Improving the modeling of tree establishment in LPJ-GUESS by introducing conspecific clustering
(2025) In Master’s Theses in Mathematical Sciences BERM02 20251Mathematics (Faculty of Sciences)
- Abstract
- This thesis explores the spatial distribution of trees in the LPJ-GUESS dynamic global vegetation model and proposes some modification to improve ecological realism. In the standard version of LPJ-GUESS, the model simplifies forest dynamics by averaging individuals into cohorts and disregarding competition, canopy interactions or resource allocation completely. To improve this, a newer version was developed, LPJ-GUESS_{expl}, that explicitly places trees along the circumference of a circle, which then represents a patch. On this ring, competition and canopy interactions, as well as resource allocation, take place. This addition to the model is important for ecological realism because these factors strongly shape the forest structure.... (More)
- This thesis explores the spatial distribution of trees in the LPJ-GUESS dynamic global vegetation model and proposes some modification to improve ecological realism. In the standard version of LPJ-GUESS, the model simplifies forest dynamics by averaging individuals into cohorts and disregarding competition, canopy interactions or resource allocation completely. To improve this, a newer version was developed, LPJ-GUESS_{expl}, that explicitly places trees along the circumference of a circle, which then represents a patch. On this ring, competition and canopy interactions, as well as resource allocation, take place. This addition to the model is important for ecological realism because these factors strongly shape the forest structure. Including such spatial interactions allows the model to reflect real life ecosystem processes that influence the productivity and resilience of the vegetation.
The modifications suggested here will continue on the LPJ-GUESS_{expl} model by introducing more realistic distribution functions to tree placement than the current uniform random placements in LPJ-GUESS_{expl}. To explore realistic clustering of the same species, some data sets from measured forest stands will also be analyzed using point pattern analysis. The two suggested modifications are: (1) an exponential distribution was used instead of a uniform distribution and (2) the distance to the nearest same species neighbor was used in the exponent to adjust the uniform sampled probability with an exponential probability.
Tests were also performed to explore whether larger patch areas improve the point pattern realism of the LPJ-GUESS model. To properly analyse the model output, a fractal approach was taken. Since the output is one-dimensional and point pattern analysis is applied to two-dimensional data, a transformation was required to address this difference. This then also allows comparison between real-world data analysis and model output analysis.
The results demonstrate that the suggested modification improves the realism of the clustering in the model. It is however not quite up to the extent of clustering seen in real forest stands. Capturing realistic clustering is important because it influences competition intensity, light distribution, and ultimately the structure and dynamics of forest stands. Thus if the clustering is not quite as seen in real forest stands, then other processes might also be over or underestimated compared to the real stands. (Less) - Popular Abstract
- Forests as we know them are not random. The placement of the trees is influenced by the trees around them since they compete for things such as sunlight, nutrition and water. There are computer models that simulate how a forest grows, like LPJ-GUESS. This is a model that can simulate vegetation globally. To be able to do this, it needs to simplify some matters that could be important ecological processes.
This thesis aims to improve the realism of the LPJ-GUESS model by introducing some clustering of trees of the same species. Some tree species prefer growing near their own species while other prefer to grow further apart. Mathematical methods such as fractal analysis, point pattern analysis and exponential probabilities will be used to... (More) - Forests as we know them are not random. The placement of the trees is influenced by the trees around them since they compete for things such as sunlight, nutrition and water. There are computer models that simulate how a forest grows, like LPJ-GUESS. This is a model that can simulate vegetation globally. To be able to do this, it needs to simplify some matters that could be important ecological processes.
This thesis aims to improve the realism of the LPJ-GUESS model by introducing some clustering of trees of the same species. Some tree species prefer growing near their own species while other prefer to grow further apart. Mathematical methods such as fractal analysis, point pattern analysis and exponential probabilities will be used to modify and analyse the output of the changed model. This development in the model can help simulate forest dynamics more realistically, helping us better understand and predict how forests respond to environmental change. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9213328
- author
- Bogers, Evelien LU
- supervisor
-
- Stefan Olin LU
- organization
- course
- BERM02 20251
- year
- 2025
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- LPJ-GUESS, conspecific clustering, spatial patterns, vegetation model, exponential sampling, tree distribution
- publication/series
- Master’s Theses in Mathematical Sciences
- report number
- LUNFBV-3008-2025
- ISSN
- 1404-6342
- other publication id
- 2025:E101
- language
- English
- id
- 9213328
- date added to LUP
- 2025-12-02 15:34:03
- date last changed
- 2025-12-02 15:34:03
@misc{9213328,
abstract = {{This thesis explores the spatial distribution of trees in the LPJ-GUESS dynamic global vegetation model and proposes some modification to improve ecological realism. In the standard version of LPJ-GUESS, the model simplifies forest dynamics by averaging individuals into cohorts and disregarding competition, canopy interactions or resource allocation completely. To improve this, a newer version was developed, LPJ-GUESS_{expl}, that explicitly places trees along the circumference of a circle, which then represents a patch. On this ring, competition and canopy interactions, as well as resource allocation, take place. This addition to the model is important for ecological realism because these factors strongly shape the forest structure. Including such spatial interactions allows the model to reflect real life ecosystem processes that influence the productivity and resilience of the vegetation.
The modifications suggested here will continue on the LPJ-GUESS_{expl} model by introducing more realistic distribution functions to tree placement than the current uniform random placements in LPJ-GUESS_{expl}. To explore realistic clustering of the same species, some data sets from measured forest stands will also be analyzed using point pattern analysis. The two suggested modifications are: (1) an exponential distribution was used instead of a uniform distribution and (2) the distance to the nearest same species neighbor was used in the exponent to adjust the uniform sampled probability with an exponential probability.
Tests were also performed to explore whether larger patch areas improve the point pattern realism of the LPJ-GUESS model. To properly analyse the model output, a fractal approach was taken. Since the output is one-dimensional and point pattern analysis is applied to two-dimensional data, a transformation was required to address this difference. This then also allows comparison between real-world data analysis and model output analysis.
The results demonstrate that the suggested modification improves the realism of the clustering in the model. It is however not quite up to the extent of clustering seen in real forest stands. Capturing realistic clustering is important because it influences competition intensity, light distribution, and ultimately the structure and dynamics of forest stands. Thus if the clustering is not quite as seen in real forest stands, then other processes might also be over or underestimated compared to the real stands.}},
author = {{Bogers, Evelien}},
issn = {{1404-6342}},
language = {{eng}},
note = {{Student Paper}},
series = {{Master’s Theses in Mathematical Sciences}},
title = {{Improving the modeling of tree establishment in LPJ-GUESS by introducing conspecific clustering}},
year = {{2025}},
}