- Challenges in developing high-resolution mesoscale models with a focus on end-users and the EUMETNET forecasting programme. Observation impact studies to assess the importance of different parts of the observing system for global and limited area NWP models.
- Numerics and physics-dynamics coupling in weather and climate models: This encompasses the development, testing and application of novel numerical techniques, the coupling between the dynamical core and physical parameterizations, variable-resolution modelling, as well as performance aspects on current and future supercomputer architectures.
- Model verification: Developments and new approaches in the use of observations and verification techniques. It covers all verification aspects from research to applications to general verification practice and across all time and space scales. Highly welcome verification subjects including high-impact, user oriented applications, warnings against adverse weather events or events with high risk or user relevance.
In the framework of the flagship EU Destination Earth initiative, the On-Demand Extremes digital twin project aims at delivering an event- and user-driven weather-induced engine for selected impact sectors. The core of the project lies in the design and development of an operational capability for an on-demand workflow on Europe's fastest supercomputers, efficiently exploiting large data volumes coming from observations and hectometric limited-area numerical weather prediction (NWP) simulations for the benefit of a variety of users. Within the project, air quality (AQ) is one of the chosen impact sectors to highlight the DT engine capabilities.
Poor air quality is a major cause of premature death and disease, and the single largest environmental health risk in Europe. AQ models are routinely run at meteorological and environmental agencies to identify contributions to air quality problems and assist in the design of effective strategies to reduce harmful air pollutants.
The talk will showcase the work done during the first phase of the project in the design and development of high-resolution HARMONIE-AROME NWP experiments within the Extremes DT infrastructure Prototype whose output is given as input for AQ model runs. Initially focusing on a historical summer heatwave case, the presentation will highlight the impact of higher resolution and other configuration parameters as well as the steps taken to deliver the NWP output variables to the AQ model developers on the target high-performance computing facilities.
Example use cases for air quality end users will be shown and next steps will be discussed in the context of the second phase of the project, which aims at delivering an integrated end-to-end, on-demand NWP-AQ workflow.
How to cite: Benacchio, T., Thorsen, S. B., Baordo, F., and Yang, X.: HARMONIE-AROME high-resolution experiments in the Destination Earth On-demand Extremes digital twin: an air quality models and end users perspective, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-550, https://doi.org/10.5194/ems2024-550, 2024.
This presentation introduces the limited-area version of the Korean Integrated Model (KIM). KIM is a global atmospheric weather prediction model which has been operational in producing medium-range forecasts for the Korea Meteorological Administration since 2019. In efforts to develop a high-resolution short-range forecasting system for the Korean Peninsula and East Asia, we developed a model framework capable of running KIM as a limited-area model (LAM). The LAM version shares model components with the global KIM model, including the cubed-sphere grid system, spectral element method solver, non-hydrostatic dynamical core, and scale-aware physics package; with the distinction that the model domain is confined to the first panel of the cubed-sphere. The related modules to accommodate this configuration in the model framework has been newly developed accordingly. Utilizing the Schmidt transformation of the cubed-sphere enables adjustment of the size and center position of the first panel through stretching and rotating the cube. The functionality of the LAM framework is demonstrated through idealized dynamics-only test runs nesting global KIM simulations. In real-case dynamical downscaling experiments refining 100 km global simulations to 25 km, the LAM exhibits stable simulations beyond 10 days of integration, closely aligning with global simulation results. Minimal generation of numerical spurious waves is observed, even without utilizing Newtonian-relaxations toward the global simulation near the lateral boundaries, suggesting the effectiveness of dynamical cores with spectral element solvers for dynamical downscaling. The presentation also addresses challenges related to high computational costs, sensitivities to varying domain size and resolution, and the model's applicability to operational kilometer-scale short-range weather prediction.
How to cite: Cho, H., Kim, J., Noh, I., and Jang, J.: Limited-Area Version of the Korea Integrated Model for the High-Resolution Regional Forecast System, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-586, https://doi.org/10.5194/ems2024-586, 2024.
The model’s mean orography acts as the boundary condition for the model dynamics and the drag from resolved orographic gravity waves can have a significant impact on the large-scale atmospheric circulation in weather and climate models. As we approach km-scale horizontal resolutions in global models, more of the orographic spectrum and the impact of orography becomes resolved. Benefits of increased resolution, for example better prediction of orographic rain, can only be harvested if the resolution of the mean orography field is also increased. However, even at kilometre scale, some of the orographic variance will not be represented on the model grid and must be parameterised.
Initial simulations for Destination Earth’s global Digital Twin at 4.4 km horizontal resolution have shown a negative wind bias over Eastern Asia and increasing forecast error compared to the operational 9 km model, indicating that the higher resolution of resolved orography causes additional small horizontal-scale orographic gravity waves breaking above the mid-latitude jet, which affects the global circulation. Therefore, we focused on improving the mean orography processing and sub-grid scale orography parameterisations with the aim to find a scale-independent formulation that maintains good forecast skill at operational model resolutions and profits from increased resolution at kilometre scale.
The processing of the source data was simplified and harmonised across resolutions using conservative interpolation of the source dataset. A new source dataset for surface elevation with 30 m resolution is used. A small increase to the spectral filtering of mean orography showed an improvement in forecast skill over the Tibetan plateau, and the updated fields describing sub-grid orographic features yield a more consistent behaviour across different resolutions. However, the sub-grid orography has significantly changed, and the parameters of the orographic parameterisation schemes needed to be optimised again considering an appropriate formulation across resolutions. We present a new approach to this parameter re-tuning, using Bayesian parameter optimisation, which enables an efficient workflow for simultaneously optimising several interdependent parameters.
