Are More Complex Physiological Models of Forest Ecosystems Better Choices for Plot and Regional Predictions?. Jin, W., He, H. S., & Thompson, F. R. 75:1–14. ![Are More Complex Physiological Models of Forest Ecosystems Better Choices for Plot and Regional Predictions? [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex [Highlights] [::] We evaluated performance of process-based forest ecosystem models. [::] A complex physiological model performed best at the plot scale. [::] A hybrid empirical-physiological model performed best at the regional scale. [Abstract] We evaluated performance of process-based forest ecosystem models. A complex physiological model performed best at the plot scale. A hybrid empirical-physiological model performed best at the regional scale. Process-based forest ecosystem models vary from simple physiological, complex physiological, to hybrid empirical-physiological models. Previous studies indicate that complex models provide the best prediction at plot scale with a temporal extent of less than 10 years, however, it is largely untested as to whether complex models outperform the other two types of models at plot and regional scale in longer timeframe (i.e. decades). We compared model predictions of aboveground carbon by one representative model of each model type (PnET-II, ED2 and LINKAGES v2.2, respectively) with field data (19-77 years) at both scales in the Central Hardwood Forests of the United States. At plot scale, predictions by complex physiological model were the most concordant with field data, suggesting that physiological processes are more influential than forest composition and structure. Hybrid model provided the best predictions at regional scale, suggesting that forest composition and structure may be more influential than physiological processes. [Excerpt: Regional scale comparisons] The percent bias of all models was larger at the regional scale than the plot scale. Abiotic environmental heterogeneity at the regional scale could be one of the factors contributing to the larger percent bias at the regional scale. Even though we used ecological subsections as our regional scale study areas-areas where the vegetation and environmental factors are considered relatively homogenous (McNab et al., 2007), the environmental heterogeneity within each subsection would still be higher than that at each plot-scale site. Small-scale abiotic environmental variations, such as difference in water availability at different slope positions of the same soil type, have been ignored in this regional scale study, and summarized environmental factors were used to represent the average physical situations across the entire subsection. [...] [\n] Predictions based on empirical relationships like those in the hybrid model may not hold true under changing environments in the future, since those relationships were established based on observations in the past (Gustafson, 2013 and Cuddington et al., 2013). However, predictions based on empirical life history attributes might retain validity in the future due to niche conservatism (Crisp et al., 2009 and Wiens et al., 2010). For example, the tolerance range of growing degree days of a given plant species may remain largely constant despite climate change. [\n] None of the models we examined simulate forest landscape processes, which are spatially continuous and temporally dynamic processes (e.g., fire disturbance). Forest landscape processes are likely to have greater contribution to forest ecosystem responses than climate variables alone (Gustafson et al., 2010, Kurz et al., 2008, Girardin and Mudelsee, 2008 and Li et al., 2013). Therefore, greater bias could occur if forest landscape processes are not included in the prediction of forest ecosystem dynamics (Reynolds et al., 2001). [...] [\n] Superiority of complex physiological model at the plot scale was achieved at the price of more detailed input data, longer time of simulation, and more simplified representation of forest composition. While the hybrid model was the best model at the regional scale, it cannot provide carbon dynamics at a fine temporal scale (e.g., daily carbon sequestration) like the complex physiological model can. The simple physiological model provided the worst prediction. Although we primarily focused on density of aboveground woody biomass, other traits associated with forest ecosystems (e.g., forest composition, basal area) are often of interest in ecosystem and landscape modeling. Data preparation and simulation time are also often, if not always, of concern.
@article{jinAreMoreComplex2016,
title = {Are More Complex Physiological Models of Forest Ecosystems Better Choices for Plot and Regional Predictions?},
author = {Jin, Wenchi and He, Hong S. and Thompson, Frank R.},
date = {2016-01},
journaltitle = {Environmental Modelling \& Software},
volume = {75},
pages = {1--14},
issn = {1364-8152},
doi = {10.1016/j.envsoft.2015.10.004},
url = {https://doi.org/10.1016/j.envsoft.2015.10.004},
abstract = {[Highlights]
[::] We evaluated performance of process-based forest ecosystem models. [::] A complex physiological model performed best at the plot scale. [::] A hybrid empirical-physiological model performed best at the regional scale.
