Enter Zoom Meeting

AS1.4

EDI
High-resolution weather and climate simulation

This session invites presentations on high-resolution simulations of weather and climate. This includes state-of-the-art global storm-resolving simulations for weather and climate prediction but also large-eddy simulations and high-resolution ocean modelling. Presentations can cover developments to improve model fidelity (e.g. via improved parametrisations), detailed studies of modelled phenomena at high-resolution (e.g. tropical cyclones) and the impacts of ocean-atmosphere coupling. However, reflecting the technical challenges of such simulations, we also welcome presentations about computational concerns, such as the effective use of heterogeneous supercomputers (including GPUs), domain-specific languages, and the development of new, efficient dynamical cores. We also welcome presentations from participants of international projects related to high-resolution weather and climate simulation, such as DYAMOND, PRIMAVERA, Destination Earth, NextGEMS and WarmWorld.

Convener: Samuel HatfieldECSECS | Co-conveners: Peter Düben, Claudia Frauen, Daniel Klocke, Vera Schemann
Presentations
| Fri, 27 May, 08:30–11:02 (CEST)
 
Room 1.34

Fri, 27 May, 08:30–10:00

Chairperson: Samuel Hatfield

08:30–08:37
Introduction & Overview of High-Resolution Numerical Weather Prediction Activities at ECMWF

08:37–08:44
|
EGU22-9111
|
Virtual presentation
Masahiro Tanoue et al.

The stable water isotopes (SWIs) (δ18O and δD) are used as an indicator of the intensity of the atmospheric hydrological cycle due to their large variability in time and space. SWIs are used for investigating the model’s bias and uncertainty. In this study, we developed a new global storm-resolving model equipped with SWIs (NICAM-WISO). We applied the new model to conduct three current climate simulations using a single-moment cloud microphysics scheme, without any convection parameterization scheme: CTRL, LRES, and HRES. These simulations used the same physical process but at a different horizontal resolution (LRES, 224 km; CTRL, 56 km; HRES, 14 km). We conducted the simulations on the supercomputer Fugaku. CTRL reproduced the seasonal means of the atmospheric hydrological cycle, as well as precipitation isotopic ratios. However, all simulation results have three types of biases. First, in tropical ocean regions, the model had a negative bias in precipitation isotopic ratios; this was caused by a negative bias in vapor isotopic ratios for the middle troposphere, which resulted from excess condensation biases during upward transportation and high-frequency deep convection. Second, all simulations overestimated precipitation isotopic ratios in the East Asia summer monsoon region due to low precipitation in the region caused by a shift in the moisture convergence zone from eastern China to the western Pacific Ocean. Third, in cold continental regions such as Siberia, Greenland, and Antarctica, the model had a positive bias in precipitation isotopic ratios due to a moisture bias and a low temperature effect; these regions also had a large positive bias in terms of precipitation deuterium excess. A particularly large bias was observed in ice clouds with low ice water content, indicating uncertainties in the vapor deposition process. Together, these results suggest that stable water isotopes are helpful for identifying biases associated with cloud microphysics and the atmospheric hydrological cycle. The unique constraints of stable water isotopes revealed cloud microphysics uncertainty and biases in the hydrological simulations.

How to cite: Tanoue, M., Yashiro, H., Takano, Y., Yoshimura, K., Kodama, C., and Satoh, M.: Modelling water isotopes using a global non-hydrostatic model with explicit convection scheme, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9111, https://doi.org/10.5194/egusphere-egu22-9111, 2022.

08:44–08:51
|
EGU22-2617
|
ECS
|
|
Virtual presentation
Yihui Zhou et al.

Simulating diurnal cycle of rainfall is a difficult challenge for general circulation models. We developed a global unstructured mesh model, Global-to-Regional Integrated forecast SysTem (GRIST), targeting at unified weather-to-climate forecast. The performance of the model in simulating the summer precipitation over East Asia has been evaluated. Yet the performance from a global perspective remains less understood. In this study, we focus on the simulations of precipitation diurnal cycle during boreal summer, and examine four AMIP simulation results of the GRIST model. These configurations mainly differ in the horizontal resolution. Thus, they reflect the direct changes due to varying resolutions. By refining the resolution over East Asia (VR-EA) and North America (VR-NA) respectively, we analyze the similarities and differences in model behaviors in simulating diurnal cycle of precipitation over these two refinement regions. VR-EA well reproduces the nocturnal rainfall, while VR-NA fails in certain regions respectively. The underlying responses to resolution of these two models are similar. For regions dominated by nocturnal rainfall, the refined resolution significantly increases the composited precipitation intensity at night up to the magnitude of the observation but has little impact on the composite percentage. The percentage of peak rainfall within 00-06h in the model over the Southern Great Plains remains lower than the observation as the resolution refines. Given the much lower occurrence frequency, the contribution of the intense precipitation to the climatological nocturnal rainfall amounts is small in VR-NA. Over East Asia, since the precipitation frequency is comparable to the observation, VR-EA benefits from the increased precipitation intensity due to higher resolution. No apparent artificial features are observed in the transition zone of the variable-resolution mesh. The results suggest that the variable-resolution modeling is cost-effective for simulating the diurnal cycle of climatological summer precipitation.

