On the likelihood of false alarms and the missing of rainfall events at remote locations in New Zealand by "scaling methods" | The Meteorological Society of New Zealand

On the likelihood of false alarms and the missing of rainfall events at remote locations in New Zealand by "scaling methods"

Year: 
2005
Volume: 
25
Author(s): 
R. Turner
A. Tait
Abstract: 

For remote regions of New Zealand where limited rainfall observations exist most methods for estimating daily rainfall rely to some extent on the scaling/interpolation of "nearby" rainfall observations. Many of these methods do account for orographic enhancement of precipitation, but they all suffer from missed rainfall events (i.e., occasions where heavy rain has fallen at the remote location, but very little or no rainfall will have fallen at any of the "nearby" observing sites. The scaling methods also produce "false alarms", (i.e., where little rain falls at the remote location, but heavy rain has fallen at all or most of the nearby sites). In this note, a contingency table technique is used to quantify the frequency at which "false alarms" and "misses" are likely to occur in various parts of New Zealand. The results are presented for four different climate regions. Results are encouraging in that for three regions, the number of false alarms per year is less than 4 days, however for the most mountainous region the number climbs to 22 days per year. The frequency of missed events was less than 3 days per year for the two drier climates (northern Southland and the eastern Wairarapa) but was
more than 10 days per year for the two wetter regions (Northland and the Lewis Pass area).

Back to top