Search Results
A short guide to increase FAIRness of atmospheric model data
The generation, processing and analysis of atmospheric model data are expensive, as atmospheric model runs are often computationally intensive and the costs of ‘fast’ disk space are rising. Moreover, atmospheric models are mostly developed by groups of scientists over many years and therefore only few appropriate models exist for specific analyses, e.g. for urban climate. Hence, atmospheric model data should be made available for reuse by scientists, the public sector, companies and other stakeholders. Thereby, this leads to an increasing need for swift, user-friendly adaptation of standards.The FAIR data principles (Findable, Accessible, Interoperable, Reusable) were established to foster the reuse of data. Research data become findable and accessible if they are published in public repositories with general metadata and Persistent Identifiers (PIDs), e.g. DataCite DOIs. The use of PIDs should ensure that describing metadata is persistently available. Nevertheless, PIDs and basic metadata do not guarantee that the data are indeed interoperable and reusable without project-specific knowledge. Additionally, the lack of standardised machine-readable metadata reduces the FAIRness of data. Unfortunately, there are no common standards for non-climate models, e.g. for mesoscale models, available. This paper proposes a concept to improve the FAIRness of archived atmospheric model data. This concept was developed within the AtMoDat project (Atmospheric Model Data). The approach consists of several aspects, each of which is easy to implement: requirements for rich metadata with controlled vocabulary, the landing pages, file formats (netCDF) and the structure within the files. The landing pages are a core element of this concept as they should be human- and machine readable, hold discipline-specific metadata and present metadata on simulation and variable level. This guide is meant to help data producers and curators to prepare data for publication. Furthermore, this guide provides information for the choice of keywords, which supports data reusers in their search for data with search engines. © 2020 The authors
UTLS temperature validation of MPI-ESM decadal hindcast experiments with GPS radio occultations
Global Positioning System (GPS) radio occultation (RO) temperature data are used to validate MPI-ESM (Max Planck Institute – Earth System Model) decadal hindcast experiments in the upper troposphere and lower stratosphere (UTLS) region between 300 hPa and 10 hPa (8 km and 32 km) for the time period between 2002 and 2011. The GPSRO dataset is unique since it is very precise, calibration independent and covers the globe better than the usual radiosonde dataset. In addition it is vertically finer resolved than any of the existing satellite temperature measurements available for the UTLS and provides now a unique one decade long temperature validation dataset. The initialization of the MPI-ESM decadal hindcast runs mostly increases the skill of the atmospheric temperatures when compared to uninitialized climate projections with very high skill scores for lead-year one, and gradually decreases for the later lead-years. A comparison between two different initialization sets (b0, b1) of the low-resolution (LR) MPI-ESM shows increased skills in b1-LR in most parts of the UTLS in particular in the tropics. The medium resolution (MR) MPI-ESM initializations are characterized by reduced temperature biases in the uninitialized runs as compared to observations and a better capturing of the high latitude northern hemisphere interannual polar vortex variability as compared to the LR model version. Negative skills are found for the b1-MR hindcasts however in the regions around the mid-latitude tropospheric jets on both hemispheres and in the vicinity of the tropical tropopause in comparison to the b1-LR variant. It is interesting to highlight that none of the model experiments can reproduce the observed positive temperature trend in the tropical tropopause region since 2001 as seen by GPSRO data.
A generalized formulation of the dynamic Smagorinsky model
A generalized formulation of the Dynamic Smagorinsky Model (DSM) is proposed as a versatile turbulent momentum diffusion scheme for Large-Eddy Simulations. The difference to previous versions of the DSM is a modified test filter range that can be chosen independently from the resolution scale to separate the impact of the test filter on the simulated flow from the impact of the resolution. The generalized DSM (gDSM) in a two-dimensional version is validated in a verification study as a horizontal momentum diffusion scheme with the Kühlungsborn Mechanistic General Circulation Model at high resolution (wavenumber 330) without hyperdiffusion. Three-day averaged results applying three different test filters in the macro-turbulent inertial range are presented and compared with analogous simulations where the standard DSM is used instead. The comparison of the different filters results in all cases in similar globally averaged Smagorinsky parameters cS ≃ 0.35 and horizontal kinetic energy spectra. Hence, the basic assumption of scale invariance underlying the application of the gDSM to parameterize atmospheric turbulence is justified. In addition, the smallest resolved scales contain less energy when the gDSM is applied, thus increasing the stability of the simulation.
Publication of Atmospheric Model Data using the ATMODAT Standard
Scientific data should be published in a way so that other scientists can benefit from these data, enabling further research. The FAIR Data Principles are defining the basic prerequisite for a good data publication: data should be Findable, Accessible, Interoperable, and Reusable. Increasingly, research communities are developing discipline-specific data publication standards under consideration of the FAIR Data Principles. A very comprehensive yet strict data standard has been developed for the climate model output within the Climate Model Intercomparison Project (CMIP), which largely builds upon the Climate and Forecast Metadata Conventions (CF conventions). There are, however, many areas of atmospheric modelling where data cannot be standardised according to the CMIP data standard because, e.g., the data contain specific variables which are not covered by the CMIP standard. Furthermore, fulfilling the strict CMIP data standard for smaller Model Intercomparison Projects (MIPs) requires much effort (in time and manpower) and hence the outcome of these MIPs often remains non-standardised. For innovative model diagnostics, preexisting standards are also not flexible enough. For that reason, the ATMODAT standard, a quality guideline for atmospheric model data, was created. The ATMODAT standard defines a set of requirements that aim at ensuring the high reusability of atmospheric model data publications. The requirements include the use of the netCDF file format, the application of the CF conventions, rich and standardised file metadata, and the publication of the data with a DataCite DOI. Additionally, a tool for checking the conformity of data and metadata to this standard, the atmodat data checker, was developed and is available on GitHub under an open licence. By using the more flexible ATMODAT standard, the publication of standardised datasets is simplified for smaller MIPs. This standardisation process is presented as an example using the data of an aerosol-climate model from the AeroCOM MIP. Furthermore, the landing pages of ATMODAT-compliant data publications can be highlighted with the EASYDAB logo. EASYDAB (Earth System Data Branding) is a newly developed quality label for carefully curated and highly standardised data publications. The ATMODAT data standardisation can easily be transferred to data from other disciplines and contribute to their improved reusability.
Fire, late frost, nun moth and drought risks in Germany's forests under climate change
Ongoing climate change affects growth and increases biotic and abiotic threats to Germany's forests. We analysed how these risks develop through the mid-century under a variety of climate change scenarios using the process-based forest model 4C. This model allows the calculation of indicators for fire danger, late frost risk for beech and oak, drought stress and nun moth risk. 4C was driven by a set of 4 simulations of future climate generated with the statistical model STARS and with 10 simulations of future climate based on EURO-CORDEX model simulations for the RCP2.6, RCP4.5 and RCP8.5 pathways. A set of about 70000 forest stands (Norway spruce, Scots pine, beech, oak, birch), based on the national forest inventory describing 98.4 % of the forest in Germany, was used together with data from a digital soil map. The changes and the range of changes were analysed by comparing results of a recent time period (1971–2005) and a scenario time period (2011–2045). All indicators showed higher risks for the scenario time period compared to the recent time period, except the late frost risk indicators, if averaged over all climate scenarios. The late frost risk for beech and oaks decreased for the main forest sites. Under recent climate conditions, the highest risk with regard to all five indicators was found to be in the Southwest Uplands and the northern part of Germany. The highest climate-induced uncertainty regarding the indicators for 2011–2045 is projected for the East Central Uplands and Northeast German Plain.