Boemre 2008 Extended Hindcast Calculation of Gulf of Mexico Circulation: Model Development, Comparisons with Observations



Download 265.82 Kb.
Page3/4
Date18.10.2016
Size265.82 Kb.
#894
1   2   3   4

4. SKILL ASSESSMENTS
The skill of the ocean model hindcast is a critical estimation of the usability of the results for other calculations. Various methods of skill assessments are in the literature. Wang et al. [2003] compares model outputs against an array of current meters over the continental shelf break and slope at the De Soto Canyon. Both means and standard deviations were compared and EOF’s were also computed. Kantha et al. [2005] present a comprehensive comparison of their forecast frontal positions (of the Loop Current and rings) with the observed fronts determined from satellite SST (sea-surface temperature), altimeter data (SSH: sea-surface height) and surface particles; comparisons of ADCP currents with the model are also presented. The authors demonstrate that their relatively simple optimal interpolation (OI) data assimilation scheme using satellite SSH-anomaly can yield good results. Chassignet et al. [2005] use ocean color data from SeaWiFS (Sea-viewing Wide Field-of-view Sensor) to assess the skills of five data-assimilative models. Various assimilation schemes are used including OI, statistical, nudging, and the Cooper and Haines’ [1996] technique. Data used include satellite SSH-anomaly, and MODAS temperature and salinity analyses. As pointed out by the authors, ocean color can characterize the performance of each data assimilation system, but do not allow for a quantitative assessment of the systems. Nonetheless, the authors show that the modeled positions of the Loop Current and rings are in good qualitative agreements with the low-chlorophyll regions representative of the Caribbean water. Oey et al. [2005] also use a relatively simple OI assimilation scheme in which the Mellor and Ezer [1991] surface-subsurface correlation functions are used to project satellite SSH-anomaly into the density field. The forecast skills are assessed by comparing the Loop Current and rings’ frontal positions with those derived from satellite SSH, SST and particles [c.f. Kantha et al. 2005]. The authors show that the inaccurate knowledge of the initial frontal position is the largest source of error, but the model has skills in that after about two weeks the forecast error is less than the persistence (keeping the front at the initial position for all time) error.
Except for the frontal-position comparisons [Kantha et al. 2005; and Oey et al. 2005], the above model-observation comparisons are largely qualitative. In particular, none of these works quantitatively evaluates the modeled currents against in situ observations. Lin et al. [2007] used Oey et al.’s [2005] method and extended it to also assimilate surface particles. An extensive skill assessment was conducted. In addition to standard comparisons against satellite data, the authors quantitatively evaluate the modeled surface currents against particle-derived currents for simulations with and without particle assimilations, and also with and without satellite SSH-anomaly assimilations. Quantitative comparisons were also made for the near-surface currents and temperature profiles against ship-board data (ADCP and temperature measurements). In addition, Lin et al. [2007] also quantitatively compared the various modeled profiles in the upper layer (0 ~ 800 m) against ADCP data at one station over the Sigsbee Escarpment (water depth  2000 m). The first two modeled EOF’s compare well with the observed, and complex vector correlations also indicated good model skill.
In this work, we follow Lin et al. [2007] to assess the skills of various data assimilation schemes described in section 3. A more recent (2003-2004) comprehensive set of observational data obtained by BOEMRE [Donohue et al. 2006] in the east-central Gulf of Mexico (Figure 4.1) is used to evaluate the models. We first compare the upper-layer currents (z > 800 m) with observations; we then present the deep-layer results.
4.1 The Upper-Layer Circulation (z > 800 m):
Figure 4.2 shows the monthly state of the Loop Current and eddy fields during the BOEMRE Exploratory Study [Donohue et al. 2006] from Apr/2003 through Feb/2004. These are model analyses using the Mellor-Ezer scheme. However, the fields are similar for those using the QEnKF scheme and with various resolutions (below). At the start of the observation, the Loop Current extends (figures 4.2a-d). Within 3 months, the Loop Current reaches as far north as the Mississippi Delta where the current reverses from weakly westward prior to the Loop’s arrival (figure 4.2a) to strongly eastward following the anticyclonic current of the Loop as the latter extends northward (figures 4.2b,c,d). A similar reversal of currents occurs over the shelf west of the Delta, i.e. over the Louisiana-Texas (LATEX) shelf, while eastward currents become stronger east of the Delta (on the Mississippi-Alabama shelf). The reversal of the currents over the entire LATEX shelf may be an important episodic example of the Loop Current’s influence on the shelf circulation; the forcing appears to be south of the Mississippi Delta. At the end of this intense Loop Current and slope/shelf interaction, on Aug/2003 (figure 4.2e), an incipient ring-shedding is seen, but no large ring is shed. Instead, the major portion of the ring reattaches to the Loop Current, and two weaker rings are shed (figure 4.2f). The reconstituted Loop Current appears to be unstable, however, and a large ring detaches (after 3~4 months) from the Loop Current as the latter makes a port-to-port path from the Yucatan Channel to the straits of Florida (figure 4.2g,h,i). The detached ring is eventually reabsorbed by the Loop Current as the latter expands, and only a small ring separates and propagates southwestward (figure 4.2j,k,l). At the same time, the two previously-shed (small) rings (i.e. figure 4.2f) merge into one ring (figure 4.2g,h,i) which weakens and drifts southwestward into the Bay of Campeche (at figure 4.2l).

