Table 1.
a.) JAS | Event forecast in System-4 | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Event in ERAI (1981–2010) | >UT | Yes | No | >M | Yes | No | <LT | Yes | No | ||
Yes | 10 | 0 | Yes | 14 | 1 | Yes | 8 | 2 | |||
No | 5 | 15 | No | 9 | 6 | No | 9 | 11 | |||
Event in IMD (1981–2002) | >UT | Yes | No | >M | Yes | No | <LT | Yes | No | ||
Yes | 6 | 1 | Yes | 11 | 0 | Yes | 5 | 2 | |||
No | 4 | 11 | No | 5 | 6 | No | 8 | 7 | |||
Event in TRMM-ERAI (1998–2010) | >UT | Yes | No | >M | Yes | No | <LT | Yes | No | ||
Yes | 3 | 1 | Yes | 5 | 1 | Yes | 3 | 1 | |||
No | 4 | 5 | No | 4 | 3 | No | 5 | 4 | |||
b.) SON | Event forecast in System-4 | ||||||||||
Event in ERAI (1981–2010) | >UT | Yes | No | >M | Yes | No | <LT | Yes | No | ||
Yes | 8 | 2 | Yes | 14 | 1 | Yes | 8 | 2 | |||
No | 5 | 15 | No | 7 | 8 | No | 5 | 15 | |||
Event in IMD (1981–2002) | >UT | Yes | No | >M | Yes | No | <LT | Yes | No | ||
Yes | 6 | 1 | Yes | 10 | 1 | Yes | 5 | 2 | |||
No | 5 | 10 | No | 5 | 6 | No | 3 | 12 | |||
Event in TRMM-ERAI (1998–2010) | >UT | Yes | No | >M | Yes | No | <LT | Yes | No | ||
Yes | 3 | 1 | Yes | 6 | 0 | Yes | 3 | 1 | |||
No | 1 | 8 | No | 3 | 4 | No | 3 | 6 |
The confusion matrix, or contingency table, collates hit, miss, false alarm and correct rejection rates in Figures 10 and 11 for the System-4 seasonal time series against the three meteorological observation/reanalysis datasets used to drive LMM for a.) the target season of July, August and September for the NE India region (forecasts issued in May) and b.) the target season of September, October and November for the NW India region (forecasts issued in July). The different event categories (read horizontally) are abbreviated as “ >UT” for malaria incidence above the upper tercile, “ >M” for malaria incidence above the median and “ <LT” for malaria incidence below the lower tercile.