Fig. 4.

Relationship between the number of years of station data and the number of stations included in the analysis for the a) 1951–80 and b) 1981–2010 normal periods. The solid and dashed lines were generated using only station-years with number of missing days ≤ 10 and ≤ 20 respectively. Cut-offs of 10 missing days per year and 10 years per normal period were used to define the station network employed in the current study (final station numbers shown by black dots).