Spatio-temporal interpolation of sub-daily (6 h) precipitation over Romania for the period 1975-2010

Abstract

A two-step interpolation framework is proposed for the first time for computing sub-daily (6 h) precipitation maps over Romania, over 36 years (1975-2010), using meteorological and ancillary data. In the first step, the monthly climatology maps were constructed. Here, the auxiliary predictors were quantified with the help of multivariate geostatistical model (Regression Kriging). In the second step - the interpolation of the 6-h anomaly maps of precipitation - three methods were compared: Multiquadratic, Ordinary Kriging and 3d Kriging. Owing to the good results in interpolating precipitation anomalies (and to the fewer steps required for producing the maps), the Multiquadratic method was chosen to construct the 6-h precipitation ratios maps for the period 1975-2010. By comparing the new gridded data set to similar sets of data, and by using precipitation data from five independent stations, it was proven that the proposed methodology was suitable for accurately estimating in space and time the 6-h precipitation values. The final outcome of this work is a gridded precipitation data set, at a 6-h time step, available in high spatial resolution (1 km × 1 km) together with the estimated accuracy.