[Abstract]
We evaluated performance of process-based forest ecosystem models. A complex physiological model performed best at the plot scale. A hybrid empirical-physiological model performed best at the regional scale. Process-based forest ecosystem models vary from simple physiological, complex physiological, to hybrid empirical-physiological models. Previous studies indicate that complex models provide the best prediction at plot scale with a temporal extent of less than 10 years, however, it is largely untested as to whether complex models outperform the other two types of models at plot and regional scale in longer timeframe (i.e. decades). We compared model predictions of aboveground carbon by one representative model of each model type (PnET-II, ED2 and LINKAGES v2.2, respectively) with field data (19-77 years) at both scales in the Central Hardwood Forests of the United States. At plot scale, predictions by complex physiological model were the most concordant with field data, suggesting that physiological processes are more influential than forest composition and structure. Hybrid model provided the best predictions at regional scale, suggesting that forest composition and structure may be more influential than physiological processes.
[Excerpt: Regional scale comparisons]
The percent bias of all models was larger at the regional scale than the plot scale. Abiotic environmental heterogeneity at the regional scale could be one of the factors contributing to the larger percent bias at the regional scale. Even though we used ecological subsections as our regional scale study areas-areas where the vegetation and environmental factors are considered relatively homogenous (McNab et al., 2007), the environmental heterogeneity within each subsection would still be higher than that at each plot-scale site. Small-scale abiotic environmental variations, such as difference in water availability at different slope positions of the same soil type, have been ignored in this regional scale study, and summarized environmental factors were used to represent the average physical situations across the entire subsection. [...]
[\textbackslash n] Predictions based on empirical relationships like those in the hybrid model may not hold true under changing environments in the future, since those relationships were established based on observations in the past (Gustafson, 2013 and Cuddington et al., 2013). However, predictions based on empirical life history attributes might retain validity in the future due to niche conservatism (Crisp et al., 2009 and Wiens et al., 2010). For example, the tolerance range of growing degree days of a given plant species may remain largely constant despite climate change.
[\textbackslash n] None of the models we examined simulate forest landscape processes, which are spatially continuous and temporally dynamic processes (e.g., fire disturbance). Forest landscape processes are likely to have greater contribution to forest ecosystem responses than climate variables alone (Gustafson et al., 2010, Kurz et al., 2008, Girardin and Mudelsee, 2008 and Li et al., 2013). Therefore, greater bias could occur if forest landscape processes are not included in the prediction of forest ecosystem dynamics (Reynolds et al., 2001). [...]
[\textbackslash n] Superiority of complex physiological model at the plot scale was achieved at the price of more detailed input data, longer time of simulation, and more simplified representation of forest composition. While the hybrid model was the best model at the regional scale, it cannot provide carbon dynamics at a fine temporal scale (e.g., daily carbon sequestration) like the complex physiological model can. The simple physiological model provided the worst prediction. Although we primarily focused on density of aboveground woody biomass, other traits associated with forest ecosystems (e.g., forest composition, basal area) are often of interest in ecosystem and landscape modeling. Data preparation and simulation time are also often, if not always, of concern.},
keywords = {*imported-from-citeulike-INRMM,~INRMM-MiD:c-13886261,~to-add-doi-URL,bias-toward-primacy-of-theory-over-reality,comparison,complexity,complexity-vs-uncertainty,ecosystem,environmental-modelling,forest-resources,local-over-complication,local-scale,model-comparison,modelling,regional-scale,system-of-systems,uncertainty}
}
Downloads: 0
{"_id":"4YenfrrEgWZ4yENN8","bibbaseid":"jin-he-thompson-aremorecomplexphysiologicalmodelsofforestecosystemsbetterchoicesforplotandregionalpredictions","authorIDs":[],"author_short":["Jin, W.","He, H. S.","Thompson, F. R."],"bibdata":{"bibtype":"article","type":"article","title":"Are More Complex Physiological Models of Forest Ecosystems Better Choices for Plot and Regional Predictions?","author":[{"propositions":[],"lastnames":["Jin"],"firstnames":["Wenchi"],"suffixes":[]},{"propositions":[],"lastnames":["He"],"firstnames":["Hong","S."],"suffixes":[]},{"propositions":[],"lastnames":["Thompson"],"firstnames":["Frank","R."],"suffixes":[]}],"date":"2016-01","journaltitle":"Environmental Modelling & Software","volume":"75","pages":"1–14","issn":"1364-8152","doi":"10.1016/j.envsoft.2015.10.004","url":"https://doi.org/10.1016/j.envsoft.2015.10.004","abstract":"[Highlights] [::] We evaluated performance of process-based forest ecosystem models. [::] A complex physiological model performed best at the plot scale. [::] A hybrid empirical-physiological model performed best at the regional