How to cite: Sützl, B., van Niekerk, A., Beljaars, A., Maciel, P., Choulga, M., Janoušek, M., Vannière, B., Forbes, R., Balsamo, G., Sandu, I., and Dueben, P.: Optimising orography for global high-resolution simulations, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-259, https://doi.org/10.5194/ems2024-259, 2024.
Atmospheric flows display phenomena on a very wide range of spatial scales that interact with each other. Many strongly localized features, such as complex orography, can only be modelled correctly if a very high spatial resolution is employed, especially in the lower troposphere, while larger scale features such as high/low pressure systems and stratospheric flows can be adequately resolved on much coarser meshes. The insufficient resolution of orographic features is compensated in numerical weather predictions (NWP) and climate models by subgrid-scale orographic drag parameterizations, which are essential for an accurate description of atmospheric flows with models using feasible resolutions. The interplay between resolved and parameterized orographic effects is critical, since many operational models employ resolutions in the so-called 'grey zone', for which some orographic effects are well resolved while others still require parameterization. Global simulations without drag parameterization have shown that the increase in forecast skill for increasing atmospheric resolution was mainly due to the improved representation of the orography.
Because of these considerations, NWP is an apparently ideal framework for adaptive numerical approaches. However, mesh adaptation strategies have only slowly found their way into the NWP literature, due to limitations of earlier numerical methods, concerns about the accuracy of the representation of atmospheric wave phenomena for variable resolution meshes, and the complexity of an efficient parallel implementation for non-uniform or adaptive meshes. We present a quantitative assessment of the static local mesh refinement capabilities of a recently proposed IMEX-DG method [1] to a number of benchmarks for atmospheric flows over both idealized and real orography. We show that simulations with adaptive meshes around orography can increase the accuracy of the local flow description without affecting the larger scales, thereby significantly reducing the overall number of degrees of freedom compared to uniform mesh simulations [2, 3]. Importantly, no spurious reflections arise at internal boundaries separating mesh regions with different resolution and correct values for the momentum flux are retrieved. Both on idealised benchmarks and on test cases over real orographic profiles, simulations using non-conforming meshes correctly reproduce the larger scale, far-field response with meshes that are relatively coarse over most of the domain. This supports the idea that locally refined meshes can be an effective tool to reduce the dependence of NWP on parametrizations of orographic effects.
[1] G. Orlando, T. Benacchio, and L. Bonaventura. “An IMEX-DG solver for atmospheric dynamics simulations with adaptive mesh refinement”. Journal of Computational and Applied Mathematics 427 (2023), p. 115124.
[2] G. Orlando, T. Benacchio, and L. Bonaventura. "Robust and accurate simulations of flows over orography using non-conforming meshes". 2024. arXiv:2402.07759.
[3] G. Orlando, T. Benacchio, and L. Bonaventura. "Impact of curved elements for flows over orography with a Discontinuous Galerkin scheme". 2024. arXiv: 2404.09319.
How to cite: Orlando, G., Benacchio, T., and Bonaventura, L.: Robust and accurate simulations of flows over orography using non-conforming variable resolution meshes, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-301, https://doi.org/10.5194/ems2024-301, 2024.
Numerical simulation of the atmosphere plays a crucial role in understanding and anticipating extreme weather events. Steady advances in computing power have made it possible to increase the complexity and range of scales represented by numerical simulation. However, the advent of heterogeneous computing architectures with multi-core central processing units (CPUs) and graphics processing units (GPUs) requires atmospheric codes to be adapted.
Here we describe the adaptation of the Meso-NH non-hydrostatic mesoscale atmospheric model of the French research community. The Fortran code is ported to GPUs by including OpenACC directives to the most computationally expensive parts of the code. This approach allows running the same code on CPUs and on hybrid CPUs/GPUs architectures. To guarantee the accuracy of the port and the absence of bugs, measures have been taken to ensure bit-to-bit reproducibility between executions on these two architectures. A critical point lies in the atmospheric pressure solver, which requires the inversion of an elliptic equation. A geometric multigrid inversion algorithm is integrated, because the fast Fourier transforms approach used in the original version of the code becomes inefficient with a high number of GPUs. Currently, the code runs on different GPU-NVIDIA and GPU-AMD platforms and scales efficiently up to at least 1,024 GPUs, achieving a 3x increase in energy efficiency compared to CPUs only.
First scientific applications focus on the simulation of extreme weather events across scales as part of a Grand Challenge GPU pilot project on the AMD-based Adastra supercomputer of GENCI (same architecture as Frontier, the 1st exascale supercomputer). Three representative storms are simulated: a North Atlantic windstorm associated with a mid-latitude cyclone, a Mediterranean convective storm characterized as a derecho, and a mesoscale convective system over the Amazon rainforest. Representation of the North Atlantic storm requires downscaling from the synoptic cyclone scale (>100 km) down to local wind gust formation (<1 km). Inversely, the representation of the Amazon storm requires upscaling from the local triggering of convective cells (<1 km), which organize and maintain the system at the mesoscale (>100 km). Finally, the Mediterranean storm involves both up- and downscaling. We show that Meso-NH successfully represents the cascade of scales for the three representative storms for horizontal grid spacing down to 100 m and grid size up to 4096x4096x128 points.
