Optimization of the Spatiotemporal Parameters in a Dynamical Marine Ecosystem Model Based on the Adjoint Assimilation

By utilizing spatiotemporal biological parameterizations, the adjoint variational method was applied to a 3D marine ecosystem dynamical model. The results of twin experiments demonstrated that the mean absolute error (MAE) of phytoplankton in the surface layer and the reduced cost function (RCF) cou...

Full description

Bibliographic Details
Published in:Mathematical Problems in Engineering
Main Authors: Li, Xiaoyan, Wang, Chunhui, Fan, Wei, Lv, Xianqing, Lv, XQ
Format: Article in Journal/Newspaper
Language:English
Published: 2013
Subjects:
Engineering
Mathematics
Multidisciplinary
Interdisciplinary Applications
Science & Technology
Technology
Physical Sciences
PHYSICAL-BIOLOGICAL MODEL
NORTH-ATLANTIC
SEAWIFS DATA
OCEAN
SEA
FILTER
BASIN
GULF
North Atlantic
Online Access:http://ir.qdio.ac.cn/handle/337002/16426
https://doi.org/10.1155/2013/373540
Description
Summary:By utilizing spatiotemporal biological parameterizations, the adjoint variational method was applied to a 3D marine ecosystem dynamical model. The results of twin experiments demonstrated that the mean absolute error (MAE) of phytoplankton in the surface layer and the reduced cost function (RCF) could be used to evaluate both the simulation results and parameter estimation. Spatiotemporal variation of key parameters (KPs) was optimized in real experiments. The RCF and MAE in each assimilation period (72 periods per year) decreased obviously. The spatially varying KP (KPS), temporally varying KP (KPT), and constant KP (KPC) were obtained by averaging KPs of spatiotemporal variation. Another type of spatiotemporal KP (KPST) was represented by KPS, KPT, and KPC. The correlation analysis of KPs, either KPS or KPT, accorded with the real ecological mechanism. Running the model with KPS, KPT, KPC, and KPST, respectively, we found that MAE was the minimum when KPs were spatiotemporal variation (KPST), while MAE reached its maximum when KPs were constant (KPC). Using spatiotemporal KPs could improve simulation precision compared with only using spatially varying KPs, temporally varying KPs, or constant KPs (these forms are the results in a previous study). KPST, a representation of spatiotemporal variation, reduces the variable number in calculation. By utilizing spatiotemporal biological parameterizations, the adjoint variational method was applied to a 3D marine ecosystem dynamical model. The results of twin experiments demonstrated that the mean absolute error (MAE) of phytoplankton in the surface layer and the reduced cost function (RCF) could be used to evaluate both the simulation results and parameter estimation. Spatiotemporal variation of key parameters (KPs) was optimized in real experiments. The RCF and MAE in each assimilation period (72 periods per year) decreased obviously. The spatially varying KP (KPS), temporally varying KP (KPT), and constant KP (KPC) were obtained by averaging KPs of spatiotemporal ...

Similar Items