Areal reduction factors from gridded data products

Abstract

Areal reduction factors (ARFs) convert a point estimate of extreme precipitation to an estimate of extreme precipitation over a spatial domain, and are commonly used in flood risk estimation. The fixed-area approach to ARF estimation considers an area of a certain size and constructs the ratio of extremes with the same exceedance probability for areal average precipitation and point precipitation at a representative location. In regions with sparse observation networks, estimates of areal average precipitation are highly uncertain if based on rain gauge data only. We construct and compare regional and seasonal ARF estimates for Norway using different gridded data products, both observation-based products (SURFdat, seNorge) and reanalysis products (NORA3). For data products that cover a sufficiently long time period, the extremes are estimated using the generalised extreme value (GEV) distribution, while the metastatistical extreme value (MEV) formulation is applied to data products with short records. The results indicate that the NORA3 reanalysis, available at a temporal resolution of 1 h and a spatial resolution of 3km, provides a good overall adequacy for the purpose of obtaining robust and reliable ARF estimates.