fig01_vertical_sun_oey_da_lce_rings_deep_v01.tif

Figure 4.1. Top: model region from fig.1.1. The red-dashed rectangle indicates BOEMRE’ 2003-2004 observations [Donohue et al. 2006]. Lower panel: mooring locations superimposed on an averaged (2003-2004) SSH color map (m). The L’s are full-depth moorings consisting of C/T/D, ADCP’s and current-meters. Other stations are deep measurements only: 500m and 100m above the bottom. Not shown are 25 PIES stations nearly evenly spaced covering the main portion 88o-92oW, 25.5o-28oN.



To see how the analyzed SSH fields compare with the AVISO data, figure 4.3 shows an example of their correlations for the QEnKFssh25 experiment. Table 4.1 lists the experiments, and also summarizes their time-mean spatial correlations (MCorr) with AVISO. The time-dependent spatial correlation (figure 4.3a) depends on the position and strength of the Loop Current and eddy fields. In general, the correlation is high (>0.8) when the Loop Current is extended into the Gulf of Mexico and rings are clearly visible in AVISO. The correlation drops during periods of incipient shedding (near the beginning of August and end of November 2003) and when the Loop Current retracts (December/2003 through February/2004).
Table 4.1. Descriptions of various experiments and the time-mean (from Mar/01/2003 through May/01/2004) spatial correlations “MCorr” of their SSHA with AVISO SSHA in the region north of 23oN and west of 84oW and in water depths > 500 m in the Gulf of Mexico; here, MCorr = Mean[<.aviso>/(<2><aviso2>)1/2],  = SSHA, <.> = spatial correlation over the stated region, and Mean[.] = time-mean. For the NoAssim experiment, the two values of MCorr correspond to time-means over the first four months and over the entire 17-month period respectively.

Experiments

Descriptions

MCorr

MEssh25

Mellor-Ezer Scheme, SSHA, 25 levels

0.86

MEssh41

Mellor-Ezer Scheme, SSHA, 41 levels

0.85

QEnKFssh25

Quasi-EnKF, SSHA, 25 levels

0.84

QEnKFssh41

Quasi-EnKF, SSHA, 41 levels

0.84

QEnKFsshT25

Quasi-EnKF, SSHA & T, 25 levels

0.84

MEssh25C

Mellor-Ezer Scheme, SSHA, 25 levels, coarse

0.87

NoAssim

“Free-running” experiment, no assimilation

0.58/0.26


fig03_el_67_12_months.tif

Figure 4.2. Model (u,v) at z = 0 m superimposed on color maps of SSH for the 12-month period Apr/2003 through Feb/2004 during the BOEMRE Exploratory Observations [Donohue et al. 2006]. The model analysis uses the Mellor-Ezer scheme. Each panel is a daily-averaged field at the indicated date and the time interval between two consecutive panels is one month. The zero-contour from AVISO altimetry SSH product is shown as the thick black line.

