Geoscience Reference
In-Depth Information
rainfall than that of the TRMM observations and the other three experiments, for
both days of the forecast. Furthermore, there are also positional errors in the location
of maximum precipitation as simulated by all the models. Even though the MODIS
run overestimates the precipitation over land, the results of the above run are in
better agreement with observations over the sea. Therefore, the MODIS run is in
better agreement with TRMM for the day-two of the forecast, since it simulates
much less rain over the sea as compared to the CTRL, the SSM/I and the ATOVS
runs.
Equitable Threat Score and Bias Score
Further quantitative analysis of the simulated rainfall is performed by calculating
the statistical skill scores namely “equitable threat score” (ETS) and “bias score”
(BS) using the contingency table ( Wilks 2006 ; Colle et al. 1999 ) . The Bias Score
(BS) is a measure of the ratio of the frequency of forecast events to the frequency of
observed events. The bias score indicates whether the forecast system has a tendency
to underpredict (Bias
) events. The bias score does
not however, measure how well the forecast corresponds to the observations. The
Equitable Threat Score (ETS) measures the fraction of observed and/or forecast
events that are correctly predicted, with a provision for hits associated purely with
“random chance”. The ETS is often used in the verification of rainfall in NWP
models since its “equitability” allows scores to be compared more fairly across
different rainfall thresholds. The ETS penalizes both misses and false alarms in the
same manner and also it does not concern itself with the source of forecast error.
While higher values of the threat score represent enhanced skill of precipitation
forecast, the maximum value of ETS is one. The ETS and BS are calculated for
all the four numerical experiments based on the 48 h accumulated precipitation for
various threshold values. The threshold values used are 40, 50 60, 70, 80, 90 and
100 mm. The results are presented in Figs. 26.8 and 26.9 . From the Fig. 26.8 ,itis
seen that the MODIS experiment exhibits highest skill of precipitation forecast for
all the various threshold values when compared to the CTRL run, while the ATOVS
and the SSM/I experiments, do show some precipitation predictability skill of the
model although lower than the MODIS run. The ATOVS experiment shows higher
skill scores at the lower threshold values which declines with increase of threshold
values. Among the four numerical experiments, the SSM/I experiment shows the
least skill score. However, all the experiments show a decrease in ETS values with
increase in the threshold, indicating the difficulty in the prediction of high intense
rainfall events. More quantitative verification of precipitation forecast is carried out
using the “BS”. Bias score of value of 1.0 implies that the model precipitation
forecast has the same frequency (areal coverage) as that of the observations. The
BS value greater than one for any model run indicates that the above run is over
estimating the precipitation while a BS value less than one signifies the under
estimation of precipitation when compared to the observation. Figure 26.9 gives
<1
) or overpredict (Bias
>1
Search WWH ::




Custom Search