How to cite: Zhou, Y., Zhang, Y., and Yu, R.: Global variable-resolution model simulation of rainfall diurnal cycle during boreal summer, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2617, https://doi.org/10.5194/egusphere-egu22-2617, 2022.

08:51–08:58
|
EGU22-2695
|
Virtual presentation
Yi Zhang et al.

This work investigates the resolution sensitivity of an explicit dynamics-microphysics coupled system using the GRIST nonhydrostatic model, with varying uniform horizontal resolutions (120 km, 60 km, 30 km, 15 km, 5 km). The experiments follow the DYAMOND (DYnamics of the Atmospheric general circulation Modeled On Non-hydrostatic Domains) winter protocol that covers a 40-day integration from UTC00, 20th  to UTC00, Jan to 29th, Feb, 2020. The five simulations did not activate parameterized convection, and no specific tuning of model physics is employed such that the direct resolution response of a fixed model system can be examined. One 120-km run with parameterized convection is done to serve as a coarse-resolution reference. Other model configurations for different runs are kept as consistent as possible except certain small differences. Results demonstrate that the model gradually improve its representation of the fine-scale features (e.g., kinetic energy spectra) as resolution increases. In terms of 40-day averaged climate, the 5-km run has an overall more realistic simulation of the rainfall distribution than lower-resolution simulations without parameterized convection (e.g., spatial distribution). Most zonally averaged climate statistics are less prone to be altered by the resolution, except those fields associated with cloud water (e.g., shortwave cloud radiative forcing). This finding was also reached by an earlier study using the ICON model. Though with better fine-scale details, the coarse-resolution averaged features of the 5-km model without parameterized convection do not necessarily (and automatically) gets better than a 120-km simulation with parameterized convection. The tropical rainfall frequency-intensity spectra become more realistic in the 5-km no-convection run, but the 120-km run with parameterized convection shows a more realistic zonally averaged mean state. This impies more development and tuning efforts are still required for global km-scale models.

How to cite: Zhang, Y., Liu, Z., and Li, J.: Resolution sensitivity of GRIST Nonhydrostatic Model During DYAMOND winter from 120 km to 5 km, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2695, https://doi.org/10.5194/egusphere-egu22-2695, 2022.

08:58–09:05
|
EGU22-1214
|
ECS
|
|
On-site presentation
Shuchang Liu et al.

Non-hydrostatic km-scale weather and climate models are promising in simulating clouds, especially convective ones. However, even km-scale models need to parameterize some physical processes and are thus subject to the corresponding uncertainty of parameters. Systematic calibration has the advantage of improving model performance with transparency and reproducibility, thus benefiting model intercomparison projects, process studies, and climate-change scenario simulations. 

In this paper, the regional atmospheric climate model COSMO v6 is systematically calibrated over the Tropical South Atlantic. First, the parameters' sensitivities are evaluated with respect to a set of validation fields (outgoing longwave radiation (OLR), outgoing shortwave radiation (OSR) and latent heat flux (LHFL)). Five of the most sensitive parameters are chosen for calibration. The objective calibration then closely follows the methodology of Bellprat et al. (2016). This includes simulations considering the interaction of all pairs of parameters and the exploitation of a quadratic-form metamodel to emulate the simulations. In the current set-up with 5 parameters, 50 simulations are required to build the metamodel. Then Latin hypercube sampling is applied and the set of parameters with the best performance score is chosen as the optimal parameter set. The model is calibrated for the year 2016 and validated in 2013. And  the optimal parameter setting lead to significant improvements for both years, especially for OSR, which is closely related to low clouds. More specifically, the domain annual mean OSR bias is reduced from 40 to 13.5 Wm-2. Moreover, when we apply the optimal setting over a larger domain with a slightly higher resolution (from 4km to 3km) in 2006, the optimal setting still works, especially for OSR and for the calibrated domain. 