scale. [Abstract] We evaluated performance of process-based forest ecosystem models. A complex physiological model performed best at the plot scale. A hybrid empirical-physiological model performed best at the regional scale. Process-based forest ecosystem models vary from simple physiological, complex physiological, to hybrid empirical-physiological models. Previous studies indicate that complex models provide the best prediction at plot scale with a temporal extent of less than 10 years, however, it is largely untested as to whether complex models outperform the other two types of models at plot and regional scale in longer timeframe (i.e. decades). We compared model predictions of aboveground carbon by one representative model of each model type (PnET-II, ED2 and LINKAGES v2.2, respectively) with field data (19-77 years) at both scales in the Central Hardwood Forests of the United States. At plot scale, predictions by complex physiological model were the most concordant with field data, suggesting that physiological processes are more influential than forest composition and structure. Hybrid model provided the best predictions at regional scale, suggesting that forest composition and structure may be more influential than physiological processes. [Excerpt: Regional scale comparisons] The percent bias of all models was larger at the regional scale than the plot scale. Abiotic environmental heterogeneity at the regional scale could be one of the factors contributing to the larger percent bias at the regional scale. Even though we used ecological subsections as our regional scale study areas-areas where the vegetation and environmental factors are considered relatively homogenous (McNab et al., 2007), the environmental heterogeneity within each subsection would still be higher than that at each plot-scale site. Small-scale abiotic environmental variations, such as difference in water availability at different slope positions of the same soil type, have been ignored in this regional scale study, and summarized environmental factors were used to represent the average physical situations across the entire subsection. [...] [\\n] Predictions based on empirical relationships like those in the hybrid model may not hold true under changing environments in the future, since those relationships were established based on observations in the past (Gustafson, 2013 and Cuddington et al., 2013). However, predictions based on empirical life history attributes might retain validity in the future due to niche conservatism (Crisp et al., 2009 and Wiens et al., 2010). For example, the tolerance range of growing degree days of a given plant species may remain largely constant despite climate change. [\\n] None of the models we examined simulate forest landscape processes, which are spatially continuous and temporally dynamic processes (e.g., fire disturbance). Forest landscape processes are likely to have greater contribution to forest ecosystem responses than climate variables alone (Gustafson et al., 2010, Kurz et al., 2008, Girardin and Mudelsee, 2008 and Li et al., 2013). Therefore, greater bias could occur if forest landscape processes are not included in the prediction of forest ecosystem dynamics (Reynolds et al., 2001). [...] [\\n] Superiority of complex physiological model at the plot scale was achieved at the price of more detailed input data, longer time of simulation, and more simplified representation of forest composition. While the hybrid model was the best model at the regional scale, it cannot provide carbon dynamics at a fine temporal scale (e.g., daily carbon sequestration) like the complex physiological model can. The simple physiological model provided the worst prediction. Although we primarily focused on density of aboveground woody biomass, other traits associated with forest ecosystems (e.g., forest composition, basal area) are often of interest in ecosystem and landscape modeling. Data preparation and simulation time are also often, if not always, of concern.","keywords":"*imported-from-citeulike-INRMM,~INRMM-MiD:c-13886261,~to-add-doi-URL,bias-toward-primacy-of-theory-over-reality,comparison,complexity,complexity-vs-uncertainty,ecosystem,environmental-modelling,forest-resources,local-over-complication,local-scale,model-comparison,modelling,regional-scale,system-of-systems,uncertainty","bibtex":"@article{jinAreMoreComplex2016,\n title = {Are More Complex Physiological Models of Forest Ecosystems Better Choices for Plot and Regional Predictions?},\n author = {Jin, Wenchi and He, Hong S. and Thompson, Frank R.},\n date = {2016-01},\n journaltitle = {Environmental Modelling \\& Software},\n volume = {75},\n pages = {1--14},\n issn = {1364-8152},\n doi = {10.1016/j.envsoft.2015.10.004},\n url = {https://doi.org/10.1016/j.envsoft.2015.10.004},\n abstract = {[Highlights]\n\n[::] We evaluated performance of process-based forest ecosystem models. [::] A complex physiological model performed best at the plot scale. [::] A hybrid empirical-physiological model performed best at the regional scale.