Search

In the framework of KNMI’s Early Warning Center (EWC), ECMWF ensemble (ENS) predictions are used to issue medium-range forecasts of severe weather. Timely forecasts of wind gusts extremes are important to prevent potential damage. However, ensemble forecasts are affected by biases and under- or over-dispersion. These errors lead to a reduction in the skill of the forecasts, especially for long lead-times and for extreme cases, such as windstorms and deep convective episodes. Hence, statistical post-processing is a fundamental step in the establishment of a skillful weather alert system for extreme wind gust events. However, weather models like ECMWF-IFS are subject to frequent updates, which include changes in the calculation of certain diagnostic variables and by consequence in statistical features of their ensemble distribution. This is the case for ECMWF wind gusts forecasts, whose bias has been reduced with the last update in October 2021. Therefore, the use of pre-update wind gusts forecasts in the training of the post-processing model must be considered with care.In the context of the development of an Ensemble Model Output Statistics (EMOS) model, this limitation has been tackled by reconstructing wind-gusts forecasts with a preliminary EMOS model. This step has been performed by including in the regression those variables that are used by ECMWF for the calculation of wind gusts, which were less affected by the update.The reconstructed wind gusts forecasts have been added to a set of summary statistics of the ensemble distribution of variables physically related to wind gusts. A process of forward selection has been applied to identify the most relevant contributions to the general EMOS model, highlighting reconstructed wind gusts as the most important predictor for all lead-times.The post-processed forecasts obtained with this experimental EMOS model have been verified and compared to those calculated with a conventional EMOS model (performed ignoring the above caveats) and with the results of a non-parametric Quantile Regression Forest. These models have been trained on the same period (2018-2021) and tested on the period that has followed the update (2021-2022), including only grid-points and stations that cover the territory of the Netherlands and distinguishing between summer and winter half-years. The method showing the best performance will be employed operationally for the post-processing of ECMWF-ENS wind gust forecasts over the Netherlands and will be used in the EWC weather alert system.

Day-to-day variations in surface air temperature affect society in many ways, but daily surface air temperature measurements are not available everywhere. Therefore, a global daily picture cannot be achieved with measurements made in situ alone and needs to incorporate estimates from satellite retrievals. This article presents the science developed in the EU Horizon 2020–funded EUSTACE project (2015–19, www.eustaceproject.org) to produce global and European multidecadal ensembles of daily analyses of surface air temperature complementary to those from dynamical reanalyses, integrating different ground-based and satellite-borne data types. Relationships between surface air temperature measurements and satellite-based estimates of surface skin temperature over all surfaces of Earth (land, ocean, ice, and lakes) are quantified. Information contained in the satellite retrievals then helps to estimate air temperature and create global fields in the past, using statistical models of how surface air temperature varies in a connected way from place to place; this needs efficient statistical analysis methods to cope with the considerable data volumes. Daily fields are presented as ensembles to enable propagation of uncertainties through applications. Estimated temperatures and their uncertainties are evaluated against independent measurements and other surface temperature datasets. Achievements in the EUSTACE project have also included fundamental preparatory work useful to others, for example, gathering user requirements, identifying inhomogeneities in daily surface air temperature measurement series from weather stations, carefully quantifying uncertainties in satellite skin and air temperature estimates, exploring the interaction between air temperature and lakes, developing statistical models relevant to non-Gaussian variables, and methods for efficient computation.

We describe a global dataset of quality‐controlled in situ daily air temperature ob-servations covering the period 1850–2015, developed in the framework of the EUSTACE (EU Surface Temperature for All Corners of Earth) project (www.eusta cepro ject.org). The dataset includes a total of 35,364 daily series of maximum and minimum temperature obtained from seven different collections. About 97% of the series are publicly available in a common format, while the remaining 3% can be obtained from the original data providers. Unlike other similar products, duplicates have been removed without blending of series, which simplifies data traceability and improves the temporal homogeneity of the individual series at the cost of a smaller average length. Residual artificial signals (breakpoints) in the series caused by sta-tion relocations, changes in instrumentation, etc., have been detected by means of the combination of four breakpoint detection tests, four variables and three temporal aggregations. The combined results give not only the most probable position of the breakpoints, but also a measure of their likelihood. The reliability of the detection was estimated for each year of each target series, based on the number of reference series and on their correlation with the target series. Moreover, its general perfor-mance was evaluated through a benchmark of synthetic series. This product will be combined with datasets of marine and ice in situ air temperature observations and with measurements from satellite to produce the first complete global statistical re-construction of daily near‐surface air temperature.

