february 98 : model velocity [cm/sec]

1
180 182 184 186 188 190 192 194 196 198 6 8 10 12 14 16 18 20 22 24 tem perature [deg celsius] calcofi base rec #1 180 182 184 186 188 190 192 194 196 198 6 8 10 12 14 16 18 20 22 24 daysofcruise calcofi base rec #2 coastal coastal coastal deep ocean deep ocean deep ocean tem perature [deg celsius] Reconstruction w ith constantRM S #1 Reconstruction using higherRM S on coastal stations#2 0 0.01 0.02 0.03 0.04 0.04 0.05 0.06 0.07 0.07 0.08 0.09 0.1 0.1 0.11 0.12 0.13 0.13 0.14 0.15 0.16 0.16 0.17 0.18 0.19 0.2 0.2 0.21 0.22 0.23 0.23 -125 -124 -123 -122 -121 -120 -119 -118 -117 -116 29 30 31 32 33 34 35 36 0 0.03 0.04 0.05 0.07 0.08 0.09 0.11 0.12 0.13 0.15 0.16 0.17 0.19 0.2 0.22 0.23 0.24 0.26 0.27 0.28 0.3 0.31 0.32 0.34 0.35 0.36 0.38 0.39 0.4 0.42 -125 -124 -123 -122 -121 -120 -119 -118 -117 -116 29 30 31 32 33 34 35 36 February 98 : Model Velocity [cm/sec] Poleward Coastal Current Modeling CalCOFI observations during El Niño: Fitting physics and biology Emanuele Di Lorenzo SIO, La Jolla, 92093 CA Arthur J. Miller SIO, La Jolla, 92093 CA John Moisan Long Island University Bruce Cornuelle SIO, La Jolla, 92093 CA Douglas J. Neilson SIO, La Jolla, 92093 CA Testing of the ecosystem module in 1D Comparison of model results with climatology demonstrates that the model is capable of resolving several of observed features. The mixed layer model is capable of resolving the seasonally varying SST and mixed-layer depths. The nitracline is well established at about 100-150 m with very low concentrations at the surface .Surface Chl-a is highest in the winter with an established Chl-a maximum at about 80-90 m.. One issue that has yet to be resolved is why does the model create such a thin Chl-a maximum at depth while the data suggest a wider feature. The oxygen profile from the model compares well with the climatologies and shows a gradual decline in oxygen levels with depth. Physical and Biological Models The physical model used is an eddy- resolving primitive equation (PE) generalized sigma-coordinate ocean circulation model (ROMS: Regional Ocean Modeling System), which is a descendent of SCRUM . The 9km model grid is curvilinear and extends about 1200 km along the coast from northern Baja to north of the San Francisco Bay area with an offshore extent normal to the coast of about 700 km as shown in the figure*. In the vertical, 20 layers reach from the free surface to the bottom, with increased resolution in the surface and bottom boundary layers. The model domain and bathymetry are shown in the “Data Sources” section. The northern, southern and western boundaries are open and with across- boundary fluxes treated by using a modified version of the Orlanski radiation scheme and/or with nudging to specified time-dependent temperature and salinity values. For the biology, seven advection diffusion equations have been added representing Phytoplankton, Zooplankton, Nitrate, Ammonia, Chl-a and two Detritus pools. At each time step source/sink terms regulate the interaction between the the biological components. Data Sources Our effort has relied heavily on the fifty year CalCOFI data record. In addition, datasets collected within the last ten years: ADCP, Topex and Drifters, are also been utilized. For our July 1997 test case, we plotted these datasets over our model domain as shown here with the model’s ETOPO5-derived bathymetry. Time Evolution of Reconstructed Temperature Offshore Eddy Rec #1: Each station is weighted equally in terms of rms- error-bounds. The inverse tends to overweight coastal stations where the misfit is greater. Rec #2: By decreasing the weight of the coastal observation misfit (i.e., increasing the rms-error-bounds on the coastal non-linear station) we improve the fit of the offshore eddies time evolution. Each day of cruise is a different location both in space and time as the ship moves along the sample grid. CalCOFI: observation data from the cruise. Base Run: integration of the model from the non corrected initial state. Rec Runs: integration of the mode after correcting the initial state according to the inverse results. Fitting Observation: an example from the JULY 97 CalCOFI El Niño Cruise The goal of this study was to reconstruct a time-dependent synoptical picture of the ocean circulation, not resolved by the current datasets, for the period of July 1997. To achieve this goal we initially constructed a synoptic, time-independent map of the July 97 ocean state from our inherently time- dependent, observational datasets. This map was used to initialize the model. Since the actual CalCOFI data is sampled over a twenty-day period, we allowed the model to integrate forward in time for twenty days (base run). The model was sampled in the same station order as the CalCOFI cruise stations (i.e., as the ship moves along the sample grid) and at the model times corresponding to the actual cruise times. As seen in the left figure this initialization and integration led to a misfit between the model data (blue line) and observations (red line). 