fig04ab_el_ssha_spattimecorr.tif

Figure 4.3. Correlations between modeled and observed (AVISO) SSHA fields for the QEnKFssh25 (see Table 1) analysis: (A) spatial correlation (north of 23oN, west of 84oW and in water depths > 500 m) and (B) time correlation (Mar/01/2003 through May/01/2004).
Table 1 shows that the time-mean spatial correlations (MCorr) are insensitive to the particular data-assimilation schemes used in the analysis. Correlations with satellite data provide a consistency check on the assimilation scheme, but they alone cannot tell how well each scheme performs [Lin et al. 2007]. Thus we do not think that the slightly different MCorr values (Table 4.1) for the various analyses to be significant. Nonetheless, the fact that satellite tends to create in the model results smoother, larger-scale fields is reflected in the high MCorr (= 0.87) value by MEssh25C. For the NoAssim experiment, the MCorr = 0.58 for the first four months, indicating some predictability for approximately the same forecast-period as that we have previously found [e.g. Yin and Oey, 2007].
The space-dependent temporal correlation (figure 4.3b) is generally high (>0.8) over the Loop Current north of the Yucatan Channel, in the broad west-southwest “corridor” region, and also over the western portion of the Gulf of Mexico. These are regions where variability due to energetic Loop Current and rings are large. Moderate correlations (between 0.5 and 0.8) are found along the northern Gulf slope and east of the Yucatan Peninsula. Smaller-scale (O(50-100 km)) eddies exist both in the model and in observations along the northern Gulf slope [Hamilton, 1996]; these eddies tend to reduce the correlation, because satellite analysis such as the AVISO data cannot resolve them. Low correlations (<0.5) exist off the northwestern Florida slope and also in the Campeche Bay in the southwestern Gulf of Mexico. These are regions where small-scale cyclones tend to develop in the model (c.f. Lin et al. 2007). Low correlations are also seen at the Yucatan Channel. Here, the model has rich variability [Ezer et al. 2003]; because of the narrowness of the channel, the AVISO maps are not reliable in the channel.
To assess how the modeled upper-layer currents compare with those observed, we compare EOF’s at each of the six full-depth moorings (i.e. the L-moorings; see figure 1); figure 4.4a compares the eigenvectors and figures 4.4b,c the time series. Only modes 1 and 2 are shown; together they constitute over 90% of the total energy. The agreements between modeled and observed orientations of eigenvectors are generally good (Fig.4.4a), and so are the agreements between the time-series for both modes (Figs.4.4b,c). The agreements are generally better for mode-1 than for mode-2, and also are better for moorings located in the east (L1 and L3) under the direct influences of the Loop Current and new rings, and also in the south (L4). The agreements are poorer at L2 (northwest), L5 (central) and L7 (near the core of the Loop Current). The L2-mooring is located over the slope and farther from the influence of large rings, and is likely affected by fluctuations from smaller features (Hamilton, 1998). The L5-mooring is located over the Sigsbee Escarpment, and the steep topographic slope is likely to be not well resolved by the model. Currents at the L7 mooring in the core of the Loop Current are likely to be sensitive to the fluctuations of the Loop Current.
fig05a_sun_oey_da_lce_rings_deep_v01.tif

Figure 4.4a. Observed (upper panel) and modeled (MEssh25 analysis; lower panel) eigenvectors of the upper-layer (z > 400 m) EOF’s at the six L-moorings shown in Fig.4.1. Red is mode 1 and green is mode 2.

fig05b_mode1_tseries_sun_oey_da_lce_rings_deep_v01.tif

Figure 4.4b. Observed (blue) and modeled (MEssh25 analysis; green) time series of the upper-layer (z > 400 m) EOF mode-1 at the six indicated L-moorings shown in Fig.4.1. The %energy and correlation coefficient between observed and modeled modes are displayed at top of each panel.

fig05c_mode2_tseries_sun_oey_da_lce_rings_deep_v01.tif

Figure 4.4c. Observed and modeled (MEssh25 analysis) time series of the upper-layer (z > 400 m) EOF mode-2 at the six indicated L-moorings shown in Fig.4.1. The %energy and correlation coefficient between observed and modeled modes are displayed at top of each panel.
It is informative to compare the EOF time-series, especially for the mode-1, with the surface (u,v) vectors superimposed on SSH maps (Figure 4.5). For example, the three positive peaks at the beginning of the mode-1 time-series for mooring L1 in April, June and July (Fig.4.4b) are associated with the Loop Current’s northward expansion and eddy-shedding events (Fig.4.5a-e): the mooring is located on the west side of the events and currents are northward and eastward consistent with the east-northeastward direction of the eigenvector (Fig.4.4a). Mode-1 becomes negative3 and fairly steady at later dates after the ring is shed; the current at L1 is then westward caused by the appearance of a cyclone that is formed after the shedding of the ring. A similar interpretation may be made to mooring L3, in which the two negative peaks in June and August correspond to strong westward flow produced by a ring. At later dates, currents at L3 are influenced by northward and eastward flow produced by the Loop Current (Figs.4.5f-g) and a small ring (Figs.4.5h-k). Similar inferences of the EOF’s may be made to other moorings (not shown).
Table 4.2 lists the correlation coefficients between observed and modeled EOF modes 1 and 2 for the three model analyses: MEssh25, QEnKFssh25 and MEssh25C. The results are similar for the three cases. However, this comparison indicates that QEnKFssh25 does not give a more accurate result than the simpler Mellor-Ezer scheme. On the other hand, the higher-resolution model MEssh25 yields slightly more accurate solution than its coarse-grid counterpart, MEssh25C.
fig06_assimssh_gomn116_el_67_12_months_with_moor_loc.tif