Growing season length (GSL) indices derived from surface air temperature are frequently used in climate monitoring applications. The widely used Expert Team on Climate Change Detection and Indices (ETCCDI) definition aims to give a broadly applicable measure of the GSL that is indicative of the duration of the mild part of the year. In this paper long‐term trends in that index are compared with an alternative measure calculated using a time series decomposition technique (empirical ensemble mode decomposition [EEMD]). It is demonstrated that the ETCCDI index departs from the mild‐season definition as its start and end dates are determined by temperature events operating within the synoptic timescale; this raises the inter‐annual variance of the index. The EEMD‐derived index provides a less noisy and more realistic index of the GSL by filtering out the synoptic‐scale variance and capturing the annual‐cycle and longer timescale variability. Long‐term trends in the GSL are comparable between the two indices, with an average increase in length of around 5 days/decade observed for the period 1965–2016. However, the results using the EEMD index display a more coherent picture of significant trends than has been previously observed. Furthermore, the EEMD‐derived growing season parameters are more closely related to variations in seasonal‐mean hemispheric‐scale atmospheric circulation patterns, with around 57% of the inter‐annual variation in the start of the growing season being connected to the North Atlantic Oscillation and East Atlantic patterns, and around 55% of variation in the end of the growing season being associated with East Atlantic/west Russia‐type patterns.

Simulation of past climate is an important tool for the validation of climate models. The comparison with observed daily values allows us to assess the reliability of their projections on climatic extremes in a future climate. The frequency and amplitude of extreme events are fundamental aspects that climate simulations need to reproduce as they have high impacts on economy and society. The ability to simulate them will help policy makers in taking better measures to face climate change. This work aims at evaluating how six models within the High Resolution Model Intercomparison Project reproduce the trends on extreme indices as they have been observed over Europe in the 1970–2014 period. Observed values are provided by the new homogenized version of the E-OBS gridded dataset. The comparison is performed through the use of indices based on seasonal averages and on exceedances of percentile-based thresholds, focusing on six subregions. Winter-average minimum temperature is generally underestimated by models (down to − 4 °C difference over Italy and Norway) while simulated trends in seasonal averages and extreme values are found to be too cold on Eastern Europe and too warm on Iberia and Southern Europe (e.g. up to a difference of − 4% per decade on the number of Cold Nights over Spain). On the other hand the models tend to overestimate summer maximum temperatures averages in the Mediterranean Area (up to + 5 °C over the Balkans) and underestimate these at higher latitudes. Iberia, Southern and Eastern Europe are simulated with too low trends in average summer temperatures. The simulated trends are too strong on the North West part and too weak on the South East part of Europe (down to − 3%/decade on the number of Warm Days over Italy and Western Balkans). These results corroborate the findings of previous studies about the underestimation of the warming trends of summer temperatures in Southern Europe, where these are more intense and have more impacts. The high-resolution versions of the models are compared to their lower-resolution counterparts, similar to those used in the CMIP5, showing a slight improvement for the simulation of extreme winter minimum temperatures, while no significant progresses have been found for extreme summer maximum temperatures.

Long and homogeneous series are a necessary requirement for reliable climate analysis. Relocation of measuring equipment from one station to another, such as from the city center to a rural area or a nearby airport, is one of the causes of discontinuities in these long series that may affect trend estimates. In this paper, an updated procedure for the composition of long series, by combining data from nearby stations, is introduced. It couples an evolution of the blending procedure already implemented within the European Climate Assessment and Dataset (ECA&D, which combines data from stations no more than 12.5 km apart from each other) with a duplicate removal, alongside the quantile matching homogenization procedure. The ECA&D contains approximately 3000 homogenized series for each temperature variable prior to the blending procedure, and approximately 820 of these are longer than 60 years; the process of blending increases the number of long series to more than 900. Three case studies illustrate the effects of the homogenization on single blended series, showing the effectiveness of separate adjustments on extreme and mean values (Geneva, Switzerland), on cases in which blending is complex (Rheinstetten, Germany), and on series that are completed by adding relevant portions of Global Telecommunications System synoptic data (Siauliai, Lithuania). A trend assessment on the whole European continent reveals the removal of negative and very large trends, demonstrating a stronger spatial consistency. The new blended and homogenized dataset will allow a more reliable use of temperature series for indices calculation and for the calculation of gridded datasets and it will be available online for users (https://www.ecad.eu).