50 m surface surface 50 m 200 m 200 m The misfit increases with time since the initial condition did not consider the time dependency of the evolving circulation field. Using an inverse technique similar to the Green’s Function Method we find a correction to our initial state that minimizes the misfit. No further corrections are made during the model integration. The model integration output from this new initial state we call the reconstruction run. It is important to realize that this inverse method assumes that the dynamics can be linearized. Since both the real system and the model are non-linear, we introduced rms-error- bounds, in the inverse method, on each observation in order to capture only the linear portion of the system. Different maps of the rms-error-bound produce different reconstruction as shown in the figure. April 98 : Model Velocity [cm/sec] Southward Coastal Current February 1998 El Nino Cruise The circulation pattern observed during the CalCOFI cruise was characterized by a strong coastal northward current. The low-salinity core of the California Current was located unusually far offshore. Two month later in the April 98 cruise the jet of the California Current was found inshore and the coastal countercurrent was absent and replaced by a southward flow. Here we present a model simulation in this same period. After setting up an initial condition for the month of February 98 we integrated the model forward and forced it using weekly wind stresses from the Leetmaa Ocean Analysis (from CDC/NOAA). Qualitatively the model captured the changes in the current as shown in the figure In April there was a well developed southward current. A more quantitative measure of the model skill will be assessed once we have completed the fit for the February 98 cruise. Figure: The upper panels show a comparison between CalCOFI observation and model output for an arbitrary month of October in the CalCOFI domain only. The horizontal length scales of the mesoscale features are comparable. The bottom panels are a snapshot of the biology for an arbitrary month of July. Figure: Snapshot of a month of October from a climatological run for the entire model grid. The surface salinity field shows a well defined southward flowing current characterized by a low-salinity core. This pattern is qualitatively similar to the observed CCS. A closer look at the model output also shows a poleward undercurrent located on the continental slope and a mean southern California Bight eddy. Model day 0 is 1 st July 1997. Conclusions Climatological integrations show that the physical model was able to capture the statistics (eg. mean flow, flow variance) of the California Current System. The 1D ecosystem model was capable of resolving several observed features. In the 3D simulations realistic features such as upwelling cells on the coastal region were evident. An error variance reduction of 68% was obtained by defining a new rms-error- bound map (higher on the non-linear coastal stations) and a better fit of the offshore eddies was achieved. Preliminary work on the fit of the February 1998 CalCOFI El Nino cruise shows that the model was sensitive to changes in local wind stress and able to qualitatively reproduce the changes in the coastal regimes observed in the data (from Feb. 98 to Apr. 98). Ongoing Work Complete the fit for the February 1998 cruise for both physics and biology. Assess the dynamical and ecosystem balances that hold in the model’s evolving fields, reconstructed by the inverse and assess their consistency with other model studies. Abstract Hydrographic and ADCP surveys of temperature, salinity and velocity from CalCOFI, altimetric measurements of sea level and drifter observations of temperature and velocity during the 1997- 98 El Nino are now being fit with an eddy- resolving ocean model of the Southern California Bight region to obtain dynamically consistent estimates of eddy variability. Skill evaluations are quantified by the model-data mismatch (rms error) during the fitting interval and eventually by forecasting independent data. Preliminary results of fitting July 1997 physical fields are discussed. The physical fields will be used to drive a three-dimensional NPZD-type model to be fit to sub-surface chlorophyll-a (Chl-a), nitrate and bulk zooplankton from CalCOFI and surface Chl-a from SeaWiFS. Preliminary results of testing the ecosystem model in one-dimensional and three-dimensional form are discussed.

