...
GTSMv3.0 uses the unstructured Delft3D Flexible Mesh software (Kernkamp et al., 2011). The spatially-varying resolution leads to high accuracy at relatively low computational costs. It has an unprecedented high coastal resolution globally (2.5 km, 1.25km in Europe, Figure 2-1). The resolution decreases from the coast to the deep ocean to a maximum of 25km. Grid resolution is refined in areas in the deep ocean with steep topography areas to enable the dissipation of barotropic energy through generation of internal tides. See User Guide (Yan et al., 2019) for more detailed regarding the model. GTSMv3.0 also has high temporal resolution producing output at 10-minute intervals. The 10-minutes time series are physically realistic since two types of forcing are used; that is, tidal and meteorological forcing. The tidal forcing is internally generated based on position of the earth, moon and sun. The meteorological forcing is available at hourly (or coarser) resolution, but is internally interpolated to the model timestep. Because tides vary at high-frequency and can non-linearly interact with storm surge (so the sum of the two is different from the individual components) we use an temporal resolution of 10 minutes. Especially for output stations with a wide and shallow continental shelf (such as the North Sea) lower temproal resolution, i.e. hourly resolution, can be too coarse and may miss the peak water levels.
...
Table 2-5: Model performance of GTSM against the UHSL dataset and the FES2012 model. The metrics used are standard deviation of errors (STDE), relative range, and correlation coefficient (R). Anchor table9 table9
Geographical Area | UHSCL tide gauge stations | FES2012 assimilative tide model | ||||||
No. of stations | STDE | Relative | R | No. of stations | STDE | Relative | R | |
Antarctic | 1 | 0.07 | 101 | 0.98 | 3 | 0.14 | 107 | 0.96 |
Arctic | 3 | 0.12 | 115 | 0.94 | 40 | 0.05 | 125 | 0.85 |
South East Asia | 27 | 0.28 | 113 | 0.90 | 0 | - | - | - |
Indian Ocean | 39 | 0.20 | 114 | 0.94 | 72 | 0.07 | 112 | 0.98 |
North Atlantic | 48 | 0.18 | 106 | 0.86 | 30 | 0.07 | 102 | 0.97 |
North Pacific | 75 | 0.15 | 102 | 0.95 | 65 | 0.07 | 104 | 0.98 |
South Atlantic | 13 | 0.16 | 114 | 0.94 | 43 | 0.05 | 111 | 0.99 |
South Pacific | 45 | 0.14 | 109 | 0.93 | 94 | 0.07 | 111 | 0.97 |
Total | 251 | 0.18 | 108 | 0.92 | 347 | 0.06 | 111 | 0.96 |
The model performance is also assessed in terms of energy budget. In general, the global and regional estimates of M2 energy dissipation through bottom friction and internal wave drag are in good agreement with satellite altimetry derived estimates by Egbert and Ray (2001). Sensitivity tests show that these energies are slightly sensitive to bottom friction coefficient changes within a range of typical values. The dissipation estimated seems quite sensitive to changes of similar order to the internal wave drag coefficient, showing a positive response in terms of STDE to increasing values of the parameter. However, it is concluded from these tests that spatially non-uniform calibration of both dissipation parameters is needed to optimize the model solution and the agreement with the observed regional dissipation estimates.
...