Porting Meso-NH to GPUs opens up new opportunities to simulate extreme weather events across scales and paves the way for future European exascale supercomputers.
How to cite: Escobar, J., Wautelet, P., Pianezze, J., Chaboureau, J.-P., Dauhut, T., Barthe, C., Brumer, S., and Pantillon, F.: Porting the Meso-NH atmospheric model to GPU architectures allows simulating extreme weather events across scales, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-726, https://doi.org/10.5194/ems2024-726, 2024.
NOAA is undertaking the development of a Seasonal Forecast System version 1 (SFSv1) to replace the current Climate Forecast System version 2 (CFSv2) due to its limitations and the need for enhanced forecast accuracy. The SFS's development is primarily based on the capabilities of the NOAA Global Ensemble Forecast System (GEFSv13) which utilizes a non-hydrostatic Finite-volume dynamical core and is tailored for extended-range forecasting at a 25 km resolution. SFSv1, intended to operate at approximately 100 km, with a target resolution of about 50 km, requires careful examination to ensure the effectiveness of the existing physics suite and dynamics at these resolutions.
Within the resolution range of 50 km to 100 km, the hydrostatic assumption remains the hydrostatic assumption retains its viability, offering reduced computational costs and intimating that the non-hydrostatic alternative may be superfluous. To comprehensively evaluate these alternatives, sensitivity experiments with both AMIP-type and fully coupled experiments, have been conducted. Adjustments to the schemes and parameters governing artificial dissipation, horizontal advection, and vertical remapping have been made to align with the hydrostatic option.
The verification of atmosphere-only sensitivity experiments indicates that the hydrostatic option exhibits comparable performance compared to non-hydrostatic in terms of the prediction of large-scale environmental patterns. It also shows superior performance in predicting the quasi-biennial zonal wind oscillation at the stratosphere. However, further model refinement is deemed necessary, particularly when coupled with the ocean model to mitigate cold SST biases over the NINO3.4 region. Ongoing investigations are also addressing other issues arising from the transition to the hydrostatic option for SFS.
How to cite: Zhou, X. and Yang, F.: Hydrostatic Seasonal Forecast System Development within the Unified Forecast System at NOAA, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-846, https://doi.org/10.5194/ems2024-846, 2024.
Several studies have shown the added value of convection-permitting climate modelling, in particular regarding extreme events, including heat waves and high precipitation events (Lind et al., 2020, Fosser et al., 2024). HCLIM consortium develops a high resolution climate model, with a non-hydrostatic version using AROME physics. Current evaluation has shown that HCLIM is a robust tool able to add value when reproducing the behaviour of the climate system over complex areas (Belusic et al., 2020).
AEMET has been a member of HCLIM since its foundation and has participated in the effort to develop the model and provide a complete and adaptable tool. In particular, we present some results on the contributions from AEMET: as many basin resources depend significantly on groundwater, works focused on improving the surface component with a new watertable parametrization. On the other hand, AEMET has contributed actively to the development of a more detailed aerosol scheme. An active line of development is currently underway, focused on the development of a fully ocean-coupled model.
Future plans include the generation of high resolution climate simulations over Spain, assessing the added value of the model and exploring different strategies for the generation of high resolution climate scenarios feeding National Adaptation Plan. Some preliminary results, evaluating the added value against CORDEX simulations, are shown. Additionally, several tests comparing different strategies regarding model configuration and the impact of the domain size and nesting strategy have been conducted, giving insight into designing the strategy for the generation of an ensemble of simulations.
References:
Belušić, Danijel & de Vries, Hylke & Dobler, A. & Landgren, Oskar & Lind, Petter & Lindstedt, David & Pedersen, Rasmus & Sánchez-Perrino, Juan & Médus, Erika & Ulft, Bert & Wang, Fuxing & Andrae, Ulf & Batrak, Yurii & Kjellström, Erik & Lenderink, Geert & Nikulin, Grigory & Pietikäinen, Joni-Pekka & Rodriguez-Camino, Ernesto & Samuelsson, Patrick & Wu, Minchao. (2020). HCLIM38: a flexible regional climate model applicable for different climate zones from coarse to convection-permitting scales. Geoscientific Model Development. 13. 1311-1333. 10.5194/gmd-13-1311-2020. http://dx.doi.org/10.5194/gmd-13-1311-2020
Fosser, G., Gaetani, M., Kendon, E.J. et al. Convection-permitting climate models offer more certain extreme rainfall projections. npj Clim Atmos Sci 7, 51 (2024). https://doi.org/10.1038/s41612-024-00600-w
Lind, P., Belušić, D., Christensen, O.B. et al. Benefits and added value of convection-permitting climate modeling over Fenno-Scandinavia. Clim Dyn 55, 1893–1912 (2020). https://doi.org/10.1007/s00382-020-05359-3
How to cite: Prieto Rico, I., Sánchez Perrino, J. C., and Rodríguez Guisado, E.: Main regional climate modelling development at AEMET using HCLIM, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-1108, https://doi.org/10.5194/ems2024-1108, 2024.