Figure 4.5. Model (u,v) at z = 0 m superimposed on color maps of SSH for the 12-month period Apr/2003 through Feb/2004, in the Loop Current and the eastern Gulf of Mexico. Locations of the L-moorings are marked. The model analysis uses the Mellor-Ezer scheme. Each panel is a daily-averaged field at the indicated date and the time interval between two consecutive panels is one month. The zero-contour from AVISO altimetry SSH product is shown as the thick black line.
Table 4.2. Correlation coefficients (CC; significant to 99% confidence value) between observation and model EOF’s modes 1 and 2 at the L-moorings (see fig.4.1 for locations) and the corresponding %energies. The observed %energies are shaded and are displayed under the mooring column. Dashes (“”; CC’s for mooring L4, for the QEnKFssh25 and MEssh25C analyses) indicate insignificant correlations.

Experiments 

 Mooring



MEssh25

Mode #1 Mode#2



QEnKFssh25

Mode#1 Mode#2



MEssh25C

Mode#1 Mode#2



L1: CC

Obs: 57% 39%

0.78 0.36

77% 22%


0.78 0.53

72% 26%


0.72 0.28

79% 19%


L2: CC

Obs: 50% 41%

0.52 0.61

65% 32%


0.38 0.60

64% 31%


0.30 0.61

59% 38%


L3: CC

Obs: 61% 34%

0.73 0.54

64% 35%


0.72 0.46

63% 35%


0.71 0.52

62% 37%


L4: CC

Obs: 59% 34%

0.62 0.64

56% 41%


 

59% 39%


 

55% 43%


L5: CC

Obs: 54% 37%

0.56 0.42

66% 30%


0.36 0.53

58% 37%


0.39 0.49

52% 46%


L7: CC

Obs: 61% 37%

0.58 0.36

69% 31%


0.67 0.49

60% 39%


0.61 0.55

68% 32%



Figure 4.6 compares stick plots at L1, L2, L3 and L4 moorings at the indicated depths for the MEssh25 model analysis. Let C = u + iv, where i = (1)1/2; the magnitude and phase of the complex or vector correlation (e.g. Lin et al. 2007) between the observed and modeled “CC’s” are plotted as a function of depth in Figure 4.7. At L1 (Fig.4.6a), currents near the surface are dominated in the first 3~4 months by the Loop Current expansion and shedding of a ring, followed by a period of weaker flow approximately to the southwest caused by the appearance of a cyclonic gyre north of the (retracted) Loop Current; these were discussed previously in conjunction with the EOF modes (Figs.4.4 and 4.5), and there is a general agreement between the observed and modeled currents. At deeper levels (z  750 m and deeper in Fig.4.6a), the agreements are poor. The |CC| values generally decrease with depth, with some near bottom improvement at some stations.
In summary, we have the following conclusions:


  1. The different data-assimilative schemes (Table 4.1) yield surprisingly similar results. The Mellor-Ezer scheme appears to have been “well-tuned” to the behaviors of Loop Current and rings (Oey et al. 2005; Oey, 2004; Yin and Oey, 2007) that, its accuracy in terms of comparison with hydrographic and current-meter data is comparable to those obtained from more complete schemes;

  2. Nonetheless, EnKF methodologies have been successfully tested, and they may offer better alternatives in producing dynamically more balanced fields, as evidenced by their better ability to reproduce higher EOF modes at some moorings;

  3. In general, the analyses tend to contain larger fraction of the energy at the large scales (than observations), as evidenced by the larger ratios of EOF mode-1 to mode-2 at the L-moorings when compared against observations;

  4. Also in general, the analyses tend to give good results in regions of active Loop Current and eddies, and less so in regions of the Gulf with weaker eddy activities such as the southwestern and northeastern portions of the Gulf of Mexico;

  5. As in Lin et al. (2007), the vertical EOF mode-1 and 2 structures appear to be well simulated in the upper approximately 500 m of the water column. At deep levels, the comparison is poor. Deep motions consist of small-scale eddies and dispersive topographic Rossby waves, and it does not appear that, at present, the model and data are sufficient to produce a good analysis field;

  6. Higher-resolution horizontal grid gives more accurate analyses. On the other hand, increasing the vertical resolution (from 25 to 41 levels) has little impact on the results. The reason is that the original vertical grid of 25 levels already has fine grid sizes near the surface and bottom, and that the upper layer is constrained by data assimilation. Increasing the vertical resolution for deep levels does not help much because deep motions are nearly depth-independent (Hamilton, 2007, 2009; Oey, 2008); and

  7. Not shown above are the results of assimilation with the PIES moorings. These do not appreciably change the accuracy. Satellite SSHA appears to have sufficiently constrained the analyses at the larger scales. At smaller scales at deeper levels that may be detected by PIES, the model cannot simulate them with good accuracy.

