Chairpersons: Mojca Dolinar, Christoph Frei
Most of the current climate grid datasets are bounded at a time resolution of one day. Particularly in the case of surface air temperature, with its pronounced sub-daily variation, this bound restricts the utility of datasets for environmental modelling. Processes with a non-linear temperature dependence, such as the melting of snow or the transpiration of plants, cannot be accurately represented with daily mean values only. Therefore, advancing into the sub-daily time resolution is highly desirable. But it is also challenging: In order to predict realistic temporal evolutions for unobserved locations, one needs to integrate observations from a whole period, not just one instant. This poster presents early results from a new project that aims at harnessing methods of spatiotemporal statistical modelling for the development of an hourly temperature grid dataset over Switzerland.
We propose and experiment with a dynamic spatiotemporal statistical model that is conceived in the framework of dynamical linear models. Conceptually, it is an extension of kriging with external drift, familiar in spatial climatology, but with time-varying and serially correlated trend coefficients. The coefficients represent characteristic temporal variations in the temperature field, such as a diurnal cycle with a gradually varying amplitude/phase, or the lifting/sinking of a temperature inversion, in response to observations. We apply the method to hourly temperature data from the operational station network in Switzerland and investigate its performance for selected weather episodes (incl. sunny summer days, a winter-time inversion period). Our results hint to more/less successful configurations of the model for hourly temperature gridding in complex terrain. Comparison against the sequential application of a purely spatial analysis of the data shows some added value of the spatiotemporal approach, i.e. a benefit of borrowing information both over space and time. We anticipate that this benefit will increase as we enhance the complexity of the model configuration later in the project.
How to cite: Frey, L. and Frei, C.: Spatial analysis of hourly surface air temperature - Entering the gate to spatiotemporal statistical modelling, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-587, https://doi.org/10.5194/ems2022-587, 2022.
Air humidity can be characterized by several variables (e.g. vapor pressure, dew point, relative humidity). Grid datasets of near-surface humidity are usually available for one of these only, say the daily mean relative humidity. If a different variable is needed in an application, say the vapor pressure deficit, the user is expected to derive it using familiar conversion formuli and a temperature dataset. A largely unnoticed problem of this procedure is that the conversion between variables is valid strictly for instantaneous values, but not for daily means considered in a daily grid dataset, because of non-linearity. In this study, the errors made with such conversion are examined using 10-minute observations at Swiss weather stations, applying the "inappropriate" conversions and comparing the results to the true daily means. The study is a preparatory step in the development of a gridded humidity dataset for Switzerland and its results point to more/less favorable decisions on the target variable.
The results show that the conversion introduces an error with a systematic component (bias) that is quantitatively significant. For example, calculating a daily mean vapor pressure deficit from daily mean relative humidity and temperature introduces a bias of about 10%. The bias is particularly large in summer and at the floor of major valleys, because of the large diurnal cycle of humidity and temperature. Our study explains the origin of the bias in terms of detail in the conversion formuli. An interesting result of our analysis is that conversions starting from dew-point temperature are less prone to error. It appears that dew-point temperature difference constitutes a relatively better choice of target variable for humidity grid datasets, at least with regard to precision of variable conversion.
How to cite: Frei, C.: Spatial analysis of near-surface air humidity. Which variable?, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-58, https://doi.org/10.5194/ems2022-58, 2022.
The amount of data observing precipitation near the Earth’s surface today is enormous and is constantly growing. Among the most interesting data sources, which are showing a greater development are the observational networks made up of stations managed by citizens. From the point of view of national meteorological services, this type of data constitutes an opportunity to integrate the networks of traditional weather stations managed by public institutions. For precipitation, especially, the availability of a dense network delivering data at hourly sampling rates, or less, allows for the reconstruction of weather phenomena occurring between the microscale and the mesoscale. Usually, a network of taditional weather stations allow for the reconstruction of atmospheric fields of a variable near the surface with a minimum spatial effective resolution of about 20 km, the use of crowdsourced observations can potentially move this lower limit down to about 2 km.