Projections from global climate models (GCMs) are regularly used to create information for climate adaptation policies and socio-economic decisions. As demand grows for accuracy in these projections, GCMs are being run at increasingly finer spatiotemporal resolution to better resolve physical processes and consequently reduce uncertainty associated with parametrizations. Yet this increase in resolution and the consequent size of the data output makes the current state-of-the-art archives (e.g., CORDEX, CMIP) unfeasible. Moreover, the current archival method has left some data consumers without their required data due to the limited number of variables stored and their lower frequency (e.g., monthly means). Initiatives like Destination Earth are investigating the novel method of data streaming, where user applications can be run as soon as the required data is produced by the climate models. Data streaming allows users to access the climate data at the highest frequency possible (e.g., hourly) and native resolution in near real model run-time. This provides an unprecedented time-scale reduction to access the climate data compared with the current simulation paradigm and the possibility of using variables and frequencies not previously available.
Yet the advent of data streaming in the climate community poses its own set of challenges. Often users require climate data that spans long periods. For example, many hydrological impact models require daily, monthly or annual maximum precipitation values, while in the wind energy sector, accurate distributions of the wind speed over long periods are essential. Obtaining statistics for periods longer than the time the climate model output is accessible can no longer be done using traditional statistical algorithms. This introduces the one-pass problem; how to compute summaries, diagnostics or derived quantities that only see each data point once (i.e., pass through the data one time)?
We present here a detailed analysis on the use of one-pass algorithms to compute statistics on streamed climate data. Unlike traditional two-pass methods, one-pass algorithms do not have access to the full time series of data needed to estimate the statistic; instead, they process data incrementally every time that the model outputs new time steps. While these algorithms have been adopted in other fields such as online trading and machine learning, they have yet to find a foothold in climate science, mainly because they have not been necessary until now. Here we show how one-pass algorithms can be harnessed for use in Earth system digital twins, generating the statistics required by users with minimal loss in accuracy and bypassing unfeasible storage requirements.
How to cite: Grayson, K., Thober, S., Roura Adserias, F., Lacima-Nadolnik, A., Sharifi, E., and Doblas-Reyes, F.: Statistical summaries for streamed climate data, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-782, https://doi.org/10.5194/ems2024-782, 2024.
A refined 4D-Var assimilation system within DestinE allows us to assimilate the Meteosat-10/SEVIRI clear-sky radiances over Europe, as well as globally at a spatial scale of 75 km instead of the previous 125 km in the ECMWF Integrated Forecasting System (IFS). Higher resolution observations can potentially improve the analysis and therefore the prediction of extreme weather events over Europe, as well as globally. The effects of using higher resolution observations have been investigated with a detailed set of experiments and the impact on wind, temperature, and humidity has been evaluated. A broad range of experiments indicate that exploiting the higher spatial density clear-sky radiances leads to an improvement of humidity sensitive fields in short- range forecasts with the IFS as independently measured for example by instruments on low-earth- orbiting satellites (IASI, CrIS, SSMIS, or ATMS). Due to a reduced displacement and representativeness error, these changes could further lead to improvements in longer range forecasts as these errors propagate upscale nonlinearly. First experiments show an upscale propagation of initially very localised increments in the analysis fields of vertical wind, as well as humidity above the Pacific or the North Atlantic. Over time, these incremental improvements from the 4D-Var system lead to an improvement in forecast scores of the IFS up to 5 days ahead.
In addition, pre-processed GOES-16/ABI and GOES-18/ABI observations by NOAA have been assimilated with 10 min sampling rates at 75 km spatial density. Exploring how to best assimilate relatively small spatial and temporal scales for one geostationary satellite, will allow us to approach these smaller scales with other satellites such as HIMAWARI/AHI above the Pacific or MTG-I/FCI above Europe. Data from both satellites will be available for us early in 2024. Preliminary experiments demonstrate the ability of IFS to assimilate observations at the highest available temporal resolution for the GOES-16 and GOES-18 satellites. Higher resolution radiances observed at these shorter time intervals naturally capture smaller scale atmospheric features such as mesoscale convective systems. In our experiments, simultaneously assimilating observations at a higher spatial and temporal resolution leads to an impact that is only marginally better than assimilating higher density observations alone, suggesting a combined investigation of optimal time-assignment, as well as assessment of the observation error are needed to optimise the integration of rapid update measurements in 4D-Var.
How to cite: Schröttle, J., Lupu, C., and Burrwos, C.: Approaching the sub-mesoscale globally at 10 min temporal resolution through assimilating clear-sky radiances measured by geostationary satellites, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-109, https://doi.org/10.5194/ems2024-109, 2024.
Convection-permitting ensemble prediction systems (EPS) are often underdispersive. To build a more reliable EPS, it is crucial to represent not only uncertainty in initial and lateral boundary conditions (IBC) but also uncertainty arising from model deficiencies. Uncertainty representations like stochastic parameterizations can add substantial variability to deterministic forecasts. However, it has been shown that when additional model uncertainty representations are implemented into an existing EPS that includes IBC uncertainty, they increase the total ensemble variance only slightly. This study aims to quantify this redundancy in ensemble variance by introducing a new "variability budget" framework. The framework decomposes the total ensemble variance into the sum of individual variances and their correlations, measuring the efficiency in increasing the total variance in relation to pre-existing variance.