In addition to the above, we have also conducted other skill-assessment analyses including detailed energetics at the L-moorings and distributions in frequency and wavenumber spaces, and also Lagragian comparisons of observed and model floats. These will be reported separately in Hamilton and Oey (2010).

fig07a_stick_uv_model_l1.tifFigure 4.6a. Observed (blue) and modeled (green; MEssh25) vector sticks at mooring L1 (see Fig.4.1 for location) at the indicated depths 96m (top 2 panels), 750m (next two panels, etc), 1000m and 1400m below the surface. The “ang” on the top panel is the clockwise angle of the local isobath from true north; it is also the y-direction of the sticks. The |CC| and  on the model panels are the magnitude and phase angle, respectively, of the complex correlation between modeled and observed sticks.

fig07b_stick_uv_model_l2.tif

Figure 4.6b. Observed (blue) and modeled (green; MEssh25) vector sticks at mooring L1 (see Fig.4.1 for location) at the indicated depths 96m (top 2 panels), 750m (next two panels, etc), 1000m and 1650m below the surface. The “ang” on the top panel is the clockwise angle of the local isobath from true north; it is also the y-direction of the sticks. The |CC| and  on the model panels are the magnitude and phase angle, respectively, of the complex correlation between modeled and observed sticks.

fig07c_stick_uv_model_l3.tif

Figure 4.6c. Observed (blue) and modeled (green; MEssh25) vector sticks at mooring L1 (see Fig.4.1 for location) at the indicated depths 96m (top 2 panels), 750m (next two panels, etc), 1000m and 2900m below the surface. The “ang” on the top panel is the clockwise angle of the local isobath from true north; it is also the y-direction of the sticks. The |CC| and  on the model panels are the magnitude and phase angle, respectively, of the complex correlation between modeled and observed sticks.

fig07d_stick_uv_model_l4.tif

Figure 4.6d. Observed (blue) and modeled (green; MEssh25) vector sticks at mooring L1 (see Fig.4.1 for location) at the indicated depths 96m (top 2 panels), 750m (next two panels, etc), 1000m and 3250m below the surface. The “ang” on the top panel is the clockwise angle of the local isobath from true north; it is also the y-direction of the sticks. The |CC| and  on the model panels are the magnitude and phase angle, respectively, of the complex correlation between modeled and observed sticks.

fig08a_vector_corr_mag_nearestpoint_ellipse_depth_interp2.tif

fig08b_vector_corr_mag_nearestpoint_ellipse_depth_interp2.tif

Figure 4.7. The magnitude (upper panel) and phase angle (lower panel), respectively, of the complex correlation between modeled (MEssh25) and observed current profiles at the six L-moorings (see Fig.4.1 for locations).

5. TOPOCAUSTICS
We have seen that data-assimilative analyses in the deep layers do not compare well with observations. Unlike the upper-layer currents, deep motions are not readily constrained by observations. There are fundamental difficulties as well as physical reasons for this. Deep motions are of smaller spatial and temporal scales and the observations and models required must also be correspondingly fine-scale in order to resolve the dynamics. However, even with high-resolution models and observations, it seems unlikely that one can derive a set of data-assimilative analysis that is accurate. Recent studies (Chang and Oey, 2010a,b,c) have indicated active interactions between surface and deep layers, and between the eastern deep Gulf (under the Loop) and western deep Gulf (say west of 90oW). These interactive dynamics must be captured well by the model, as well as by the data that are used to constrained the model. For example, there may not be sufficient data to compute the coupling between the upper and layer transports through the Yucatan Channel. Therefore, a slight mismatch in phasing (for example) between the upper and deep layers can result in large errors. Moreover, deep motions in the Gulf of Mexico are dominated by complex interaction of small-scale eddies with topography that also excite highly dispersive topographic Rossby waves (TRW’s). This section will describe a special but important aspect of this interaction: topocaustics. Details are given in Oey et al. (2009; included as Appendix 2). Here we summarize the essential ideas, with emphasis on the usefulness of the data-assimilative analysis of the upper-layer dynamics (described in section 4) in producing the physics of deep energetic currents.
Topocaustic, or topographic caustic, is due to the focusing of TRW’s caused by sloping topography coupled with bending isobaths, which then results in locally large accumulation of deep-level energy. Mathematically, topocaustics occur near regions of maximum NT = N|Ñh| (N = Brunt-Vaisala frequency; h = water depth). Although the idea draws on a simple mathematical equivalence between the propagation of internal waves and of TRW’s, topocaustics are nevertheless uniquely dependent upon the earth’s rotation, which constraints TRW’s to propagate “one-sided” only, and which therefore allows energy to accumulate at the “western” end of closed contours of NT. The concepts are illustrated in the Fig.5.1 which describes the similarities and contrasts between (non-hydrostatic!) internal waves and (hydrostatic!) topgraphic Rossby waves.
figure02_topocaustics_v09

Figure 5.1. Similarities and differences between (A) internal waves trapped in a thermocline (upper panel; for clarity only one set of rays are sketched) and (B) topographic Rossby waves in a TRW-valley (lower panel).


