Homogenisation and interpolation of subdaily bright sunshine duration time series in Slovenia

Homogenisation and interpolation of bright sunshine duration time series in Slovenia have recently been renewed at Slovenian Environment Agency (ARSO). First, the quality of measured hourly to daily data at meteorological stations for the period 1960–2022 was checked. QC data was then aggregated to monthly sums, obtaining 27 stations’ time series, separately for three subdaily intervals. Monthly data was normalized relative to the maximum possible sunshine duration and transformed by arcus sinus function to improve the homogenisation with the HOMER software tool. Homogenisation was run in several iterations, using metadata and applying additive correction model with seasonally dependent corrections. Resulting homogenised time series were then back-transformed to the measurement units (hours).

Original subdaily values of sunshine duration were adjusted to match the monthly values of homogenised data. The first step of the adjustment procedure was the normalization of the subdaily data (i.e. division with the maximum possible sunshine duration for the day of year). Resulting ratio values from 0 to 1 were transformed by arcus sinus functions and then shifted by such a value that the monthly sum of back-transformed values matched the homogenised monthly sum. Shifted values below –π/2 were set to –π/2, whereas values above π/2 were set to π/2. This way all the original ratio values of 0 (no sunshine) and 1 (maximum possible sunshine) remain unchanged. This adjustment procedure takes into account both the sunshine duration frequency distribution and the influence of expected inhomogenities.

Missing subdaily data was replaced by spatially interpolated values of homogenised daily ratios. The distribution of raw interpolated values was partly adjusted to the distribution of the corresponding homogenised data in order to improve the statistics of extreme values.

The resulting collection of homogenised and interpolated subdaily data of bright sunshine duration was statistically analysed. Time series of the data show a statistically significant positive linear trend on an annual scale throughout the 63-year period. The trend is stronger for morning (around 3.7 % per decade) than for midday (2.0 % per decade) and afternoon values (2.5 % per decade). For autumn the trend is weak and insignificant, whereas for winter morning hours the trend reaches around 8 % per decade in the lowlands of eastern Slovenia.

Homogenised an interpolated time series have been used for the calculation of climate normals for the latest WMO standard reference period (1991–2020) and are planned to be used to improve climate projections for Slovenia.

Keywords: sunshine duration, climate change, homogenisation, spatial interpolation