Advanced

LPJ-GM 1.0 : Simulating migration efficiently in a dynamic vegetation model

Lehsten, Veiko LU ; Mischurow, Michael LU ; Lindström, Erik LU ; Lehsten, Dorte LU and Lischke, Heike (2019) In Geoscientific Model Development 12(3). p.893-908
Abstract

Dynamic global vegetation models are a common tool to assess the effect of climate and land use change on vegetation. Though most applications of dynamic global vegetation models use plant functional types, some also simulate species occurrences. While the current development aims to include more processes, e.g. the nitrogen cycle, the models still typically assume an ample seed supply allowing all species to establish once the climate conditions are suitable. Pollen studies have shown that a number of plant species lag behind in occupying climatological suitable areas (e.g. after a change in the climate) as they need to arrive at and establish in the newly suitable areas. Previous attempts to implement migration in dynamic vegetation... (More)

Dynamic global vegetation models are a common tool to assess the effect of climate and land use change on vegetation. Though most applications of dynamic global vegetation models use plant functional types, some also simulate species occurrences. While the current development aims to include more processes, e.g. the nitrogen cycle, the models still typically assume an ample seed supply allowing all species to establish once the climate conditions are suitable. Pollen studies have shown that a number of plant species lag behind in occupying climatological suitable areas (e.g. after a change in the climate) as they need to arrive at and establish in the newly suitable areas. Previous attempts to implement migration in dynamic vegetation models have allowed for the simulation of either only small areas or have been implemented as a post-process, not allowing for feedbacks within the vegetation. Here we present two novel methods simulating migrating and interacting tree species which have the potential to be used for simulations of large areas. Both distribute seeds between grid cells, leading to individual establishment. The first method uses an approach based on fast Fourier transforms, while in the second approach we iteratively shift the seed production matrix and disperse seeds with a given probability. While the former method is computationally faster, it does not allow for modification of the seed dispersal kernel parameters with respect to terrain features, which the latter method allows.

We evaluate the increase in computational demand of both methods. Since dispersal acts at a scale no larger than 1 km, all dispersal simulations need to be performed at maximum at that scale. However, with the currently available computational power it is not feasible to simulate the local vegetation dynamics of a large area at that scale. We present an option to decrease the required computational costs through a reduction in the number of grid cells for which the local dynamics are simulated only along migration transects. Evaluation of species patterns and migration speeds shows that simulating along transects reduces migration speed, and both methods applied on the transects produce reasonable results. Furthermore, using the migration transects, both methods are sufficiently computationally efficient to allow for large-scale DGVM simulations with migration.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
in
Geoscientific Model Development
volume
12
issue
3
pages
16 pages
publisher
Copernicus Gesellschaft Mbh
external identifiers
  • scopus:85062691207
ISSN
1991-959X
DOI
10.5194/gmd-12-893-2019
language
English
LU publication?
yes
id
9f677265-f001-42bf-8aa5-f760a57ddc54
date added to LUP
2019-03-19 11:08:43
date last changed
2019-04-12 10:54:21
@article{9f677265-f001-42bf-8aa5-f760a57ddc54,
  abstract     = {<p>Dynamic global vegetation models are a common tool to assess the effect of climate and land use change on vegetation. Though most applications of dynamic global vegetation models use plant functional types, some also simulate species occurrences. While the current development aims to include more processes, e.g. the nitrogen cycle, the models still typically assume an ample seed supply allowing all species to establish once the climate conditions are suitable. Pollen studies have shown that a number of plant species lag behind in occupying climatological suitable areas (e.g. after a change in the climate) as they need to arrive at and establish in the newly suitable areas. Previous attempts to implement migration in dynamic vegetation models have allowed for the simulation of either only small areas or have been implemented as a post-process, not allowing for feedbacks within the vegetation. Here we present two novel methods simulating migrating and interacting tree species which have the potential to be used for simulations of large areas. Both distribute seeds between grid cells, leading to individual establishment. The first method uses an approach based on fast Fourier transforms, while in the second approach we iteratively shift the seed production matrix and disperse seeds with a given probability. While the former method is computationally faster, it does not allow for modification of the seed dispersal kernel parameters with respect to terrain features, which the latter method allows. </p><p>We evaluate the increase in computational demand of both methods. Since dispersal acts at a scale no larger than 1&amp;thinsp;km, all dispersal simulations need to be performed at maximum at that scale. However, with the currently available computational power it is not feasible to simulate the local vegetation dynamics of a large area at that scale. We present an option to decrease the required computational costs through a reduction in the number of grid cells for which the local dynamics are simulated only along migration transects. Evaluation of species patterns and migration speeds shows that simulating along transects reduces migration speed, and both methods applied on the transects produce reasonable results. Furthermore, using the migration transects, both methods are sufficiently computationally efficient to allow for large-scale DGVM simulations with migration.</p>},
  author       = {Lehsten, Veiko and Mischurow, Michael and Lindström, Erik and Lehsten, Dorte and Lischke, Heike},
  issn         = {1991-959X},
  language     = {eng},
  month        = {03},
  number       = {3},
  pages        = {893--908},
  publisher    = {Copernicus Gesellschaft Mbh},
  series       = {Geoscientific Model Development},
  title        = {LPJ-GM 1.0 : Simulating migration efficiently in a dynamic vegetation model},
  url          = {http://dx.doi.org/10.5194/gmd-12-893-2019},
  volume       = {12},
  year         = {2019},
}