We suggest that deep (~2000 m) energetic short-period (5~12 days) motions observed by Hamilton and Lugo-Fernandez (2001) and Hamilton (2007) near the Sigsbee escarpment (Fig.5.2) in the Gulf of Mexico are the result of topocaustics. The process is demonstrated in Oey et al. (2009; also in appendix 2) using the analysis product of section 4 for the upper layer. Thus a Gulf of Mexico simulation initialized with a data-assimilated analysis covering the period of the Sigsbee observation is conducted. To correctly simulate the forcing dynamics unconstrained by observations during the period of occurrence of topocautics, the data-assimilative analysis was only applied to initialize the model, approximately 3 months before the high energetic current event. We first show that the assimilation is accurate (Fig.5.3). We then show that the observation mooring is near a localized maximum NT (Fig.5.2). We then carried out the Intrinsic Mode Functions analysis (Huang et al 1999; fig.5.4) and confirm the existence of energetic bursts of short-period deep current events, both in the observations as well as in the model results. We further demonstrated that the strong currents are locally forced from above, either when an extended Loop Current is over the site, or when a warm ring passes (fig.5.5).
figure1abc_topocaustics09

Figure 5.2. (A) AVHRR (Advanced Very High Resolution Radiometry; http://fermi.jhuapl.edu/avhrr/gm/averages/) seven-day composite sea-surface temperature in the Gulf of Mexico in Feb/1993 showing the Loop Current and a warm-core ring further west; (B) contours (dark lines and color shading) of maximum allowable TRW frequency NT = N|Ñh| (cycles/day or cpd) in the vicinity of the Sigsbee escarpment (box in panel A). Here the N = 610-4 s-1. Thin brown contours are isobaths, and the dotted line = 2000 m. The escarpment is identified with the band of high NT oriented northeast to southwest, approximately along the 2000 m isobaths in the north and along the 3000 m isobaths in the south. The “*” is one mooring from Hamilton (2007) and the along-isobath velocity component (daily-averaged, 200 m above the bottom) is plotted in (C) which also shows the corresponding vector sticks. The dashed arrowed lines in the time-series plot in (C) indicate periods when bursts of TRW’s were identified by Hamilton (2007). (Data courtesy of Dr. Peter Hamilton).


fig07_withi1_topocaustics10_972003word_cleaned

Figure 5.3. A comparison of observed and modeled current variance ellipses at six tall (L) moorings in the east-central Gulf of Mexico (left inset) for the period Apr/2003 through Apr/2004, and also at the Sigsbee (I1) mooring for the period Aug/1999 through Aug/2000, i.e. the “*” mooring shown in Fig.1. Vectors at ellipse centers are 1-year mean velocities but these are not representatives of the long-term means because they were dominated by a few strong current events. Note that scales are different above and below z = 700 m (as indicated by the dotted line). The observational data are courtesy of Dr. Peter Hamilton of SAIC.




fig10_topocaustics10_972003word_cleaned

Figure 5.4. Observed (A) and modeled (B) first six Intrinsic Mode Functions (IMF’s) of the south-to-north (v) component velocity at the Sigsbee mooring. Units are m/s (ordinate) and days since Sep/02/1999 (abscissa). The velocity is 200 m above the bottom (water depth  2000 m).




fig08abc_topocaustics10_972003word_cleaned

Figure 5.5. Modeled sea-surface height (color; red = +0.6m, blue = 0.6 m) and velocity (vectors) on (A) t = 50 day showing the warm-core ring Eddy Juggernaut over the mooring location at the Sigsbee escarpment (solid dot). Panels (B) and (C) show the corresponding plots 130 and 260 days later, respectively. The time “t” is Julian day from Sep/02/1999, the same as that used in Figure 5.4.