In this study, we consider the observations of hourly precipitation collected by a network of European citizens from September 2019 to August 2020. All the observations have been measured by Netatmo weather stations. The stations can be purchased in shops and are then installed in private homes for the most varied purposes, including home automation for instance.
We present an exploratory analysis of the data collected aiming at setting up a quality control system for hourly precipitation. In particular, we will show statistics describing the dataset in terms of: data availability over Europe; rain/no-rain probability; (local) variability of precipitation as a function of its intensity; spatial statistics characterizing the length scales of precipitation over time.
How to cite: Baietti, E. and Lussana, C.: Exploratory analysis of hourly precipitation data measured by citizen observations over Europe, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-240, https://doi.org/10.5194/ems2022-240, 2022.
Precipitation data predicted by numerical models is used in a wide variety of applications. For civil protection, numerical weather prediction provides key information for decision making and as input to hydrological modelling. For climatology, climate bulletins based on reanalyses reconstructing past weather events are increasingly gaining ground. Nevertheless, it is often the case that observed precipitation data from several data sources are available over the domain covered by the numerical simulations. Sometimes the observational data have been used by the numerical models, however usually measurements from observational networks of weather stations or radar-derived precipitation estimates constitute an independent source of information for precipitation. The combination of numerical model output and observational data aiming at a more accurate and precise representation of precipitation has already become a classic of post-processing of numerical models.
The work we will present is based on the application of Inverse problem theory to the spatial analysis of precipitation. Numerical model precipitation is the background field. The observations may have been measured or estimated by more than one observational system, such as weather stations or remote sensing, and they are considered to be more reliable than numerical model precipitation. At the same time, observations are used to add locally more details to the precipitation field, thus increasing its effective spatial resolution. The final product, the combined field of precipitation, has to provide information on the precipitation uncertainty because this is needed for most applications.
Examples of the statistical approaches investigated at the Norwegian meteorological institute for hourly precipitation will be provided. The numerical model is a high-resolution ensemble local area model. The observations are radar-derived estimates from the Norwegian mosaic of weather radars and the rain-gauge measurements used have been collected by an heterogeneous network of traditional and private stations.
How to cite: Lussana, C., Nipen, T. N., and Seierstad, I. A.: A statistical approach to the spatial analysis of precipitation by combining multiple data sources, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-246, https://doi.org/10.5194/ems2022-246, 2022.
Challenge. Meteorological observations are fundamental to sustain a wide range of applications at national meteorological and hydrological services (NMHSs), including gridded climate datasets. Typically, NMHSs use high-quality observations from the networks they operate. However, in recent years, alternative weather data sources are becoming available. Government agencies, organizations, or citizens are joining the effort of monitoring the weather by placing sensors in public or private spaces. The promise of such alternative networks is that they are available at higher spatial resolution that the official ones, which implies the weather observations could contribute to add more spatial detail to the Climate Services. Although the alternative measurements are indeed provided at a higher spatial resolution, they also contain substantial bias and noise, which needs to be addressed.
Approach. The quantification of bias and noise, or measurement errors, is becoming a central theme in science. Measurement error can have a large effect on reliable spatial regression of multifidelity data, therefore it is important to provide a quantified bias and noise level to the regression algorithm. This proper treatment of bias and noise in multi-fidelity data is required, for example, when delivering gridded datasets. Such error treatment does not only improve the reliability of the predicted mean grid (ensemble mean), but also has a notable effect on the predicted uncertainty in the grid (ensemble spread). Since in most cases no quantified bias or noise level is available from the datasets, we propose to use a proxy-based model to learn the bias and noise from the data. Proxies for bias and noise of multi-fidelity meteorological data can be ‘covariates’ like population density, forest cover, radiation intensity, or similar.
Results. In the present work, we use multi-fidelity kriging to combine three datasets in the Netherlands: the network of the National Meteorological Service KNMI, data measured along the national road network by Rijkswaterstaat (Directorate-General for Public Works and Water Management), and crowd-sourced data from the WOW-NL network (http://wow.knmi.nl). In our study, we test our methods on synthetic data and then apply those methods to observed data. We investigate and quantify the improvements in gridded temperature (mean and uncertainty of grid) which we observe when applying a proxy-based error model.