The variance budget framework is applied to a convective scale EPS based on operational IBC uncertainties, augmented by two model uncertainty representations, the physically-based stochastic perturbations scheme (PSP) and microphysical parameter perturbations (MPP). By adding these two model uncertainty representations we find a marginal increase in total variance of wind, temperature and humidity at various heights. In particular, PSP introduces variability at convection initiation, while MPP prolongs the lifetime of variance. Both model uncertainties work in sub regions of areas influenced by IBC uncertainty and impact convective activity especially at small scales. Since the impact of PSP and MPP is mostly negatively correlated with the existing impact, this only leads to a slight increase in the total variance. A flow-dependent assessment based on the strength of convective forcing reveals that the model uncertainties show larger variances in weakly-forced conditions but with stronger negative correlations. This negative correlation is primarily attributed to random displacements of convection, with a stronger effect during weak forcing when convection is more intermittent compared to strong forcing.
How to cite: Matsunobu, T., Keil, C., and Craig, G.: A Variance Budget to estimate the Growth and Interaction of Uncertainties on Convective Scales , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-353, https://doi.org/10.5194/ems2024-353, 2024.
Natural hazards associated with mesoscale processes such as summertime convective events pose a considerable threat to people and property. Yet, forecasting such events remains a challenge, even for the latest generation of high-resolution, convective-scale numerical weather prediction models. Their forecast quality in such cases will likely benefit from improved initial conditions in form of an improved data assimilation analysis. Thus, the assimilation of high-resolution measurements of the lower troposphere has a high potential to enhance the predictability of convective conditions. For example, recent studies by the German Weather Service (DWD) in Aachen and Lindenberg suggest a positive influence of additional Doppler wind lidars (DWLs) on the analysis. However, a thorough investigation of the impact of additional ground-based observations in complex terrain is still lacking.
The ‘Swabian MOSES’ campaign took place from June to August 2023 in the German Black Forest mountain range and deployed a spatially distributed network of instruments to observe the dynamic and thermodynamic characteristics of the lower troposphere. Among them was a network of 12 DWLs, which together have never been used for data assimilation experiments before. Here, we present a 3-months campaign re-analysis dataset that uses a wealth of remote sensing and in-situ campaign observations. These data are added to the regional forecasting system of the DWD, which employs the non-hydrostatic model ICON at 2 km resolution (ICON-D2) and the Kilometer Scale Ensemble Data Assimilation system (KENDA) with 40 ensemble members that uses a Local Ensemble Transform Kalman Filter. We assimilate additional vertical profiles of the horizontal wind retrieved from the DWLs, targeted radiosoundings released from two sites during intensive observation periods, ground based zenith path-delay observations from a (not yet operational) German-wide network of Global Navigation Satellite Systems receivers, and 2-meter temperature and relative humidity as well as surface pressure observations.
In this contribution, we present our experimental setup, address challenges associated with the assimilation of non-operational observations, and demonstrate how different observations influence the campaign re-analysis. Moreover, we compare the campaign re-analysis with a quasi-operational reference re-analysis without additional observations. Our main focus is on the representation of convective summertime conditions.
How to cite: Thomas, J., Geppert, G., Reich, H., Steinert, T., Anlauf, H., Keller, J., Knippertz, P., and Oertel, A.: A campaign re-analysis for ‘Swabian MOSES 2023’: how do high-resolution observations change the analysis of a convective-scale data assimilation system?, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-853, https://doi.org/10.5194/ems2024-853, 2024.
We formulate a novel spatial forecast verification methodology grounded in the geometric principles underlying optimal transport (OT). In its original form OT seeks to minimise the transport of mass between two distributions in space, providing a cost of transportation. Canonical examples come from logistics, such as finding the optimal route to distribute bread from bakeries to cafes. Since its initial formulation, OT theory has found many varying applications from signal processing and economics to meteorology and machine learning, notably through the famed Wasserstein distance. Its success is first due to Kantorovich’s dual formulation and more recently due to novel algorithms and GPU compute. Which combined allow regularised OT problems to be solved efficiently.
In this work we consider a precipitation field as a measure in 2D space and compute the unbalanced OT distance between a 2D observation and forecast field. By leveraging this geometric formulation, we find a summary metric (the objective), which itself has constituent parts indicating performance. Additionally, the methodology allows us to form transport maps. These maps highlight regions in the field which require the most transport to align with the observation. Alongside providing a visual representation of the error, this provides physical insight into transport error as the map will traverse along geodesics of the underlying space. This has direct implications for operational forecasters, giving clear, easy to understand illustrations of error, whilst simultaneously providing important information to researchers at forecasting services. This research is supported by UKRI NERC, Imperial’s Grantham Institute and the MET Office UK.
How to cite: Francis, J., Cotter, C., and Mittermaier, M.: Optimal Transport Tools for Spatial Forecast Verification, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-63, https://doi.org/10.5194/ems2024-63, 2024.
Location errors in precipitation forecasts are ubiquitous in high-resolution weather forecasts due to the misplacement of convective cells but also of mesoscale or synoptic-scale features such as convergence lines or low-pressure systems. However, those kinds of forecast errors never appear in isolation and are usually mixed with intensity errors and more generally with systematic model biases. Therefore, correctly disentangling the contribution of location errors in verification scores is challenging, but also essential to advance forecast quality due to a couple of reasons. Firstly, location errors will incur a double penalty in traditional point-by-point verification (one false alarm event and one missed event). As a result, traditional metrics have the undesirable property of penalizing more a forecast with a correct feature in the wrong location than a forecast that misses the feature. In second place, once a forecast has a location error, traditional metrics are insensitive to the magnitude of the displacement. Hence, traditional metrics are not good at detecting improvements in the size of location errors (i.e. they lack discrimination). Both problems imply that intrinsically better forecasts do not necessarily get better scores.
