Effects of undetected data quality issues on climatological analyses

Abstract

Abstract. Systematic data quality issues may occur at various stages of the data generation process. They may affect large fractions of observational datasets and remain largely undetected with standard data quality control. This study investigates the effects of such undetected data quality issues on the results of climatological analyses. For this purpose, we quality controlled daily observations of manned weather stations from the Central Andean area with a standard and an enhanced approach. The climate variables analysed are minimum and maximum temperature and precipitation. About 40 % of the observations are inappropriate for the calculation of monthly temperature means and precipitation sums due to data quality issues. These quality problems undetected with the standard quality control approach strongly affect climatological analyses, since they reduce the correlation coefficients of station pairs, deteriorate the performance of data homogenization methods, increase the spread of individual station trends, and significantly bias regional temperature trends. Our findings indicate that undetected data quality issues are included in important and frequently used observational datasets and hence may affect a high number of climatological studies. It is of utmost importance to apply comprehensive and adequate data quality control approaches on manned weather station records in order to avoid biased results and large uncertainties.

A new homogenized daily data set for temperature variability assessment in Italy

“…Careful quality controls of climate data set are recommended by WMO (Klein Tank et al, 2009;WMO, 2011;Hunziker et al, 2017), as erroneous data can affect different aspects of climatic analysis. As an example, low data quality can produce inhomogeneities in time series and reduce the crosscorrelation between pairs of series, with a negative effect on the performance of any homogenization process (Hunziker et al, 2018). Serious data errors can also compromise the execution of homogenization programs (Domonkos and Coll, 2017a).…”

“…In SCIA the quality controls for temperature and precipitation daily records follow the work of Durre et al (2010) for the Global Historical Climatological Network (GHCN). These controls are cited in several climatological studies, like Ashcroft et al (2018) and Hunziker et al (2018). SCIA quality controls include (a) single series tests, which check the selfconsistency of each time series (i.e., basic integrity checks, outlier presence, temporal coherency); (b) multiple series tests, which assess the consistency of each time series with respect to a set of neighbouring stations (spatial controls).…”

“…Long-term observations are fundamental for monitoring ongoing climate changes, as well as their impacts on human and natural systems. The analysis of climate evolution and its variability is particularly critical in countries characterized by geographically complex territories and a variety of climatic conditions (Hunziker et al, 2018). According to the Köppen-Geiger classification system (Köppen, 1936), Italy's climate ranges from warm temperate with dry summer to polar and it is heavily influenced by the presence of two mountain chains (Alps and Apennines), which control wind circulation and precipitation patterns (Marchetti et al, 2017).…”

Homogenization of a combined hourly air temperature dataset over Romania

“…The increasing number of professional and amateur weather stations which provide meteorological information at high spatial and temporal resolution (i.e., sub-daily data), make fully automatic quality control (QC) and homogenization methods an essential tool for identifying and removing spurious measurements, as well as for adjusting the inhomogeneities from the series (Hunziker et al, 2018;Napoly et al, 2018). Numerous air temperature homogenization studies have been dedicated to monthly and daily data, including extremes, and focusing mainly on local and country scales.…”

Random trend errors in climate station data due to inhomogeneities

“…Besides homogenization, quality control is an important issue in data processing, especially for the determination of extremes. However, Hunziker et al (2018) showed that the error of mean temperature trends depends strongly on whether the data are homogenized or not, whereas the kind of applied quality control is less important. As the aim is to derive trend errors of the raw data neither homogenization nor specific quality control is applied.…”