The changes of European climate have serious effects on society and economy. A thorough climatological analysis is fundamental to provide reliable and accurate assessments of these changes. Extreme temperature events, such as heatwaves and cold spells, have considerable effects on e.g. health systems, energy consumption and phenological cycles. Their changes in frequency and severity over the last centuries can be studied using daily temperature series from in-situ weather stations. However, these series suffer from external interventions to the measuring stations, such as relocations and modifications to the instruments, and from changes in their surroundings (growing trees, new buildings). These induced changes in the recorded values are not related to climatic events, making the series inhomogeneous and unreliable. With the aim of producing solid temperature databases, several works in the past decades have introduced techniques for the identification of such changes and their correction (homogenization). Within this thesis, a new procedure has been developed, taking inspiration from the Quantile Matching method [Trewin, 2013]. This is based on the calculation of different adjustments for average and extreme values and, in this project, has been revisited and modified, introducing new aspects aimed at making it more flexible, more heuristic and more faithful to the originally observed data. The new homogenization is applied to the European Climate Assessment & Dataset (ECA&D), a pan-European dataset providing observations from all European National Meteorological Services. The method is validated with a comparison to acknowledged homogenization methods against a benchmark dataset, proving its robustness and the quality of the results. The homogenized temperature series, thanks to their high reliability, are then analyzed, performing trend analyses focused on the extreme events. Finally the homogenized series are used to create a homogenized version of E-OBS, gridded dataset obtained with the interpolation of ECA&D station data. The homogenized E-OBS is then employed to compare the trends on average and extreme values of the last decades with those simulated in the same period by climate models, which in other studies are used for predictions of future climate under different emission scenarios.The Introduction (Chapter 1) explains the context in which this thesis is developed. The current state and knowledge about climate change is introduced, followed by the issues implied by the presence of inhomogeneities in temperature series. The aims of the thesis and the expected results and further applications are then exposed in detail.Chapter 2 describes in detail the statistical bases of the homogenization method.Itis composed by a break detection that statistically identifies the timing of the occurred inhomogeneities. This is needed since the dataset is too big to handle the available docu- mentation of reported changes occurred to the stations (metadata), which would require a long and labourious process. The second step is the adjustment calculation, developed following previous studies on the technique of Quantile Matching. This calculates different adjustments for the daily data according to their position in the temperature distribution, thus handling differently average and extreme values. The reported case studies prove the effectiveness of the routine, showing clear improvement in the quality of the series. The difference in the trends of several indices of minimum and maximum temperature between homogenized and raw series show limited changes in average (between +0.01 and +0.02) and no geographical patterns. Moreover, the trends of homogenized series present a clear improvement of the geographical consistency and a considerable narrowing of their distribution. This proves the increased quality of the dataset.The work reported in Chapter 3 describes the process of blending of series. This involves, for example, the series of the station in a city centre that was ended and the new one that was started in a close-by rural area or in an airport. The blending procedure here described joins these series by concatenating them and by mutually filling their gaps. While on one side this process generates long series, on the other hand the blended series are not necessarily homogeneous. For this reason, the homogenization process exposed in Chapter 2 is adapted and applied to these series. The results of this process is a set of long and homogeneous series that are fundamental for thorough historical climatic inspections. Three case studies help exposing the complexity of the process and its benefits. Finally a trend assessment on the new homogenized blended series has been performed. Similarly to what reported by previous studies, this has revealed steep trends in summer maximum temperatures over the Mediterranean and in winter minimum temperatures in Eastern Europe. The latter is connected with a narrowing of the winter minimum temperatures, while in Central Europe a relevant widening of summer maximum temperatures is observed.The Quantile Matching homogenization procedure is compared with other methods in Chapter 4. Here two benchmark datasets are generated, concatenating data from homo- geneous neighbouring series in the national network of Czech Republic and among series specifically selected within the ECA&D. Two benchmark datasets allow to compare situ- ations with very good data quality and station density (Czech dataset) and with scarcer station density and presence of missing data (European dataset). Three well known methods (DAP, HOM, SPLIDHOM) are evaluated together with the Quantile Matching, making use of a set of metrics, such as Root Mean Square Error, percentage of adjusted data and evaluation of trends in average and extreme values. On the Czech Dataset almost all methods perform very well, proving the quality of their statistical features in favourable conditions. The European Dataset allows to test the robustness of the methods in challenging conditions. Here some methods show difficulties in the homogenization of warm extremes and large percentages of missed adjustments of biased data. The Quantile Matching works very well in both cases, showing good performances, comparable to the results of a prestigious method as SPLIDHOM.The homogeneous blended series are the bases for a new version of the gridded dataset E-OBS, which is a valuable tool for the validation of climate simulations, such as the ones developed in the frame of the High Resolution Model Intercomparison Project. These models aim at simulating the climate of the period after 1950 and can be compared to observed values to detect how well they reproduced climate variability and trends. Studies of previous versions of climate simulations highlighted underestimations in the trends of (especially warm) extreme events. In Chapter 5 this comparison is performed taking the difference of the trends in average values and in the number of warm (or cold) extreme events above (or below) percentile-based thresholds. The studied models simulate the trends generally well, though they show underestimation of the strong reduction of cold events in Eastern Europe and of the steep increase of warm events in the Mediterranean area.In the Synthesis (Chapter 6) the obtained results are summarized and discussed, focusing on how they have accomplished the aims of the research. The homogenization method based on the Quantile Matching has shown to work very well on the individual series and on the whole network, reducing the presence of anomalous trends and increasing spatial coherence of the data. The comparison with other methods against a benchmark dataset has validated the quality of the new method and given reliability to the studies performed on the homogenized dataset. These have confirmed the severe warming processes over Europe, highlighting the increased distribution width of summer daily temperatures over Central Europe and the narrowing of the distribution of winter daily temperatures over the Alps and Eastern Europe. Finally, one of the very powerful uses of the results of this thesis has been shown. This is the evaluation of climate simulations against a ho- mogenized gridded dataset, which has allowed to inspect how well the models are able to reproduce the statistical features of the extreme temperature events over the last decades. Moreover, in the Synthesis possible improvements for the homogenization method are ex- posed together with concluding remarks. The main conclusions of this thesis are the acknowledgement of the high efficiency of the developed method, of the high quality of the obtained dataset and of the important added value that homogenization processes like this provide to climatological analyses and to the solidity of the evidences of climate change.