Here we showcase some novel ways to tackle those two issues from different perspectives. Regarding the first issue, we present a new decomposition of the Mean Squared Error into three positive definite terms, one of which is linked to the amount of double penalty. Then we show how this allows screening forecasts for their levels of double penalty.
Regarding the second issue, recent advances in spatial verification techniques have enabled estimating location errors of global precipitation forecasts by approximating the Wasserstein distance between unbiased fields. This technique constructs a transport plan to move all the precipitation water in the forecasts to the correct locations according to observations, from which a mean location error is computed. We have extended this technique to be able to work with biased fields and at the same time curtail the location errors to prevent large displacement errors that result from biases in regions far away. This enables us to detect improvements at global or regional scale in ECMWF forecasts.
Finally, we have also shown that there is a relationship between the travel plan employed to compute the mean location error and the Mean Absolute Error. This allows us to go back to the first issue again and assess what proportion of MAE is contributed by location errors of a certain scale.
How to cite: Lledó, L., Skok, G., and Haiden, T.: Quantifying double-penalty effects and mitigating their impact on forecast verification, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-50, https://doi.org/10.5194/ems2024-50, 2024.
As science and technology develop, the operation of global numerical weather prediction models at high-resolution is considered to demonstrate more accurate forecast performance. However, when using the high-resolution numerical forecast model, the size of the grid becomes smaller, the number of grids increases by the square of the horizontal resolution change rate, and a relatively small time-step size must be applied according to the Courant-Friedrichs-Lewy condition. For these reasons, the overall amount of calculation increases and the limits of computer resources are encountered.
Now, in this study, we intend to use a variable resolution system based on a stretched global grid in Korean Integrated Model (KIM) to see the effect of high-resolution numerical forecasts at least in the region of interest without increasing the number of grids due to resolution changes. This system uses the Schmidt transform to expect a high-resolution effect using fine grids for the area of interest depending on the contraction/relaxation magnification, and shows a low-resolution effect using coarse grids for the other side of the Earth. In other words, in a variable resolution system, only the size of grid in each region changes depending on the contraction/relaxation ratio without structural modification of the dynamical core in a uniform grid system, and the number of grids is the same as the number of uniform grids. For the real case, we will use the variable resolution system to show the effect of the high-resolution in the region of interest (surrounding of the Korea Peninsula) by adjusting the contraction/relaxation scale. As a primary result, this study aims to analyze the numerical results of forecast performance and computational efficiency in variable grid systems compared to uniform grid system in KIM.
How to cite: Nam, H. and Hahm, J.-H.: Effect of Variable Resolution System Based on Stretched Global Grid in KIM, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-395, https://doi.org/10.5194/ems2024-395, 2024.
Korean Integrated Model (KIM), a global weather prediction model, plans to extend into a variable resolution system covering the short to medium forecast range. To improve forecast skill, we intend to increase vertical resolution from 91 to 137 layers in KIM, which has been updated with an enhanced horizontal resolution of 8 km. The vertical coordinates of 137 layers (L137) followed those of the European Centre for Intermediate Forecasting (ECMWF) Integrated Forecasting System (IFS), with reduced vertical grid spacing throughout the troposphere and stratosphere compared to the 91 layers (L91). Here, we focus on understanding the sensitivity of performance to vertical resolution in the medium-range forecast. The overall conclusion for increased vertical resolution is that it shows mainly neutral in the northern and southern hemispheres, but also some impacts in the tropics, against the IFS analysis. One of the notable impacts of changes in vertical resolution is a decrease of temperature in the lower and upper troposphere, which results in a considerable degradation of the upper troposphere over the tropics. This is because the physical processes rely on vertical grid spacing, which induces radiative cooling while increasing cloud hydrometeors. The refinement of the vertical grid increased the planetary boundary layer (PBL) height, which enhanced vertical mixing results in cooling/moistening above the PBL and warming/drying within the PBL. This increases the amount of cloud water content, influencing radiative fluxes and contributing to the lower troposphere cooling. The refined vertical resolution increased the amount of cloud ice content remained in cumulus convection, wherever weak or robust in its occurrence. This contributes to the cooling in the upper troposphere over the tropics. Increasing vertical resolution is needed in improvement of physical processes.
How to cite: Park, J.-R., Lee, E., and Bae, S. Y.: Temperature sensitivity to vertical resolutions over tropics in KIM, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-397, https://doi.org/10.5194/ems2024-397, 2024.
The Korea Meteorological Administration (KMA) operates the Korea Local Analysis and Prediction System (KLAPS), a very short-range forecast model providing real-time and 12-hour forecasts information at 10-minute intervals at 5 km horizontal resolution.
The analysis process of KLAPS, based on the LAPS (Local Analysis and Prediction System) from NOAA, is the data assimilation system that ingests radar, satellite, aircraft, surface observations, and regional atmospheric model data to analyze 3-D atmospheric conditions. After that, to ensure that the momentum and mass fields are consistent with the cloud-derived vertical motions during the diabatic initialization process, optimizing the thermodynamic balance between wind and cloud fields is performed.
This diabatic initialization process is critical to the initial condition of KLAPS for improving performance of model forecasting. In this step, user-defined weights, derived from the known error characteristics of analysis and background models, are utilized to minimize the variational cost function. By fine-tuning the weights of the variational cost function, the gaps between the model’s spin-up and current weather conditions and short-term extrapolation are reduced.
