Limits...
A 3-D variational assimilation scheme in coupled transport-biogeochemical models: Forecast of Mediterranean biogeochemical properties.

Teruzzi A, Dobricic S, Solidoro C, Cossarini G - J Geophys Res Oceans (2014)

Bottom Line: The assimilation solution is found in a reduced dimensional space, and the innovation for the biogeochemical variables is obtained by the sequential application of the covariance operators.The computational costs of the assimilation scheme adopted are low compared to other assimilation techniques, and its modular structure facilitates further developments.The 3D-VAR scheme results especially suitable for implementation within a biogeochemistry operational forecast system.

View Article: PubMed Central - PubMed

Affiliation: Istituto Nazionale di Oceanografia e di Geofisica Sperimentale Trieste, Italy.

ABSTRACT

[1] Increasing attention is dedicated to the implementation of suitable marine forecast systems for the estimate of the state of the ocean. Within the framework of the European MyOcean infrastructure, the pre-existing short-term Mediterranean Sea biogeochemistry operational forecast system has been upgraded by assimilating remotely sensed ocean color data in the coupled transport-biogeochemical model OPATM-BFM using a 3-D variational data assimilation (3D-VAR) procedure. In the present work, the 3D-VAR scheme is used to correct the four phytoplankton functional groups included in the OPATM-BFM in the period July 2007 to September 2008. The 3D-VAR scheme decomposes the error covariance matrix using a sequence of different operators that account separately for vertical covariance, horizontal covariance, and covariance among biogeochemical variables. The assimilation solution is found in a reduced dimensional space, and the innovation for the biogeochemical variables is obtained by the sequential application of the covariance operators. Results show a general improvement in the forecast skill, providing a correction of the basin-scale bias of surface chlorophyll concentration and of the local-scale spatial and temporal dynamics of typical bloom events. Further, analysis of the assimilation skill provides insights into the functioning of the model. The computational costs of the assimilation scheme adopted are low compared to other assimilation techniques, and its modular structure facilitates further developments. The 3D-VAR scheme results especially suitable for implementation within a biogeochemistry operational forecast system.

No MeSH data available.


(a) Assimilation run (shaded areas) and in situ (contours) chlorophyll concentration along the BOUM 2008 trans-Mediterranean section from the NWM to the LEV subregions (x axis is the distance from the first measurement station). Model results have been interpolated into the geographical coordinates of the observation stations (Figure 1). (b) Winter and summer profiles of chlorophyll of AR (blue), CR (green), and climatological data (black) [Manca et al., 2004] for the western (solid lines) and eastern (dotted lines) Mediterranean Sea. The model profiles of the two regions are the averages of the corresponding subregions. The climatological data are the averages of the DS4, DF1, and DF3 regions for the western MS and of the DJ5, DJ7, DJ8, DH3, DL1, DL3, DL4 domains for the eastern MS [Manca et al., 2004, Figure 6].
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4508912&req=5

fig08: (a) Assimilation run (shaded areas) and in situ (contours) chlorophyll concentration along the BOUM 2008 trans-Mediterranean section from the NWM to the LEV subregions (x axis is the distance from the first measurement station). Model results have been interpolated into the geographical coordinates of the observation stations (Figure 1). (b) Winter and summer profiles of chlorophyll of AR (blue), CR (green), and climatological data (black) [Manca et al., 2004] for the western (solid lines) and eastern (dotted lines) Mediterranean Sea. The model profiles of the two regions are the averages of the corresponding subregions. The climatological data are the averages of the DS4, DF1, and DF3 regions for the western MS and of the DJ5, DJ7, DJ8, DH3, DL1, DL3, DL4 domains for the eastern MS [Manca et al., 2004, Figure 6].

Mentions: [56] The skill of the model in reproducing the vertical structure of chlorophyll has been evaluated using available in situ data (BOUM 2008 cruise [Moutin et al., 2012]) and climatological data [Manca et al., 2004]. The comparison with the BOUM data set, which refers to a section from the NWM to the LEV subregions of July 2008, is shown in Figure 8a. The model correctly represents the west-east gradient of the DCM during late spring-summer 2008. The eastward deepening of the DCM is a permanent structure of the MS during spring-summer, which is well reproduced by the OPATM-BFM model [Lazzari et al., 2012], and the use of assimilation very slightly improves the performance of the model (Table 3). However, it is worth noting that the vertical structure of the chlorophyll field, which is strongly driven by physical processes (depth of pycnocline and mixed layer depth), is only partly affected by the assimilation.