\n\n[Abstract]\n\nWe evaluated performance of process-based forest ecosystem models. A complex physiological model performed best at the plot scale. A hybrid empirical-physiological model performed best at the regional scale. Process-based forest ecosystem models vary from simple physiological, complex physiological, to hybrid empirical-physiological models. Previous studies indicate that complex models provide the best prediction at plot scale with a temporal extent of less than 10 years, however, it is largely untested as to whether complex models outperform the other two types of models at plot and regional scale in longer timeframe (i.e. decades). We compared model predictions of aboveground carbon by one representative model of each model type (PnET-II, ED2 and LINKAGES v2.2, respectively) with field data (19-77 years) at both scales in the Central Hardwood Forests of the United States. At plot scale, predictions by complex physiological model were the most concordant with field data, suggesting that physiological processes are more influential than forest composition and structure. Hybrid model provided the best predictions at regional scale, suggesting that forest composition and structure may be more influential than physiological processes.\n\n[Excerpt: Regional scale comparisons]\n\nThe percent bias of all models was larger at the regional scale than the plot scale. Abiotic environmental heterogeneity at the regional scale could be one of the factors contributing to the larger percent bias at the regional scale. Even though we used ecological subsections as our regional scale study areas-areas where the vegetation and environmental factors are considered relatively homogenous (McNab et al., 2007), the environmental heterogeneity within each subsection would still be higher than that at each plot-scale site. Small-scale abiotic environmental variations, such as difference in water availability at different slope positions of the same soil type, have been ignored in this regional scale study, and summarized environmental factors were used to represent the average physical situations across the entire subsection. [...]\n\n[\\textbackslash n] Predictions based on empirical relationships like those in the hybrid model may not hold true under changing environments in the future, since those relationships were established based on observations in the past (Gustafson, 2013 and Cuddington et al., 2013). However, predictions based on empirical life history attributes might retain validity in the future due to niche conservatism (Crisp et al., 2009 and Wiens et al., 2010). For example, the tolerance range of growing degree days of a given plant species may remain largely constant despite climate change.\n\n[\\textbackslash n] None of the models we examined simulate forest landscape processes, which are spatially continuous and temporally dynamic processes (e.g., fire disturbance). Forest landscape processes are likely to have greater contribution to forest ecosystem responses than climate variables alone (Gustafson et al., 2010, Kurz et al., 2008, Girardin and Mudelsee, 2008 and Li et al., 2013). Therefore, greater bias could occur if forest landscape processes are not included in the prediction of forest ecosystem dynamics (Reynolds et al., 2001). [...]\n\n[\\textbackslash n] Superiority of complex physiological model at the plot scale was achieved at the price of more detailed input data, longer time of simulation, and more simplified representation of forest composition. While the hybrid model was the best model at the regional scale, it cannot provide carbon dynamics at a fine temporal scale (e.g., daily carbon sequestration) like the complex physiological model can. The simple physiological model provided the worst prediction. Although we primarily focused on density of aboveground woody biomass, other traits associated with forest ecosystems (e.g., forest composition, basal area) are often of interest in ecosystem and landscape modeling. Data preparation and simulation time are also often, if not always, of concern.},\n keywords = {*imported-from-citeulike-INRMM,~INRMM-MiD:c-13886261,~to-add-doi-URL,bias-toward-primacy-of-theory-over-reality,comparison,complexity,complexity-vs-uncertainty,ecosystem,environmental-modelling,forest-resources,local-over-complication,local-scale,model-comparison,modelling,regional-scale,system-of-systems,uncertainty}\n}\n\n","author_short":["Jin, W.","He, H. S.","Thompson, F. R."],"key":"jinAreMoreComplex2016","id":"jinAreMoreComplex2016","bibbaseid":"jin-he-thompson-aremorecomplexphysiologicalmodelsofforestecosystemsbetterchoicesforplotandregionalpredictions","role":"author","urls":{"Paper":"https://doi.org/10.1016/j.envsoft.2015.10.004"},"keyword":["*imported-from-citeulike-INRMM","~INRMM-MiD:c-13886261","~to-add-doi-URL","bias-toward-primacy-of-theory-over-reality","comparison","complexity","complexity-vs-uncertainty","ecosystem","environmental-modelling","forest-resources","local-over-complication","local-scale","model-comparison","modelling","regional-scale","system-of-systems","uncertainty"],"downloads":0},"bibtype":"article","biburl":"https://tmpfiles.org/dl/58794/INRMM.bib","creationDate":"2020-07-02T22:41:10.491Z","downloads":0,"keywords":["*imported-from-citeulike-inrmm","~inrmm-mid:c-13886261","~to-add-doi-url","bias-toward-primacy-of-theory-over-reality","comparison","complexity","complexity-vs-uncertainty","ecosystem","environmental-modelling","forest-resources","local-over-complication","local-scale","model-comparison","modelling","regional-scale","system-of-systems","uncertainty"],"search_terms":["more","complex","physiological","models","forest","ecosystems","better","choices","plot","regional","predictions","jin","he","thompson"],"title":"Are More Complex Physiological Models of Forest Ecosystems Better Choices for Plot and Regional Predictions?","year":null,"dataSources":["DXuKbcZTirdigFKPF"]}