Chapter 6. Conclusions and Recommendation
The circulation of the Gulf of Mexico is dominated by the Loop Current and the eddies it sheds, and by the passage of intense weather systems in all seasons. The highly variable and intense circulation, together with river discharges, impacts the marine ecosystems of the Gulf. Intensive and extensive field and modeling studies have increased the understanding of the circulation and provide a basis for skill assessing numerical circulation models. In this project, the Princeton Ocean Modeling group at Princeton University conducted a high-resolution model hindcast of the circulation in the Gulf of Mexico from year 2000 to 2007, a period that is an extension of the hindcast period from 1993 to 1999 previously conducted for BOEMRE [Oey, 2004]. The Princeton’s circulation model is three-dimensional and time-dependent, and includes realistic topography, surface fluxes (wind, heat and salt fluxes), ocean temperature and salinity fields, as well as 34 daily river discharges from the northern Gulf of Mexico. Satellite data are assimilated into the model to give an estimate of the actual ocean’s state – the positions and strengths of Loop Current and rings in particular - during the hindcast years 2000 to 2007.
We have delivered to the Bureau of Ocean Energy Management, Regulation and Enforcement (BOEMRE) the model codes, input and forcing files and all surface and subsurface current (and other ocean) data. These information and analysis improve our understanding of the physical oceanography of the Gulf of Mexico, and can be used by BOEMRE for environmental assessments such as in the oil spill trajectory analysis.
In this project, we have examined both surface and deep currents and have subjected the model dataset to rigorous model-data comparison for the surface currents [as in previous work: Oey, 2004], as well as for deep observations. Since there is much less data in the deep layers of the Gulf of Mexico than surface, this project will also deal with some fundamental questions on the characteristics of deep currents, as well as their connection with topography. It is hoped that these more process-oriented studies will yield knowledge that is useful for the scientific community in general, and for BOEMRE’ ongoing and future research in the Gulf of Mexico in particular.
The project has resulted in 2 peer-reviewed publications: Oey et al. [2009] and Berntsen and Oey [2010] and one manuscript to be submitted.
Recommendations for Future Studies
Our assessments of the simulated currents show that the model can provide good skills near the surface – in the upper 500 m in the Gulf of Mexico. While there has been an impressive gain (since the last hindcast study of Oey, 2004) in simulating the current energetics in deeper layers (deeper than approximately 500m) – in that the intensity of the currents are now comparable to those observed – subsurface features basically have shorter temporal and spatial scales and are much more difficult to hindcast. More deep observations will be necessary and future models should incorporate these observations.

REFERENCES
Atlas, R., R. N. Hoffman, S. C. Bloom, J. C. Jusem, and J. Ardizzone, 1996: A multiyear global surface wind velocity data set using SSM/I wind observations. Bull. Amer. Meteor. Soc., 77, 869–882.

Awaji, T., N. Imasato and H. Kunishi (1980): Tidal exchange through a strait: A numerical experiment using a simple model basin. J. Phys. Oceanogr., 10, 1499–1508.

Berntsen, J. and L.-Y. Oey, 2010: Estimation of the internal pressure gradient in σ-coordinate ocean models: comparison of second-, fourth-, and sixth-order schemes. Ocean Dyn. 60, 317-330. DOI 10.1007/s10236-009-0245-y.

Bunge, L., J. Ochoa, A. Badan, J. Candela, and J. Sheinbaum, 2002: Deep flows in the Yucatan Channel and their relation to changes in the Loop Current extension. J. Geophys. Res., 107, C12, 3233, doi:10.1029/2001JC001256.

Candela J., J. Sheinbaum, J. Ochoa, A. Badan and R. Leben, 2002. The potential vorticity flux through the Yucatan Channel and the Loop Current in the Gulf of México, Geophys. Res. Lett. 29, 2059, doi:10.1029/2002GL015587.

Cetina-Heredia P., J. Candela, J. Sheinbaum, J. Ochoa and A. Badan, 2006: Circulation along the Caribbean Coast of Mexico. J. Geophys. Res., 111, C8, C08021, doi:10.1029/2005JC003056, 2006.



Chang,Y.-L. and L.-Y. Oey, 2010a: Why can wind delay the shedding of Loop Current eddies? J. Phys. Oceanogr, 40, 2481-2495.

Chang,Y.-L. and L.-Y. Oey, 2010b: Eddy and wind-forced heat transports in the Gulf of Mexico. JPO, 40, 2728-2742.

Chang,Y.-L. and L.-Y. Oey, 2011: Loop Current Cycle. J. Phys. Oceanogr, 41, 458-471.

Chang, Y.-L., L.-Y. Oey, F.-H. Xu, H.-F. Lu & A. Fujisaki, 2011: 2010 Oil Spill - trajectory projections based on ensemble drifter analyses. Ocean Dynamics, DOI 10.1007/s10236-011-0397-4.

Cochrane, J.D. and F.J. Kelley . 1986 . Low-frequency circulation on the Texas-Louisiana continental shelf. Journal of Geophysical Research. 91(C9) :10,645-10,659.