A 3-D variational assimilation scheme in coupled transport-biogeochemical models: Forecast of Mediterranean biogeochemical properties.

Teruzzi A, Dobricic S, Solidoro C, Cossarini G - J Geophys Res Oceans (2014)

(a) Assimilation run (shaded areas) and in situ (contours) chlorophyll concentration along the BOUM 2008 trans-Mediterranean section from the NWM to the LEV subregions (x axis is the distance from the first measurement station). Model results have been interpolated into the geographical coordinates of the observation stations (Figure 1). (b) Winter and summer profiles of chlorophyll of AR (blue), CR (green), and climatological data (black) [Manca et al., 2004] for the western (solid lines) and eastern (dotted lines) Mediterranean Sea. The model profiles of the two regions are the averages of the corresponding subregions. The climatological data are the averages of the DS4, DF1, and DF3 regions for the western MS and of the DJ5, DJ7, DJ8, DH3, DL1, DL3, DL4 domains for the eastern MS [Manca et al., 2004, Figure 6].
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4508912&req=5

fig08: (a) Assimilation run (shaded areas) and in situ (contours) chlorophyll concentration along the BOUM 2008 trans-Mediterranean section from the NWM to the LEV subregions (x axis is the distance from the first measurement station). Model results have been interpolated into the geographical coordinates of the observation stations (Figure 1). (b) Winter and summer profiles of chlorophyll of AR (blue), CR (green), and climatological data (black) [Manca et al., 2004] for the western (solid lines) and eastern (dotted lines) Mediterranean Sea. The model profiles of the two regions are the averages of the corresponding subregions. The climatological data are the averages of the DS4, DF1, and DF3 regions for the western MS and of the DJ5, DJ7, DJ8, DH3, DL1, DL3, DL4 domains for the eastern MS [Manca et al., 2004, Figure 6].
Mentions: [56] The skill of the model in reproducing the vertical structure of chlorophyll has been evaluated using available in situ data (BOUM 2008 cruise [Moutin et al., 2012]) and climatological data [Manca et al., 2004]. The comparison with the BOUM data set, which refers to a section from the NWM to the LEV subregions of July 2008, is shown in Figure 8a. The model correctly represents the west-east gradient of the DCM during late spring-summer 2008. The eastward deepening of the DCM is a permanent structure of the MS during spring-summer, which is well reproduced by the OPATM-BFM model [Lazzari et al., 2012], and the use of assimilation very slightly improves the performance of the model (Table 3). However, it is worth noting that the vertical structure of the chlorophyll field, which is strongly driven by physical processes (depth of pycnocline and mixed layer depth), is only partly affected by the assimilation.

Bottom Line: The assimilation solution is found in a reduced dimensional space, and the innovation for the biogeochemical variables is obtained by the sequential application of the covariance operators.The computational costs of the assimilation scheme adopted are low compared to other assimilation techniques, and its modular structure facilitates further developments.The 3D-VAR scheme results especially suitable for implementation within a biogeochemistry operational forecast system.

View Article: PubMed Central - PubMed

Affiliation: Istituto Nazionale di Oceanografia e di Geofisica Sperimentale Trieste, Italy.

ABSTRACT

[1] Increasing attention is dedicated to the implementation of suitable marine forecast systems for the estimate of the state of the ocean. Within the framework of the European MyOcean infrastructure, the pre-existing short-term Mediterranean Sea biogeochemistry operational forecast system has been upgraded by assimilating remotely sensed ocean color data in the coupled transport-biogeochemical model OPATM-BFM using a 3-D variational data assimilation (3D-VAR) procedure. In the present work, the 3D-VAR scheme is used to correct the four phytoplankton functional groups included in the OPATM-BFM in the period July 2007 to September 2008. The 3D-VAR scheme decomposes the error covariance matrix using a sequence of different operators that account separately for vertical covariance, horizontal covariance, and covariance among biogeochemical variables. The assimilation solution is found in a reduced dimensional space, and the innovation for the biogeochemical variables is obtained by the sequential application of the covariance operators. Results show a general improvement in the forecast skill, providing a correction of the basin-scale bias of surface chlorophyll concentration and of the local-scale spatial and temporal dynamics of typical bloom events. Further, analysis of the assimilation skill provides insights into the functioning of the model. The computational costs of the assimilation scheme adopted are low compared to other assimilation techniques, and its modular structure facilitates further developments. The 3D-VAR scheme results especially suitable for implementation within a biogeochemistry operational forecast system.

No MeSH data available.