Sensitivity experiments are performed to estimate optimal weights of the variational balance processes, achieving improved initial fields for KLAPS. The predictability of precipitation is also evaluated based on the improved initial fields to verify the effects of weight optimization in adiabatic initialization process. This indicates the significant impacts of improved initial fields on forecast accuracy. The optimization of weights in initialization process of KLAPS is expected that initial state of the model will be improved and these results can contribute to enhancement of predictability performance for precipitation of the KLAPS.
Acknowledgement: This work was supported by Development of Numerical Weather Prediction and Data Application Techniques (KMA2018-00721)
How to cite: Lee, S., Won, H. Y., Lee, K. H., and Kim, S.: A study on optimizing the diabatic initialization of very short-range forecast model in KMA, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-617, https://doi.org/10.5194/ems2024-617, 2024.
The Korea Peninsula is surrounded by the sea on three sides and is made up of mountainous areas more than 70% of the land, making it difficult to predict and verify precipitation using numerical models. To overcome these topographical features, a dense observation network and 10 radar site have been operated in Korea Meteorological Administration (KMA). Also, a regional prediction system based on the Korea Intergrated Model (RDAPS-KIM) has been operated since May 2022. RDAPS-KIM covers the East Asia region with a finer horizontal resolution(3km) than the global model KIM(12km) and simulates real complex terrain closely.
Traditional skill score to verify models assesses the rainfall prediction performance for the grid closest to single point locations, which has the double penalty problem that the forecast precipitation exhibits the same pattern as the observation but leads to worse prediction evaluations due to missing spatial displacement. This double penalty issue occurs more frequently in regional model when the rainbands were shifted, so it can lead to misconception that regional model underperform global model. In order to avoid this problem, spatial verification methods to evaluate probability of rainfall forecast in the surrounding area have been suggested in the previous study (Ebert, 2008). The Fraction Skill Score (FSS) is one of the spatial verification methods suggested Roberts and Lean (2008), which allows the comparison of predicted precipitation with a spatial truth fields such as radar data. Accordingly, by verifying prediction with probability, it is expected that the shortcomings of verification of high resolution models can be overcome. Also, the FSS is recommended for verification in numerical weather prediction models by World Meteorological Organization (WMO) (JWGFVR, 2013).
In this study, the models of KMA (KIM and RDAPS-KIM) are assessed using FSS method compared with different sized neighborhoods and various rainfall threshold for 3h-accumulated precipitation. Analysis indicated how the spatial scale influences the FSS values, showing that FSS increases as neighborhood size increases. The use of selected one proper neighborhood size pointed out that RDAPS-KIM lead to a high FSS than KIM, because it fits the rainbands better. The result shows that RDAPS-KIM is more efficient for heavy precipitation and local rainfall than KIM. In the future, we will operate the higher resolution 1km model to improve prediction performance in simulating severe weather events, and evaluate in combination with traditional techniques and probabilistic forecasts. It is expected that interpretation of accuracy of precipitation forecasts from various perspectives will be possible.
How to cite: Jeong, S., Lee, J., Lee, E., and Kim, S.: Analysis of the Fraction Skill Score for the rainfall verification of high resolution model in KMA, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-618, https://doi.org/10.5194/ems2024-618, 2024.
The Korea Meteorological Administration (KMA) has been operated a very short-range forecast system, named Korean Local Analysis and Prediction System (KLAPS), which runs every 10 minutes at 5 km horizontal resolution since 2019. To reflect rapidly changing atmosphere state in real-time, numerous data sources such as surface observations, upper-air measurement, radar data, satellite, aircraft reports, wind profiler data are ingested into the three-dimensional analysis. To date, 13 different types of observation data have been utilized, and for recent two years, two more observation data are tried to utilize newly to improve the forecast performances. The first one is a freezing level (0oC) data interpreted by a bright band due to the enhanced radar reflectivity. Another one is the commercial aircraft-based meteorological information from Automatic Dependent Surveillance Broadcast (ADSB) Mode-Selective (Mode-S). In this study, we present the method to utilize these data and the result of the very short-range precipitation forecast.
To ingest the freezing level data to KLAPS system, firstly each polar-form data at 10 radar station over the domain is converted into a cartesian coordinate system. And then the individual freezing level data are merged to three-dimension grid on KLAPS, and finally utilized for the temperature analysis with a climatological quality control procedure. Meanwhile, the ADSB Mode-S data are simply applied to KLAPS analysis system by referring to the Aircraft Meteorological Data and Relay (AMDAR) data package. But the Mode-S observation are required a data thinning process to reduce observation redundancy and correlated biases due to its high dense resolution in both space and time (Met Office, 2019). In this study, a median value of the Mode-S data within each 5km horizontal grid, 22-vertical level, and 10-minute time window, is newly considered as a thinning method. During the conference, the improved performances of KLAPS are going to be presented in details.
How to cite: Won, H. Y., Lee, K. H., and Kim, S.: A utilization of new observation data in KLAPS (Korean Local Analysis and Prediction System) and the impacts on a very short-range precipitation forecast , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-619, https://doi.org/10.5194/ems2024-619, 2024.