Upload: asta

Post on 12-Jan-2016

28 views

Category:

Documents


0 download

DESCRIPTION

February 98 : Model Velocity [cm/sec]. Poleward Coastal Current. Modeling CalCOFI observations during El Niño: Fitting physics and biology. John Moisan Long Island University. Bruce Cornuelle SIO, La Jolla, 92093 CA. Douglas J. Neilson SIO, La Jolla, 92093 CA. Emanuele Di Lorenzo - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: February 98 : Model Velocity [cm/sec]

180 182 184 186 188 190 192 194 196 1986

8

10

12

14

16

18

20

22

24

tem

per

ature

[deg

cel

sius] calcofi

base rec #1

180 182 184 186 188 190 192 194 196 1986

8

10

12

14

16

18

20

22

24

days of cruise

calcofibase rec #2

coastal coastal coastal deep ocean deep ocean deep ocean te

mper

ature

[deg

cel

sius]

Reconstruction with constant RMS #1

Reconstruction using higher RMS on coastal stations #2

0

0.01

0.02

0.03

0.04

0.04

0.05

0.06

0.07

0.07

0.08

0.09

0.1

0.1

0.11

0.12

0.13

0.13

0.14

0.15

0.16

0.16

0.17

0.18

0.19

0.2

0.2

0.21

0.22

0.23

0.23

-125 -124 -123 -122 -121 -120 -119 -118 -117 -11629

30

31

32

33

34

35

36

0

0.03

0.04

0.05

0.07

0.08

0.09

0.11

0.12

0.13

0.15

0.16

0.17

0.19

0.2

0.22

0.23

0.24

0.26

0.27

0.28

0.3

0.31

0.32

0.34

0.35

0.36

0.38

0.39

0.4

0.42

-125 -124 -123 -122 -121 -120 -119 -118 -117 -11629

30

31

32

33

34

35

36

February 98 : Model Velocity [cm/sec]

Poleward Coastal Current

Modeling CalCOFI observations during El Niño: Fitting physics and biology Emanuele Di Lorenzo

SIO, La Jolla, 92093 CAArthur J. Miller

SIO, La Jolla, 92093 CAJohn Moisan

Long Island UniversityBruce Cornuelle

SIO, La Jolla, 92093 CADouglas J. NeilsonSIO, La Jolla, 92093 CA

Testing of the ecosystem module in 1D Comparison of model results with climatology demonstrates that the model is capable of resolving several of observed features. The mixed layer model is capable of resolving the seasonally varying SST and mixed-layer depths. The nitracline is well established at about 100-150 m with very low concentrations at the surface.Surface Chl-a is highest in the winter with an established Chl-a maximum at about 80-90 m..

One issue that has yet to be resolved is why does the model create such a thin Chl-a maximum at depth while the data suggest a wider feature. The oxygen profile from the model compares well with the climatologies and shows a gradual decline in oxygen levels with depth.

Physical and Biological ModelsThe physical model used is an eddy-resolving primitive equation (PE) generalized sigma-coordinate ocean circulation model (ROMS: Regional Ocean Modeling System), which is a descendent of SCRUM . The 9km model grid is curvilinear and extends about 1200 km along the coast from northern Baja to north of the San Francisco Bay area with an offshore extent normal to the coast of about 700 km as shown in the figure*. In the vertical, 20 layers reach from the free surface to the bottom, with increased resolution in the surface and bottom boundary layers.The model domain and bathymetry are shown in the “Data Sources” section.The northern, southern and western boundaries are open and with across-boundary fluxes treated by using a modified version of the Orlanski radiation scheme and/or with nudging to specified time-dependent temperature and salinity values.For the biology, seven advection diffusion equations have been added representing Phytoplankton, Zooplankton, Nitrate, Ammonia, Chl-a and two Detritus pools. At each time step source/sink terms regulate the interaction between the the biological components.Running the coupled model for several years, using climatological, forcing shows that the model is able to capture the statistics of the California Current System.

Data SourcesOur effort has relied heavily on the fifty year CalCOFI data record. In addition, datasets collected within the last ten years: ADCP, Topex and Drifters, are also been utilized. For our July 1997 test case, we plotted these datasets over our model domain as shown here with the model’s ETOPO5-derived bathymetry.

Time Evolution of Reconstructed Temperature

Offshore Eddy

Rec #1: Each station is weighted equally in terms of rms-error-bounds. The inverse tends to overweight coastal stations where the misfit is greater.

Rec #2: By decreasing the weight of the coastal observation misfit (i.e., increasing the rms-error-bounds on the coastal non-linear station) we improve the fit of the offshore eddies time evolution.

Each day of cruise is a different location both in space and time as the ship moves along the sample grid.CalCOFI: observation data from the cruise.Base Run: integration of the model from the non corrected initial state.Rec Runs: integration of the mode after correcting the initial state according to the inverse results.

