3 Increasing Trends in Tropical Cyclone Induced Surge Impacts ...
41
After successfully installing the JMA-MRI storm surge model, past and future
TCs are simulated. The model requires seven inputs data which are (a) date: time
(month, day, hour; mmddhh); (b) Pcenter: central pressure (hPa); (c) lon, lat: point of
TC center (in degree); (d) Rref: radius of referenced pressure contour (in degree); (e)
R0: radius of maximum wind (in km); (f) C f : coefficients c1 (=0.70 in usual case);
and (g) P∞: environmental pressure (hPa). The Rref usually set as 1000 hPa or it
can set from R0.
3.4 Model Validation
The performance of the JMA storm surge model was reported in several studies
(Choudhury 2014; Tsuboki et al. 2015; Mio and Tetsuo 2015). Currently, the
Bangladesh Meteorological Department (BMD) has adopted this model into the
operational forecasting systems and reported reasonable results for the demonstration areas (Kohno et al. 2018). One of the salient features of the model is that it
considers astronomical tide during surge simulation. To examine model performance
for the entire NIO, we compare the reported TC-induced maximum surge heights
(m) with modelled values. Reported surge levels were available only for 16 TCs out
of 28 from different sources (Bangladesh Meteorological Department; Dube et al.
2009; Needham et al. 2013; Pakistan Meteorological Department 2010; Roy et al.
1999; Unisys 1999, 2007, 2008).
The results show that the modelled surge levels are in reasonable agreement with
reported surge levels (Fig. 3.1). The correlation coefficient between reported and
model estimated surge level is 0.69, which is highly significant (p < 0.05 at 99%
confidence level). This correlation value clearly indicates that the model performs
well. The model underestimates surge levels for all cases except one TC. Both the
model error and the quality of the reported data cause deviations between modelled
and reported surge levels. Various factors, for instance, radius of maximum wind,
wind speed, storm track, storm’s central pressure, landfall location, coastal elevation
and morphology of the coast largely affect the surge level. Furthermore, we only
estimated maximum surge levels within 48 h of TC landfall that may also contribute
to the lower surge levels.
In conclusion, the JMA-MRI storm surge model can be applied for simulating
surge levels for the NIO as the simulated values are comparable with the reported
values. This conclusion is based on our model validation and on the other validations
reported in other studies.
41
After successfully installing the JMA-MRI storm surge model, past and future
TCs are simulated. The model requires seven inputs data which are (a) date: time
(month, day, hour; mmddhh); (b) Pcenter: central pressure (hPa); (c) lon, lat: point of
TC center (in degree); (d) Rref: radius of referenced pressure contour (in degree); (e)
R0: radius of maximum wind (in km); (f) C f : coefficients c1 (=0.70 in usual case);
and (g) P∞: environmental pressure (hPa). The Rref usually set as 1000 hPa or it
can set from R0.
3.4 Model Validation
The performance of the JMA storm surge model was reported in several studies
(Choudhury 2014; Tsuboki et al. 2015; Mio and Tetsuo 2015). Currently, the
Bangladesh Meteorological Department (BMD) has adopted this model into the
operational forecasting systems and reported reasonable results for the demonstration areas (Kohno et al. 2018). One of the salient features of the model is that it
considers astronomical tide during surge simulation. To examine model performance
for the entire NIO, we compare the reported TC-induced maximum surge heights
(m) with modelled values. Reported surge levels were available only for 16 TCs out
of 28 from different sources (Bangladesh Meteorological Department; Dube et al.
2009; Needham et al. 2013; Pakistan Meteorological Department 2010; Roy et al.
1999; Unisys 1999, 2007, 2008).
The results show that the modelled surge levels are in reasonable agreement with
reported surge levels (Fig. 3.1). The correlation coefficient between reported and
model estimated surge level is 0.69, which is highly significant (p < 0.05 at 99%
confidence level). This correlation value clearly indicates that the model performs
well. The model underestimates surge levels for all cases except one TC. Both the
model error and the quality of the reported data cause deviations between modelled
and reported surge levels. Various factors, for instance, radius of maximum wind,
wind speed, storm track, storm’s central pressure, landfall location, coastal elevation
and morphology of the coast largely affect the surge level. Furthermore, we only
estimated maximum surge levels within 48 h of TC landfall that may also contribute
to the lower surge levels.
In conclusion, the JMA-MRI storm surge model can be applied for simulating
surge levels for the NIO as the simulated values are comparable with the reported
values. This conclusion is based on our model validation and on the other validations
reported in other studies.