High resolution wind speed forecasts are crucial for a range of applications, including the management of onshore wind power generation. Conventional wind speed forecasting is bound to the coarse spatial resolution of NWP models of 2-30 km. The wind speed complementarity model (WiCoMo)* is a high-resolution wind downscaling model that provides distributions of annual wind speeds across Germany at a horizontal resolution of 25 m x 25 m. This work aims to combine high resolution wind downscaling with numerical weather prediction (NWP) models to improve accuracy and resolve local effects, particularly in complex terrain. Quantile mapping was used to derive a transfer function at each 25 m x 25 m grid cell based on annual historical wind speeds calculated by WiCoMo and the NWP models respectively. The function was then applied to hourly time series of NWP models to simulate downscaled predictions. In addition, power curves of wind turbines were used to calculate the onshore wind power output of Germany from the high-resolution forecast. Validation metrics were used to compare the performance of the WiCoMo-enhanced NWP models with raw NWP outputs. The analysis demonstrates that the WiCoMo-enhanced NWP models outperform raw NWP across all tested models. For the year 2022, the MAE of NEMS4 was reduced from 1.66 m/s to 1.13 m/s and for NEMSGLOBAL it improved by almost 33%. The MBE was reduced to near 0 in all cases. Furthermore, spatial evaluations show that local wind speed effects often falling below the grid size in NWP models, such as hilltop speed-up or sheltering valleys, are resolved by the downscaling. The study suggests that localized wind speeds at wind turbine sites improve the accuracy of wind power output predictions. However, several limitations are identified, including challenges in applying corrections during specific weather conditions. Additionally, the modelled wind power output could not be validated at single turbine sites, limiting the validity of estimates for the entire country. The study demonstrates the potential of WiCoMo-enhanced NWP models in improving wind speed forecasting capabilities. The findings have important implications for various applications, including renewable energy planning and risk assessment.
* Christopher Jung and Dirk Schindler. Introducing a new wind speed complementarity model. Energy, 265:126284, 2023. ISSN 0360-5442. doi: https://doi.org/10.1016/j.energy.2022.126284.
How to cite: Wenzel, C. and Pierotti, N.: Combining high-resolution wind downscaling with numerical weather prediction models, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-653, https://doi.org/10.5194/ems2024-653, 2024.
The NWP model-named the Korean Integrated Model (KIM), which was developed through the first-phase of Korean Institute of Atmospheric Prediction Systems (KIAPS) was adopted for the operational forecast at the Korea Meteorological Administration (KMA) in April 2020. KIM is a non-hydrostatic spectral-element model based on a cubed-sphere grid with the physics package for the medium-range prediction, and a hybrid four-dimensional ensemble-variation data assimilation (DA) system. The second-phase KIAPS has been developing both model physics, dynamics and DA to improve prediction performance. There have been two recent updates, one being a DA update including extending the usage of several satellite observations, which is mainly done by the KMA, and the other being a model physics done by the KIAPS. In this study, we evaluate the impact of model physics update, especially focusing on the high-impact weather events over the the Korean Peninsula.
In a previous study, we found that when the forecast lead time is longer, KIM simulates mostly rainfall in the West Sea without heavy rainfall inland. In addition, most of precipitation is produced by the cumulus parameterization scheme (CPS). We confirmed that suppression of deep convection by adjusting the trigger condition in the CPS can improve precipitation distribution and intensity for a heavy rainfall event on the Korean Peninsula. Therefore, including the modification of the CPS process, modified physical processes are updated.
In this study, we performe a full-cycle experiment for 10-day forecasts over each 2-month summer and winter, and evaluate the overall performance according to the update. By analyzing precipitation events, we will discover the key factors affecting performance of precipitation forecast. These results are expected to show improved forecasting performance compared to the previous version.
How to cite: Cho, S., Seol, K.-H., and Lee, E.-H.: Evaluation of numerical model simulations of precipitation events over Korean Peninsula, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-662, https://doi.org/10.5194/ems2024-662, 2024.
The on-demand Digital Twin Extreme (DTE) is part of the European Destination Earth initiative which, during phase 1 of the project (2022-2024), has involved the cooperation of a large number of European National Weather Centers. The goal of the on-demand DTE is to provide a technical framework which is capable of running different flavours of the ACCORD forecast models (e.g. HARMONIE-AROME, AROME and ALARO) at very high resolution. The forecast predictions are expected to be triggered on demand for high-impact weather events within a regional domain (the so-called Limited Area Model) at subkilometre scale (e.g. horizontal resolutions of 750, 500 or 200 metres). In this poster, we will present some case studies which were explored at the Danish Meteorological Institute (DMI) in the contest of DTE phase 1. We will provide an overview of the challenges we faced looking at different weather events (such as flooding, extreme winds and precipitation) when the high resolution forecast model is initialised by lateral boundary conditions which can be from either the operational ECMWF Integrated Forecasting System (about 9 km resolution) or the Global Continuous Extreme Digital Twin (about 4.5 km resolution). Development and testing of data assimilation within the high resolution domain were not considered in the phase 1 of the project. Hence, in the case studies we focused on exploring the benefits and limitations of the LAM high resolution forecasts when the system is simply initialised with one of the ECMWF global systems. Overall, we will show the predictability capacity of the on-demand DTE and the added value of the increased model horizontal resolution.
How to cite: Baordo, F., Benacchio, T., Alerskans, E., Yang, X., and Thorsen, S.: Destination Earth On-demand Extremes case studies: results, challenges and lessons learnt at DMI, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-690, https://doi.org/10.5194/ems2024-690, 2024.