Homogenization of daily temperature series is a fundamental step for climatological analyses. In the last decades, several methods have been developed, presenting different statistical and procedural approaches. In this study, four homogenization methods (together with two variants) have been tested and compared. This has been performed constructing a benchmark dataset, where segments of homogeneous series are replaced with simultaneous measurements from neighboring homogeneous series. This generates inhomogeneous series (the test set) whose homogeneous version (the benchmark set) is known. Two benchmark datasets are created. The first one is based on series from the Czech Republic and has a high quality, high station density, and a large number of reference series. The second one uses stations from all Europe and presents more challenges, such as missing segments, low station density, and scarcity of reference series. The comparison has been performed with pre-defined metrics which check the statistical distance between the homogenized versions and the benchmark. Almost all homogenization methods perform well on the near-ideal benchmark (maximum relative root mean square error (rRMSE): 1.01), while on the European dataset, the homogenization methods diverge and the rRMSE increases up to 1.87. Analyses of the percentages of non-adjusted inhomogeneous data (up to 39%) and substantial differences in the trends among the homogenized versions helped identifying diverging procedural characteristics of the methods. These results add new elements to the debate about homogenization methods for daily values and motivate the use of realistic and challenging datasets in evaluating their robustness and flexibility.

Observations are the foundation for understanding the climate system. Yet, currently available land meteorological data are highly fractured into various global, regional, and national holdings for different variables and time scales, from a variety of sources, and in a mixture of formats. Added to this, many data are still inaccessible for analysis and usage. To meet modern scientific and societal demands as well as emerging needs such as the provision of climate services, it is essential that we improve the management and curation of available land-based meteorological holdings. We need a comprehensive global set of data holdings, of known provenance, that is truly integrated both across essential climate variables (ECVs) and across time scales to meet the broad range of stakeholder needs. These holdings must be easily discoverable, made available in accessible formats, and backed up by multitiered user support. The present paper provides a high-level overview, based upon broad community input, of the steps that are required to bring about this integration. The significant challenge is to find a sustained means to realize this vision. This requires a long-term international program. The database that results will transform our collective ability to provide societally relevant research, analysis, and predictions in many weather- and climate-related application areas across much of the globe.

The daily maximum and minimum temperature series of the European Climate Assessment & Dataset are homogenized using the quantile matching approach. As the dataset is large and the detail of metadata is generally missing, an automated method locates breaks in the series based on a comparison with surrounding series and applies adjustments which are estimated using homogeneous segments of surrounding series as reference. A total of 6,500 series have been processed and after removing duplicates and short series, about 2,100 series have been adjusted. Finally, the effect of the homogenization of daily maximum and minimum temperature on trend estimation is shown to produce a much more spatially homogeneous and then plausible picture.