- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Quasi-experimental methods for wildfire impact quantification...
Open Collections
UBC Theses and Dissertations
UBC Theses and Dissertations
Quasi-experimental methods for wildfire impact quantification : applications of distance-adjusted propensity score matching to forest inventory data Woo, Hyeyoung
Abstract
Quantifying wildfire impacts on forest ecosystems is challenging due to the lack of pre-fire data or controls from experiments over a large landscape. Quasi-experimental methods have been popular in various fields of science where experiments are difficult to implement. However, the application of quasi-experimental methods to ecological data have not yet been fully explored. In this dissertation, I applied quasi-experimental methods to quantify wildfire impacts on aboveground forest woody carbon mass using national forest inventory data from the United States of America (USA) and British Columbia (BC), Canada. First, I compared distance-adjusted propensity score matching (DAPSM) with propensity score matching (PSM) and spatial matching (SM) to quantify the changes in forest woody carbon mass due to wildfires in Washington and Oregon, USA. Incorporating spatial information in addition to environmental covariates was essential to account for both observed and unobserved environmental covariates in matching. Thus, DAPSM was favored over PSM and SM. Second, I conducted a sensitivity analysis on the performance of DAPSM with different data availability to provide a practical guide of sample size and environmental covariates required to quantify wildfire impacts. I found that the inclusion of the spatial distance compensated for the omission of key covariates, but this compensation was not effective for small sample sizes. Third, I applied DAPSM with and without replacement to three datasets with small sample sizes collected for case-studies of wildfire impacts in south-central BC. DAPSM with replacement using BC forest inventory plot data enabled balancing the environmental covariates between burned and control plots under certain circumstances. The controls produced by DAPSM captured the trends in the amount of woody carbon masses under different fire severities, implying that they may replace the pre-burn data once the propensity scores are adequately addressed. Overall, the implementation of DAPSM allowed the assessment of wildfire impacts on forest carbon by building a causal relationship from observational forest inventory data. Based on applied examples, this dissertation provides guidelines for employing propensity score matching to quantify the impacts of natural disturbances. This research contributes to future studies considering quasi-experimental approaches for analyzing ecological data where controlled experiments are impossible.
Item Metadata
Title |
Quasi-experimental methods for wildfire impact quantification : applications of distance-adjusted propensity score matching to forest inventory data
|
Creator | |
Supervisor | |
Publisher |
University of British Columbia
|
Date Issued |
2021
|
Description |
Quantifying wildfire impacts on forest ecosystems is challenging due to the lack of pre-fire data or controls from experiments over a large landscape. Quasi-experimental methods have been popular in various fields of science where experiments are difficult to implement. However, the application of quasi-experimental methods to ecological data have not yet been fully explored. In this dissertation, I applied quasi-experimental methods to quantify wildfire impacts on aboveground forest woody carbon mass using national forest inventory data from the United States of America (USA) and British Columbia (BC), Canada.
First, I compared distance-adjusted propensity score matching (DAPSM) with propensity score matching (PSM) and spatial matching (SM) to quantify the changes in forest woody carbon mass due to wildfires in Washington and Oregon, USA. Incorporating spatial information in addition to environmental covariates was essential to account for both observed and unobserved environmental covariates in matching. Thus, DAPSM was favored over PSM and SM.
Second, I conducted a sensitivity analysis on the performance of DAPSM with different data availability to provide a practical guide of sample size and environmental covariates required to quantify wildfire impacts. I found that the inclusion of the spatial distance compensated for the omission of key covariates, but this compensation was not effective for small sample sizes.
Third, I applied DAPSM with and without replacement to three datasets with small sample sizes collected for case-studies of wildfire impacts in south-central BC. DAPSM with replacement using BC forest inventory plot data enabled balancing the environmental covariates between burned and control plots under certain circumstances. The controls produced by DAPSM captured the trends in the amount of woody carbon masses under different fire severities, implying that they may replace the pre-burn data once the propensity scores are adequately addressed.
Overall, the implementation of DAPSM allowed the assessment of wildfire impacts on forest carbon by building a causal relationship from observational forest inventory data. Based on applied examples, this dissertation provides guidelines for employing propensity score matching to quantify the impacts of natural disturbances. This research contributes to future studies considering quasi-experimental approaches for analyzing ecological data where controlled experiments are impossible.
|
Genre | |
Type | |
Language |
eng
|
Date Available |
2022-01-06
|
Provider |
Vancouver : University of British Columbia Library
|
Rights |
Attribution-NonCommercial-NoDerivatives 4.0 International
|
DOI |
10.14288/1.0406182
|
URI | |
Degree | |
Program | |
Affiliation | |
Degree Grantor |
University of British Columbia
|
Graduation Date |
2022-05
|
Campus | |
Scholarly Level |
Graduate
|
Rights URI | |
Aggregated Source Repository |
DSpace
|
Item Media
Item Citations and Data
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International