A note of caution for consistency checking and correcting methods of point precipitation records

D. Koutsoyiannis, A note of caution for consistency checking and correcting methods of point precipitation records, International Precipitation Conference (IPC10), Coimbra, Portugal, doi:10.13140/RG.2.2.34667.75044, 2010.



Point precipitation data are routinely subject to consistency checking and adjustment of emerging inconsistencies, a process also known as data homogenization. The double mass curve is the most popular method of this type. While this is a graphical and empirical method with a high degree of subjectivity, there exist more objective and statistically sound versions. However, all versions tacitly rely on the assumption that precipitation is independent in time over long (e.g. annual) time scales. On the other hand, long precipitation time series reveal that they may exhibit long-range dependence, also known as the Hurst-Kolmogorov (HK) behaviour. A simulation study is performed, which shows that under HK behaviour different slopes appearing in the double mass curve are regular and do not necessarily indicate data inconsistency or inhomogeneity. Thus, application of the routine method to correct the data in fact modifies correct measurements, which are rendered inconsistent. Thus, if we hypothesize that the HK behaviour is common in precipitation, application of such methods may enormously distort correct data, based on a vicious circle logic: (a) we assume time independence of the rainfall process; (b) we interpret manifestation of dependence (the HK behaviour in particular) as incorrectness of data; (c) we modify the data so as to remove the influence of dependence; (d) we obtain series much closer to the faulty assumption of independence. The caution derived from the simulation study is that such methods should never be applied blindly. Unless information on local conditions and station archive justify that inconsistencies or errors exist, corrections of data should be avoided.

PDF Full text:

See also: http://dx.doi.org/10.13140/RG.2.2.34667.75044

Other works that reference this work (this list might be obsolete):

1. Lanza, L.G., and L. Stagi, Non-parametric error distribution analysis from the laboratory calibration of various rainfall intensity gauges, Water Science and Technology, 65 (10), 1745-1752, 2012.