Craig, P. D., and M. L. Banner, 1994: Modeling wave-enhanced turbulence in the ocean surface layer. J. Phys. Oceanogr., 24, 2546–2559.

Csanady, G.T., 1983: Dispersal by randomly varying currents. J. Fluid Mech., 132, 375-394.

DeHaan, C., and W. Sturges, 2005: Deep Cyclonic Circulation in the Gulf of Mexico. J. Phys. Oceanogr., 35(10), 1801-1812.

DiMarco, S.F., W.D. Nowlin Jr., and R.O. Reid, 2005: A statistical description of the velocity fields from upper ocean particles in the Gulf of Mexico. InCirculation in the Gulf of Mexico: Observations and Models,” W. Sturges, A. Lugo-Fernandez, Eds, Geophysical Monograph Series, Vol.161, 360pp, 2005.

Donohue, K., P. Hamilton, K. Leaman, R. Leben, M. Prater, D. R. Watts, and E. Waddell, 2006: Exploratory study of deepwater currents in the Gulf of Mexico. Volume ii: Technical reportOCS Study MMS 2006-074.

Ezer, T. and Mellor, G.L., 1994: Continuous assimilation of Geosat altimeter data into a three-dimensional primitive equation Gulf Stream model. J. Phys. Oceanogr., 24(4): 832-847.

Ezer, T., L-Y. Oey, H-C. Lee, and W. Sturges, 2003: The variability of currents in the Yucatan Channel: Analysis of results from a numerical ocean model. Journal of Geophysical Research, 108(C1), 3012, 10.1029/2002JC001509.

Fan, S., Oey, L.-Y., and Hamilton, P., 2004. Assimilation of Particle and Satellite Data in a Model of the Northeastern Gulf of Mexico. Continental Shelf Research, 24(9), 1001-1013.

Gutierrez de Velasco, G. and C. D. Winant, 1996. Seasonal patterns of a wind stress curl over the Gulf of Mexico. J. Geophys. Res., 101, 18127 - 18140.

Hamilton, P., 2007: Deep current variability near the Sigsbee escarpment in the Gulf of Mexico. J. Phys. Oceanogr., 37, 708–726.

Hamilton, P., 2009: Topographic Rossby waves in the Gulf of Mexico. Progress Oceanogr., 82,1-31.

Hamilton, P., J.C. Larsen, K.D. Leaman, T.N. Lee, and E. Waddell, 2005: Transports through the Straits of Florida, J. Phys. Oceanogr., 35, 308-322.

Hendry, R.M., 1982: On the structure of the deep Gulf Stream. Journal of Marine Research, 40, 119-142.

Hoffman, R. N.s, 1984: SASS wind ambiguity removal by direct minimization. Part II: Use of smoothness and dynamical constraints. Mon. Wea.Rev., 112, 1829–1852.

Hogg, N.G., R.S. Pickart, R.M. Hendry, and W.J. Smethie Jr., 1986: The Northern Recirculation Gyre of the Gulf Stream. Deep-Sea Research, 33, 1139-1165.

Hogg, N.G., 1992: On the transport of the Gulf Stream between Cape Hatteras and the Grand Banks. Deep-Sea Research, 39, 1231-1246.

Hogg, N.G. and W.E. Johns, 1995: Western boundary currents. U.S. National Report to Internatonal Union of Geodesy and Geophysics 1991-1994, Supplement to Reviews of Geophysics, 33, 1311-1334.

Hurlburt, H. E., and J. D. Thompson, 1980: A numerical study of Loop Current intrusions and eddy shedding, J. Phys. Oceanogr., 10, 1611– 1651.

Kantha, L., J.-K. Choi, K.J.Schaudt, and C.K. Cooper, 2005: A regional data-assimilative model for operational use in the Gulf of Mexico. InCirculation in the Gulf of Mexico: Observations and Models,” W. Sturges, A. Lugo-Fernandez, Eds, Geophysical Monograph Series, Vol.161, 360pp, 2005.
Kourafalou, V.H., L.-Y. Oey, J.D. Wang and T.N. Lee, 1996.  The fate of river discharge on the continental shelf. Part I: modeling the river plume and the inner-shelf coastal current. J. Geophys. Res, 101(C2), 3415-3434.

Lin, X.-H., L.-Y. Oey and D.-P. Wang, 2007: Altimetry and particle data assimilations of Loop Current and eddies. J. Geophys. Res. 112, C05046, doi:10.1029/2006JC 003779, 2007.




Download 265.82 Kb.

Share with your friends:
1   2   3   4




The database is protected by copyright ©ininet.org 2024
send message

    Main page