How to cite: de Baar, J., Garcia-Marti, I., and van der Schrier, G.: Spatial regression of multi-fidelity meteorological observations using a proxy-based measurement error model, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-53, https://doi.org/10.5194/ems2022-53, 2022.
The provision of authoritative information about the past, present, and future climate in Europe is one of the principal objectives of the Copernicus Climate Change Service (C3S). In the past decades, the KNMI has been actively contributing to this endeavor through the ECA&D and E-OBS initiatives, which, among responsibilities, periodically release via C3S European-wide gridded observational datasets (E-OBS) of the principal weather variables (e.g. temperature, precipitation) that are fundamental for climate research. Recently, the KNMI has released E-OBS version 25.0e, which includes the first release of gridded daily mean wind speed that covers most of Europe.
The E-OBS wind speed data set covers the period 1980 – 2021. This newly developed gridded dataset applies a combination of spatial regression methods with local uncertainty estimates using a 20-member ensemble. When creating this gridded data set, the daily mean wind speed is modeled using an array of auxiliary variables (e.g. topography, distance to coast, surface roughness) and the ensemble ensures that a measure of uncertainty is provided along with the mean wind speed. In this presentation, we focus on three main parts: 1) general characteristics of the wind dataset; 2) improvements with respect to the previous version (e.g. spatial coverage); and 3) how the use of auxiliary variables characterization of the surroundings of the monitoring station with environmental variables improves the spatial resolution of the grid.
Nevertheless, a pertinent question that follows the creation of the E-OBS daily mean wind dataset is: how does it compare to other well-consolidated data products stemming from numerical weather prediction? In this work, we carry out different statistical comparisons between the daily wind speed provided by E-OBS and ERA5-Land to gain insights about the observed similarities and differences. One of the aspects we consider are possible climatological trends in wind speed over Europe. For example, in the E-OBS data set we see a reduction in yearly mean wind speed around 1990. Is this effect significant, and do we also observe it in ERA5-Land? The comparison of E-OBS and ERA5-Land data for wind speed can provide useful information about their strengths and weaknesses, suggesting directions for future model improvement and improvement of the analysis of observations. Furthermore, this information may be valuable for the proper interpretation of both ERA5-Land and E-OBS data by various users.
How to cite: de Baar, J., Nhat Luu, L., van der Schrier, G., van den Besselaar, E., and Garcia-Marti, I.: Recent improvements in the E-OBS gridded data set for daily mean wind speed over Europe in the period 1980-2021 , EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-423, https://doi.org/10.5194/ems2022-423, 2022.
The focus of this work is to create homogenised mean monthly air temperature series, produce monthly temperature grids and derive climate monitoring products to assess the state of air temperature and the observed temperature change in Croatia. Having that in mind, monthly mean temperatures from 122 Croatian stations are homogenised and high resolution monthly gridded data are developed for the 1981-2018 period. The hierarchical clustering is introduced to define climate regions in Croatia needed for homogenisation. The breaks of homogeneity are detected by the standard normal homogeneity test. Further on, the regression kriging is applied to produce 1 km x 1 km monthly grids for each month in the analysed period. The quality of the interpolation was assessed by leave-one-out cross-validation and the root mean square error of 0.7°C. The quality of spatial interpolation is estimated with normalised error maps. Climate normals and trends are derived from homogenised station data and monthly grids. After 2000, average annual anomalies from the 30-years climate normal 1981-2010 were positive and up to 1.4°C warmer than the average, and just occasionally negative. The significant strong warming is observed and mapped over the entire Croatian territory in April, June, July, August and November, being stronger inland than on the coast. Annual trends were significant and in the range from 0.3°C/decade to 0.7°C/decade. That suggests that our region could face consequences such as devastating heatwaves, water shortages, loss of biodiversity and risks to food production, especially as being part of the Mediterranean where it seems that the observed trends are 2-2.5 times stronger than the global mean. We can hope that some of the presented climate monitoring products can help in assessing the vulnerability and the risk from climate change and help with the mitigation of the potentially affected sectors like forestry, agronomy, tourism, water management, energy production or consumption, health or others.
