

4.2 Introduction

4.2.1 Post-fire succession and dynamics

4.2.2 Impacts of tundra fires on communities - lichen ecological and cultural relevance

4.2.3 Tundra fire modeling

4.3 Methods

4.3.1 Site - the Anaktuvuk River fire

4.3.2 Calibrating DVM-DOS-TEM

4.3.3 Structural changes to f(foliage)

4.3.4 Fire modeling and evaluation

4.4 Results

4.4.1 Net carbon uptake

4.4.2 Vegetation carbon

4.4.3 Soil thermal dynamics

4.4.3.1 Soil temperature

4.4.3.2 Active layer depth

4.5 Discussion and conclusion

4.5.1 Soil

4.5.2 Vegetation

4.5.3 Implications and conclusions

4.6 References. Chapter 5: General conclusions.

5.1 References.

Abstract

Terrestrial biosphere modeling in high latitudes is of high relevance, especially as the region is undergoing rapid changes with significant local and global consequences. Modeling pre­dictions are often the foundation of policy decisions, and thus their reliability needs to be assessed. As model uncertainties arise from several sources and different steps of the modeling process, a holistic approach should be taken to ensure a realistic assessment of all possible outcomes, as well as the best chances of model improvement. This includes the re-framing of uncertainty as an opportunity for growth, as well as studying the interactions of different sources of model uncertainty. Chapter 1 provides an overview of different perspectives on and classifications of uncertainty in science and modeling in particular. In Chapter 2, I combine a data availability and needs analysis with a structural network analysis of a causal loop dia­gram which assesses the parameters structural vulnerability. The resulting measure gauges a parameters need for additional field observations based on parameter- and structural-based uncertainty. In Chapter 3, I additionally analyze initial conditions-based uncertainty, which is often overlooked but may be more relevant than assumed in high-latitude ecosystems underlain by permafrost. I find that the Terrestrial ecosystem model with dynamic vegeta­tion and dynamic organic soil (DVM-DOS-TEM) preserves initial conditions from the last glacial maximum into the year 2100, and that structural choices impact the propagation. Additionally, parameter-based uncertainty is modulated by both structural choices and ini­tial conditions. In Chapter 4, I move to a specific assessment of whether DVM-DOS-TEM is adequate to represent the post-fire tundra ecosystem. I incorporate structural changes to better represent GPP for small vegetation carbon, which in particular improves the repre­sentation of Lichen recovery post-fire, which is of high local relevance. I conclude with a general discussion of uncertainty assessments and model improvement in Chapter 5.

Recommended Citation

Mevenkamp, Hannah, "Multiple sources of uncertainty in high-latitude terrestrial ecosystem modeling" (2025). Biological Sciences. 529.
https://scholarworks.alaska.edu/uaf_grad_bio_sciences/529