Woods Hole Coastal and Marine Science Center


next up previous contents
Next: Statistical Comparison of Up: Model/Data Comparison Previous: TemperatureSalinity and

Velocity Time Series Comparison at the Mooring Locations

The quality of the comparison between modeled and observed currents varies substantially with space and time. At the Boston Buoy, for example, events are frequently well-represented during the winter (Figure 3.11), but are only occasionally well-represented during the summer (Figure 3.12).

  Plots of Boston Buoy velocity comparison: Jan 28, 1991 -- Mar 25, 1991
Figure 3.11: Boston Buoy velocity comparison: Jan 28, 1991 -- Mar 25, 1991.

  Plots of Boston Buoy velocity comparison: Jun 18, 1990 -- Aug 13, 1990
Figure 3.12: Boston Buoy velocity comparison: Jun 18, 1990 -- Aug 13, 1990.

As explained in Signell et al (1994), this is because the flow regime in winter is largely wind-driven with large spatial scales. Thus small deviations in the structure or timing of the circulation between between the model and reality will not result in large errors at the mooring locations. During the stratified seasons, however, the spatial scales are considerably smaller due to upwelling, intrusions of Gulf of Maine river water, etc. Small deviations in the structure or timing during the stratified seasons will therefore result in larger errors at the mooring locations. For example, if the model representation of an intrusion is off by 5 km, the whole signal could be missed at a particular site like the future outfall location.

The degree of agreement between model and data can be quantified by the complex correlation coefficient, which is high during the winter and considerably lower during the summer (Figures 3.13 and 3.14).

  Plots of Boston Buoy Model/Data velocity statistics at 5 m depth
Figure 3.13: Boston Buoy Model/Data velocity statistics at 5 m depth.

  Plots of Boston Buoy Model/Data velocity statistics at 23 m depth
Figure 3.14: Boston Buoy Model/Data velocity statistics at 23 m depth.

Although the correlations are considerably lower in the summer, the modeled mean and standard deviation of the low-frequency currents are comparable to observed values and do not show any clear bias.

The relatively good correlation between observed and modeled currents at the Boston Buoy is also seen at the other western Massachusetts Bay stations Scituate and Manomet (Figure 3.15).

  Plots of Model/Data velocity correlations at selected regional stations
Figure 3.15: Model/Data velocity correlations at selected regional stations.

The correlations are significantly lower in Cape Cod Bay and eastern Massachusetts Bay.

Another way of assessing the degree to which the model compares with data is to look at the empirical orthogonal functions (EOF's) of the model and data. At mooring locations where at least 60% of the time period May 1990 -- June 1991 contained good data, time domain EOF's were computed from the east/west and north/south components of the sub-tidal velocity time series. The first modes of the observed and modeled velocities explain 50% and 69% of the variance, respectively, and have reassuringly similar structure (Figure 3.16). This indicates that surface currents in western Massachusetts Bay constitute the largest amount of the energy in both the model and the data.

  Plots of Model/Data comparison of the 1st EOF structure of 
the velocity time series
Figure 3.16: Model/Data comparison of the 1st EOF structure of the velocity time series


next up previous contents
Next: Statistical Comparison of Up: Model/Data Comparison Previous: TemperatureSalinity and

Skip Navigation

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo USAGov logo U.S. Department of the Interior | U.S. Geological Survey

URL: woodshole.er.usgs.gov/operations/modeling/mbayopen/node24.html
Page Contact Information: WHSC Webmaster
This page last modified on Monday, 14-Jan-2013 04:50:21 EST