How to cite: Perčec Tadić, M., Pasarić, Z., and Guijarro, J. A.: CroMonthlyGrids - monthly air temperature grids for climate monitoring and climate change detection, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-620, https://doi.org/10.5194/ems2022-620, 2022.
The MISH (Meteorological Interpolation based on Surface Homogenized Data Basis; Szentimrey and Bihari) software was developed at the Climate Department of Hungarian Meteorological Service (OMSZ) specifically for interpolation of climate elements. At OMSZ the MISH is currently using to interpolate temperature, precipitation, wind, global radiation, pressure and humidity data. The software consists of two major parts: the modelling of climate statistical parameters and the interpolation subsystem. A good quality modelling requires long homogenized data sets, but the representativity of the station system is also very important. Therefore, as many measurements as possible should be considered in the modelling process.
Interpolation in MISHv1.03 is based on statistical parameters that are determined in advance in the modelling part by using climate data series. It means that this modelling has to be done only once, before the interpolation. If the modelling is made from a homogenised data set of many stations, it is possible to interpolate well with few predictors. At the Hungarian Meteorological Service, archived data are continuously recorded and new meteorological stations are being installed. The previous MISH modelling procedure for temperature data series was done using 58 stations, but this has been renewed and, recently we estimate climate statistical parameters using data of 112 stations.
In this presentation, we present the results of the new modelling for mean, minimum and maximum temperature values. In all three cases, our results show a significant improvement on previous results. We can say that we can provide more accurate estimates with few predictors for locations in Hungary where measurements are not available. The new modelling also gives us a better understanding of the climate of the past centuries, since we can make more accurate estimates from fewer measurements. It also allows us to report climate change in more detail.
How to cite: Izsak, B., Szentes, O., and Bihari, Z.: Modelling climate statistical parameters by MISH interpolation procedure, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-422, https://doi.org/10.5194/ems2022-422, 2022.
Precipitation is a globally fundamental parameter that has an influence in many fields and determines the hydro-meteorological cycle.
This year, the Global Precipitation Climatology Centre (GPCC) updates the daily in-situ precipitation product Full Data Daily (FDD). The FDD is a reliable raster product, based on global land-surface precipitation totals. The database includes data provided by national meteorological and hydrological services, regional and global data collections as well as WMO GTS-data. The FDD is characterized in particular by the detailed quality control of the input data.
The new FDD v2022 covers a period of 39 years from Jan 1982 - Dec 2020 with daily temporal resolution and 1° spatial resolution based on the modified SPHEREMAP interpolation scheme.
This paper introduces a test study on the methodological decisions for FDD and its effects on the extreme value indices (ETCCDI). For this purpose, we tested different methods for a selected time period in conjunction with the update of the GPCC FDD v2022. The study aims to analyze 1) the impact of the interpolation method on the ETCCDI. We recalculated a sub-period of FDD using modified Spheremap or krigging as interpolation scheme. Two two global data sets are generated each with a different interpolation method. 2) We examine the influence of the computing order on the ETCCDI (1°x1° resolution) by a) calculating the extreme value indices ETCCDI station based first and interpolating later and b) interpolating first and then calculating the ETCCDI. An existing difference is expected, as these are both answers to different questions. However, the quantification of this difference is intended to answer frequently asked questions regarding the data set.
This gives us the opportunity to estimate and quantify the influence of the methodical decisions on the ETCCDI and to improve the understanding of the FDD.
How to cite: Rustemeier, E., Ziese, M., Schneider, U., Finger, P., and Hänsel, S.: Evaluation of the interpolation scheme and order of calculation for extreme value indices based on the updated GPCC Full Data Daily v2022, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-484, https://doi.org/10.5194/ems2022-484, 2022.
