Advanced

Global parameters sensitivity analysis of modeling water, energy and carbon exchange of an arid agricultural ecosystem

Wu, Mousong LU ; Ran, Youhua; Jansson, Per Erik; Chen, Peng; Tan, Xiao and Zhang, Wenxin LU (2019) In Agricultural and Forest Meteorology 271. p.295-306
Abstract


Agricultural ecosystems are important for regulating terrestrial hydrological and carbon cycles. Hydrological and carbon processes in agricultural ecosystem models are complex due to interactions between parameters. It is therefore crucial to identify parameter sensitivity before a process-based model is applied for simulations and predictions of water, energy and carbon fluxes in agricultural ecosystems. In this study, we investigated the sensitivity and equifinality of the CoupModel parameters in modeling an arid agricultural ecosystem in northwestern China. In total, 27 model parameters were analyzed using a global parameters sensitivity analysis approach and a combination... (More)


Agricultural ecosystems are important for regulating terrestrial hydrological and carbon cycles. Hydrological and carbon processes in agricultural ecosystem models are complex due to interactions between parameters. It is therefore crucial to identify parameter sensitivity before a process-based model is applied for simulations and predictions of water, energy and carbon fluxes in agricultural ecosystems. In this study, we investigated the sensitivity and equifinality of the CoupModel parameters in modeling an arid agricultural ecosystem in northwestern China. In total, 27 model parameters were analyzed using a global parameters sensitivity analysis approach and a combination of multiple in situ and remotely sensed data sets. Among the five major model processes, we found that the energy balance process account for much of the importance in the model, followed by soil hydrology, plant growth, soil heat, and soil carbon processes. Meanwhile, parameters from the plant growth process exhibited higher equifinalities than other processes. We found that net ecosystem exchange (NEE) is controlled by soil heat, soil hydrology and energy balance processes, which is mainly due to a high equifinality (0.91) between the parameters g
max
(maximal stomatal conductance) and V
cmax
(maximal carboxylation rate). The equifinalities between different parameters result in a trade-off in model performance metrics (i.e. determination coefficient R
2
and mean error ME) in the water, energy and carbon balance simulations. We revealed that daytime and yearly accumulated eddy fluxes (sensible heat H
s
, latent heat LE and NEE) can constrain the model parameters better. Remotely sensed data were also promising as additional constraints on soil water contents and energy fluxes. This study introduced a systematic global parameter sensitivity analysis approach together with the equifinality identification in an ecosystem model. The approach proposed here is applicable to other studies and the equifinalities detected in this study can be important implications for modelling arid agricultural ecosystems. Additional exploration on remotely sensed data in constraining the model from different aspects are highly recommended in modeling agricultural ecosystems.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Arid agricultural ecosystem, Equifinality, Parameter sensitivity index, Remote sensing, Water-carbon coupling
in
Agricultural and Forest Meteorology
volume
271
pages
12 pages
publisher
Elsevier
external identifiers
  • scopus:85063072630
ISSN
0168-1923
DOI
10.1016/j.agrformet.2019.03.007
language
English
LU publication?
yes
id
cf126fdc-6278-4dcd-ade8-679679e6bba6
date added to LUP
2019-03-27 12:26:06
date last changed
2019-04-23 04:47:21
@article{cf126fdc-6278-4dcd-ade8-679679e6bba6,
  abstract     = {<p><br>
                                                         Agricultural ecosystems are important for regulating terrestrial hydrological and carbon cycles. Hydrological and carbon processes in agricultural ecosystem models are complex due to interactions between parameters. It is therefore crucial to identify parameter sensitivity before a process-based model is applied for simulations and predictions of water, energy and carbon fluxes in agricultural ecosystems. In this study, we investigated the sensitivity and equifinality of the CoupModel parameters in modeling an arid agricultural ecosystem in northwestern China. In total, 27 model parameters were analyzed using a global parameters sensitivity analysis approach and a combination of multiple in situ and remotely sensed data sets. Among the five major model processes, we found that the energy balance process account for much of the importance in the model, followed by soil hydrology, plant growth, soil heat, and soil carbon processes. Meanwhile, parameters from the plant growth process exhibited higher equifinalities than other processes. We found that net ecosystem exchange (NEE) is controlled by soil heat, soil hydrology and energy balance processes, which is mainly due to a high equifinality (0.91) between the parameters g                             <br>
                            <sub>max</sub><br>
                                                          (maximal stomatal conductance) and V                             <br>
                            <sub>cmax</sub><br>
                                                          (maximal carboxylation rate). The equifinalities between different parameters result in a trade-off in model performance metrics (i.e. determination coefficient R                             <br>
                            <sup>2</sup><br>
                                                          and mean error ME) in the water, energy and carbon balance simulations. We revealed that daytime and yearly accumulated eddy fluxes (sensible heat H                             <br>
                            <sub>s</sub><br>
                                                         , latent heat LE and NEE) can constrain the model parameters better. Remotely sensed data were also promising as additional constraints on soil water contents and energy fluxes. This study introduced a systematic global parameter sensitivity analysis approach together with the equifinality identification in an ecosystem model. The approach proposed here is applicable to other studies and the equifinalities detected in this study can be important implications for modelling arid agricultural ecosystems. Additional exploration on remotely sensed data in constraining the model from different aspects are highly recommended in modeling agricultural ecosystems.                         <br>
                        </p>},
  author       = {Wu, Mousong and Ran, Youhua and Jansson, Per Erik and Chen, Peng and Tan, Xiao and Zhang, Wenxin},
  issn         = {0168-1923},
  keyword      = {Arid agricultural ecosystem,Equifinality,Parameter sensitivity index,Remote sensing,Water-carbon coupling},
  language     = {eng},
  pages        = {295--306},
  publisher    = {Elsevier},
  series       = {Agricultural and Forest Meteorology},
  title        = {Global parameters sensitivity analysis of modeling water, energy and carbon exchange of an arid agricultural ecosystem},
  url          = {http://dx.doi.org/10.1016/j.agrformet.2019.03.007},
  volume       = {271},
  year         = {2019},
}