High‐resolution monthly precipitation climatologies over Norway (1981–2010): Joining numerical model data sets and in situ observations

Abstract

The 1981-2010 monthly precipitation climatologies for Norway at 1 km resolution are presented. They are computed by an interpolation procedure (HCLIM+RK) combining the output from a numerical model with the in situ observations. Specifically, the regional climate model data set HCLIM-AROME, based on the dynamical downscaling of the global ERA-Interim reanalysis onto 2.5 km resolution, is considered together with 2009 rain-gauges located within the model domain. The precipitation climatologies are defined by superimposing the grid of 1981-2010 monthly normals from the numerical model and the kriging interpolation of station residuals. The combined approach aims at improving the quality of gridded climatologies and at providing reliable precipitation gradients also over those remote Norwegian regions not covered by observations, especially over the northernmost mountainous areas. The integration of rain-gauge data greatly reduces the original HCLIM-AROME biases. The HCLIM+RK errors obtained from the leave-one-out station validation turn out to be lower than those provided by two considered interpolation schemes based on observations only: a multi-linear local regression kriging (MLRK) and a local weighted linear regression (LWLR). As average over all months, the mean absolute (percentage) error is 10.0 mm (11%) for HCLIM+RK, and 11.4 (12%) and 11.6 mm (12%) for MLRK and LWLR, respectively. In addition, by comparing the results at both station and grid cell level, the accuracy of MLRK and LWLR is more sensitive to the spatial variability of station distribution over the domain and their interpolated fields are more affected by discontinuities and outliers, especially over those areas not covered by the rain-gauge network. The obtained HCLIM+RK climatologies clearly depict the main west-to-east gradient occurring from the orographic precipitation regime of the coast to the more continental climate of the inland and it allows to point out the features of the climatic subzones of Norway.

seNorge_2018, daily precipitation, and temperature datasets over Norway

“…(7) is based on the statistics of the innovation (i.e., observation minus background) at station locations. As described by Desroziers et al (2005), the elements of the background error covariance matrix at station locations, which is modeled by us as σ 2 b S, should match the innovation sample covariances. In Tables 1-3, the values of the parameters determining σ 2 b S are shown for a selection of years (1960,1970,1980,1990,2000,2010) in the assumption of a constant D h (i.e., D h i = Dh, ∀i = 1, .…”

“…The innovation distributions are used to investigate the properties of the background error at grid points. On the other hand, the statistics of analysis residuals reveal the filtering properties of the statistical interpolation at station locations that are related to the observation representativeness error (Lussana et al, 2010;Lorenc, 1986;Desroziers et al, 2005). The mean absolute error (MAE) and the root-meansquare error (RMSE) quantify the average mean absolute deviation and the spread, respectively, of a variable.…”

Spatial interpolation of two‐metre temperature over Norway based on the combination of numerical weather prediction ensembles and in situ observations

“…() combining precipitation from regional reanalyses and in situ observations from gauges; and more recently Crespi et al . () demonstrating the benefits of the combination of model data and in situ observations for the reconstruction of monthly precipitation climatologies in Norway. In the context of the European project Uncertainties in Ensemble of Regional Reanalysis (UERRA, uerra.eu), the work presented by Soci et al .…”

“…Examples of OI applications to the combination of numerical model output and independent (i.e. not used in the data assimilation cycle) in situ observations have been described for example by: Mahfouf et al (2007) presenting the Canadian Precipitation Analysis (CaPA); Soci et al (2016) combining precipitation from regional reanalyses and in situ observations from gauges; and more recently Crespi et al (2019) demonstrating the benefits of the combination of model data and in situ observations for the reconstruction of monthly precipitation climatologies in Norway. In the context of the European project Uncertainties in Ensemble of Regional Reanalysis (UERRA, uerra.eu), the work presented by Soci et al (2016) has been applied to two-metre temperature, among the other variables, and the gridded fields have been used for hydrological simulations.…”

seNorge_2018, daily precipitation and temperature datasets over Norway

“…Instead the reference is derived from long-term averages calculated from the output of a high-resolution numerical model. We have used a regional climate simulation with a resolution of 2.5 km, based on the dynamical downscaling of the global reanalysis ERAInterim and available for the time period 2003-2016, to derive the monthly reference fields, as this has been proven skillful for such an application (Crespi et al, 2018).…”