J. Estévez

Spatial regression test for ensuring temperature data quality in southern Spain

Quality assurance of meteorological data is crucial for ensuring the reliability of applications and models that use such data as input variables, especially in the field of environmental sciences. Spatial validation of meteorological data is based on the application of quality control procedures using data from neighbouring stations to assess the validity of data from a candidate station (the station of interest). These kinds of tests, which are referred to in the literature as spatial consistency tests, take data from neighbouring stations in order to estimate the corresponding measurement at the candidate station. These estimations can be made by weighting values according to the distance between the stations or to the coefficient of correlation, among other methods. The test applied in this study relies on statistical decision-making and uses a weighting based on the standard error of the estimate. This paper summarizes the results of the application of this test to maximum, minimum and mean temperature data from the Agroclimatic Information Network of Andalusia (southern Spain). This quality control procedure includes a decision based on a factor f, the fraction of potential outliers for each station across the region. Using GIS techniques, the geographic distribution of the errors detected has been also analysed. Finally, the performance of the test was assessed by evaluating its effectiveness in detecting known errors.

Introduction to the special issue on “hydro-meteorological time series analysis and their relation to climate change”

Observed changes in the climate system are unequivocal: atmosphere and ocean warming, sea level rising, amounts of snow and ice diminution and extreme weather events increasing are some examples (IPCC 2014). The impact of these phenomena on eco-hydrological processes is being studied all over the world (Tahir et al. 2011; Willems and Vrac 2011; Ficklin et al. 2016; Wu et al. 2016). Under this context, the study of hydro-meteorological time series is crucial to understand and characterize the behaviour of important variables such as rainfall and its related patterns and consequences (droughts and floods episodes), river flow, temperature, etc. Recent works show that small changes in temperature, precipitation and snow can have a large impact at the basin scale, being these variables the most affected by climate change (Wang et al. 2014; Zarenistanak et al. 2014; Faiz et al. 2017). Considering the need to deeply know the evolution of hydrological series, it is important to note that water scarcity is one of the most significant challenges that society has to face today and in the coming years, being the key resource for socio-economic development and the natural ecosystems sustainability (Machiwal and Jha 2012). Due to unbridled urbanization and industrialization and the global growth of the population, the water demand is progressively growing in different locations around the world (UNESCO 2009; Grafton and Hussey 2011). With the idea of modelling climate behaviour and forecasting more accurately certain meteorological variables, the use of different techniques can be found in the scientific literature. In this sense, many methods can be applied to detect trends and break points, to obtain scale properties and other characterization parameters. From the identification of significant cyclical components of solar irradiance and temperature (Boland 1995) using Fourier transform (spectral analysis) to the detection of long-range correlations in nonstationary hydro-meteorological time series using multifractal approach (Kalamaras et al. 2017; Krzyszczak et al. 2017), numerous techniques have been used to describe in detail the relevant natural processes. They range from the classical deterministic models such as Box–Jenkins approach or ARIMA to the most current ones using Artificial Neural Networks, Wavelet analysis, Support Vector Machine or Genetic Algorithms (Bărbulescu 2016). To guarantee the reliability of the results obtained from time series analysis, it is necessary to work with validated data sets. Thus, different quality control procedures should be previously applied to hydro-meteorological series, identifying incorrect values, gaps or inconsistent records (Estévez et al. 2011; Fiebrich et al. 2010; López-Lineros et al. 2014). Since hydro-meteorological variables exhibit a widely different behaviour in time and space, a detailed analysis of historical data series in different places of the world is needed. It is then a challenge for scientists to be able to understand how climate change is affecting hydro-meteorological datasets or vice versa, how the different behaviour of these variables can impact on the actual and future climate. This special issue aims at contributing to the understanding of such relationship, providing the most recent results in the analysis of time series of temperature, rainfall, drought, river flow, recorded worldwide and investigated with various statistical methods to disclose deep dynamical climate-linked properties and patterns. & Luciano Telesca luciano.telesca@imaa.cnr.it

Obtaining Homogeneous Regions by Determining the Generalized Fractal Dimensions of Validated Daily Rainfall Data Sets

Extreme rainfall data are widely used in several hydrological models and civil engineering design. Despite high temporal resolution rainfall data are not commonly available, daily rainfall data series are easily found. When these available data series are short in length the Regional Frequency Analysis (RFA) is a good tool to enlarge them by joining stations into homogeneous regions. This is by far, the most complicated step in RFA. This work presents a new method to form homogeneous regions of extreme annual daily rainfall data series. Daily rainfall data series from 53 weather stations in the Maule Region (Chile) have been used. Their fractal dimensions spectra have been obtained by applying the box counting method. Each station has been characterized by the fractal dimensions D1 and D2. A cluster analysis has been carried out based on these at-site characteristics and three regions have been obtained. After performing a RFA of extreme daily annual rainfall data series within each region they have shown as homogeneous. Only one of the available stations has not been possible to be included into any homogeneous regions, being the local frequency analysis the only suitable method to be applied at this location.

Local Analysis of the Characteristics and Frequency of Extreme Droughts in Málaga Using the SPI (Standardized Precipitation Index)

Drought is a natural phenomenon worldwide that triggers significant economic, social and environmental impacts. It is characterized by persistent time periods with recorded rainfall data below the mean, and it is one of the major climate-related hazards. The SPI (Standardized Precipitation Index) is a simple and probabilistic meteorological index widely used to identify the duration and severity of droughts. It can be used in risk management and decision analysis, it can be tailored to time periods of interest to the user and it only needs rainfall data. In this work, the duration, intensity and magnitude of extreme droughts for the period 1945–2011 have been characterized in Malaga (Airport) by using monthly rainfall data. The drought events were identified and characterized using the SPI applied to time scales of 1, 3, 6 and 12 consecutive months. The estimation of the magnitude and intensity quantiles of the drought events for a return period from 2 to 200 years were obtained—using the L-moments method—by a frequency analysis of the maximum annual time series