Comments on “Breakpoint Representation of Rainfall”

Abstract

In a recent paper, Sansom presents a new representation of subhour rainfall variations as derived from pluviographs from tilting-siphon raingages. This representation is based on breakpoints where the trace changes slope abruptly. He draws on experience from New Zealand, which shows that rainfall intensity normally changes abruptly between approximately steady states. In this comment, some data from Kenya are presented that largely corroborate that observation, but the data also suggest that rainfall intensity sometimes changes in a more gradual way, which is less suitable for a simple breakpoint representation. Problems that may arise if a breakpoint representation is used fog. smooth changes are discussed, and an extension that accommodates both abrupt changes and smooth changes is suggested. For this extension, a digitized point is designated as either being a ?true? breakpoint (where the trace changes slope abruptly) or an ?ordinary? point, which is needed to represent smooth changes in the trace. The resolution of the autographing tilting-siphon raingage and the tipping-bucket raingage are compared. Two rainfall events from Lund, Sweden, are used to discuss the possibility of using the (extended) breakpoint representation for data from tipping-bucket gauges.