The effects of ignoring clustered data structure in allometric biomass models on large forest area biomass estimation
The aim of this study was to assess the effects of ignoring the clustered data structure on large area biomass estimation, when model uncertainty is included or not in the biomass prediction process. We used a Monte Carlo error propagation procedure to combine the uncertainty from allometric model predictions with the uncertainty from plot-to-plot variation, to produce estimates of mean AGB per hectare and standard error of the mean. An alternative procedure that ignores model prediction uncertainty was also used, therefore, showing only uncertainty due to differences between plots. Ignoring the clustered data structure, (i.e., fitting allometric models using ordinary least squares), the estimates of mean biomass per hectare were approximately 11% less than the estimates based on mixed effects models (that accounted for the clustered data structure), regardless of including or not the model prediction uncertainty. The estimates of uncertainty were also affected by ignoring the clustered data structure. When including model prediction uncertainty, ignoring the clustered data structure resulted in underestimation of standard error by 30%, whereas when model uncertainty was not included, the underestimation was 13%. Therefore, ignoring the clustered data structure, may affect both, the accuracy and the precision of biomass estimations over large forest areas. Keywords: Monitoring and data collection, Climate change ID: 3616826
| Main Author: | |
|---|---|
| Format: | Document biblioteca |
| Language: | English |
| Published: |
FAO ;
2022
|
| Online Access: | https://openknowledge.fao.org/handle/20.500.14283/cc4415en http://www.fao.org/3/cc4415en/cc4415en.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The aim of this study was to assess the effects of ignoring the clustered data structure on large area biomass estimation, when model uncertainty is included or not in the biomass prediction process. We used a Monte Carlo error propagation procedure to combine the uncertainty from allometric model predictions with the uncertainty from plot-to-plot variation, to produce estimates of mean AGB per hectare and standard error of the mean. An alternative procedure that ignores model prediction uncertainty was also used, therefore, showing only uncertainty due to differences between plots. Ignoring the clustered data structure, (i.e., fitting allometric models using ordinary least squares), the estimates of mean biomass per hectare were approximately 11% less than the estimates based on mixed effects models (that accounted for the clustered data structure), regardless of including or not the model prediction uncertainty. The estimates of uncertainty were also affected by ignoring the clustered data structure. When including model prediction uncertainty, ignoring the clustered data structure resulted in underestimation of standard error by 30%, whereas when model uncertainty was not included, the underestimation was 13%. Therefore, ignoring the clustered data structure, may affect both, the accuracy and the precision of biomass estimations over large forest areas.
Keywords: Monitoring and data collection, Climate change
ID: 3616826 |
|---|