Fitting Observation: an example from the JULY 97 CalCOFI El Niño CruiseThe goal of this study was to reconstruct a time-dependent synoptical picture of the ocean circulation, not resolved by the current datasets, for the period of July 1997. To achieve this goal we initially constructed a synoptic, time-independent map of the July 97 ocean state from our inherently time-dependent, observational datasets. This map was used to initialize the model. Since the actual CalCOFI data is sampled over a twenty-day period, we allowed the model to integrate forward in time for twenty days (base run). The model was sampled in the same station order as the CalCOFI cruise stations (i.e., as the ship moves along the sample grid) and at the model times corresponding to the actual cruise times. As seen in the left figure this initialization and integration led to a misfit between the model data (blue line) and observations (red line).

50 m

surface

surface

50 m

200 m

200 m

The misfit increases with time since the initial condition did not consider the time dependency of the evolving circulation field. Using an inverse technique similar to the Green’s Function Method we find a correction to our initial state that minimizes the misfit. No further corrections are made during the model integration. The model integration output from this new initial state we call the reconstruction run. It is important to realize that this inverse method assumes that the dynamics can be linearized. Since both the real system and the model are non-linear, we introduced rms-error-bounds, in the inverse method, on each observation in order to capture only the linear portion of the system. Different maps of the rms-error-bound produce different reconstruction as shown in the figure.

April 98 : Model Velocity [cm/sec]

Southward CoastalCurrent

February 1998 El Nino CruiseThe circulation pattern observed during the CalCOFI cruise was characterized by a strong coastal northward current. The low-salinity core of the California Current was located unusually far offshore. Two month later in the April 98 cruise the jet of the California Current was found inshore and the coastal countercurrent was absent and replaced by a southward flow.Here we present a model simulation in this same period. After setting up an initial condition for the month of February 98 we integrated the

model forward and forced it using weekly wind stresses from the Leetmaa Ocean Analysis (from CDC/NOAA). Qualitatively the model captured the changes in the current as shown in the figure In April there was a well developed southward current. A more quantitative measure of the model skill will be assessed once we have completed the fit for the February 98 cruise.

Figure: The upper panels show a comparison between CalCOFI observation and model output for an arbitrary month of October in the CalCOFI domain only.

The horizontal length scales of the mesoscale features are comparable.

The bottom panels are a snapshot of the biology for an arbitrary month of July.

 

Figure: Snapshot of a month of October from a climatological run for the entire model grid. The surface salinity field shows a well defined southward flowing current characterized by a low-salinity core. This pattern is qualitatively similar to the observed CCS. 

A closer look at the model output also shows a poleward undercurrent located on the continental slope and a mean southern California Bight eddy.  

Model day 0 is 1st July 1997.

ConclusionsClimatological integrations show that the physical model was able to capture the statistics (eg. mean flow, flow variance) of the California Current System.  The 1D ecosystem model was capable of resolving several observed features. In the 3D simulations realistic features such as upwelling cells on the coastal region were evident.  An error variance reduction of 68% was obtained by defining a new rms-error-bound map (higher on the non-linear coastal stations) and a better fit of the offshore eddies was achieved. Preliminary work on the fit of the February 1998 CalCOFI El Nino cruise shows that the model was sensitive to changes in local wind stress and able to qualitatively reproduce the changes in the coastal regimes observed in the data (from Feb. 98 to Apr. 98).

Ongoing WorkComplete the fit for the February 1998 cruise for both physics and biology. Assess the dynamical and ecosystem balances that hold in the model’s evolving fields, reconstructed by the inverse and assess their consistency with other model studies. Forecast independent data in subsequent CalCOFI hydrographic and ADCP surveys, TOPEX datasets and SeaWiFS observations.

AbstractHydrographic and ADCP surveys of temperature, salinity and velocity from CalCOFI, altimetric measurements of sea level and drifter observations of temperature and velocity during the 1997-98 El Nino are now being fit with an eddy-resolving ocean model of the Southern California Bight region to obtain dynamically consistent estimates of eddy variability. Skill evaluations are quantified by the model-data mismatch (rms error) during the fitting interval and eventually by forecasting independent data. Preliminary results of fitting July 1997 physical fields are discussed. The physical fields will be used to drive a three-dimensional NPZD-type model to be fit to sub-surface chlorophyll-a (Chl-a), nitrate and bulk zooplankton from CalCOFI and surface Chl-a from SeaWiFS. Preliminary results of testing the ecosystem model in one-dimensional and three-dimensional form are discussed.