The results thus show that parameter calibration is a useful and efficient tool for model improvement. We will also discuss potential limitations and highlight how the approach could be extended to global atmospheric models. Calibrating over a larger domain might help improve the overall performance, but would potentially also lead to compromises among different regions and variables, and require more computational resources.

How to cite: Liu, S., Zeman, C., Sørland, S. L., and Schär, C.: Systematic Calibration of A Convection-Resolving Model: Application over Tropical Atlantic, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1214, https://doi.org/10.5194/egusphere-egu22-1214, 2022.

09:05–09:12
|
EGU22-1924
|
ECS
|
On-site presentation
Claudia Stephan et al.

Eleven 40-day long integrations of five different global models with horizontal resolutions of less than 9 km are compared in terms of their global energy spectra. The method of normal-mode function decomposition is used to distinguish between balanced (Rossby wave; RW) and unbalanced (inertia-gravity wave; IGW) circulation. The simulations produce the expected canonical shape of the spectra, but their spectral slopes at mesoscales, and the zonal scale at which RW and IGW spectra intersect differ significantly. The partitioning of total wave energies into RWs an IGWs is most sensitive to the turbulence closure scheme and this partitioning is what determines the spectral crossing scale in the simulations, which differs by a factor of up to two. It implies that care must be taken when using simple spatial filtering to compare gravity wave phenomena in storm-resolving simulations, even when the model horizontal resolutions are similar. In contrast to the energy partitioning between the RWs and IGWs, changes in turbulence closure schemes do not seem to strongly affect spectral slopes, which only exhibit major differences at mesoscales. Despite their minor contribution to the global (horizontal kinetic plus potential available) energy, small scales are important for driving the global mean circulation. Our results support the conclusions of previous studies that the strength of convection is a relevant factor for explaining discrepancies in the energies at small scales. The models studied here produce the major large-scale features of tropical precipitation patterns. However, particularly at large horizontal wavenumbers, the spectra of upper tropospheric vertical velocity, which is a good indicator for the strength of deep convection, differ by factors of three or more in energy. High vertical kinetic energies at small scales are mostly found in those models that do not use any convective parameterisation.

How to cite: Stephan, C., Duras, J., Harris, L., Klocke, D., Putman, W. M., Taylor, M., Wedi, N. P., Žagar, N., and Ziemen, F.: Atmospheric energy spectra in global kilometre-scale models, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1924, https://doi.org/10.5194/egusphere-egu22-1924, 2022.

09:12–09:19
|
EGU22-2905
|
ECS
|
|
On-site presentation
Isabel H. Smith et al.

Atmospheric turbulence has a serious, dangerous, and costly impact on aviation. Turbulence makes up most weather-related in-flight accidents and costs the global aviation sector up to US$1 billion every year. Upper level turbulence can be broken down into four main types: Clear-Air Turbulence (CAT), Convectively Induced Turbulence (CIT), Near-Cloud Turbulence (NCT), and Mountain Wave Turbulence (MWT). Aviation is often impacted by CAT, which is not visible on radar and is therefore extremely hard to detect in advance of an encounter. Previous literature has shown that climate change is strengthening CAT globally, with increased severity particularly over the North Atlantic, a busy flight route, within the winter months. These findings have been based on CMIP3 and CMIP5 climate models, which have now been superseded by CMIP6 (Coupled Model Intercomparison Project Phase 6) models with higher resolution. 

In this presentation we build and develop these previous findings further by using the CMIP6 HighResMIP PRIMAVERA simulations, which have grid spacings from 135km to 25km. CAT has not previously been investigated with models that come this close to resolving individual patches of turbulence. Comparisons between several resolutions have given us a better understanding of how different climate models, and their grid spacings, represent turbulence. Despite some multidecadal and yearly variability, CAT is found to increase in frequency, in all turbulent severities, in time and with increased near-surface temperatures. Interestingly, atmosphere-only global climate models predict a smaller increase in CAT, in comparison to coupled atmosphere-ocean models. Our findings suggest that an increasing mean near-surface temperature over the North Atlantic will lead to further light to severe turbulence events, which results in extremely bumpy air travel, longer travel times, and increased CO2 emissions into the atmosphere. 

How to cite: Smith, I. H., Williams, P. D., and Schiemann, R.: Using high-resolution climate models to predict increases in atmospheric turbulence , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2905, https://doi.org/10.5194/egusphere-egu22-2905, 2022.

09:19–09:26
|
EGU22-4433
|
ECS
|
Virtual presentation
Ann Kristin Naumann and Monika Esch

In global storm-resolving models (SRMs), that resolve convection explicitly instead of parameterizing it, microphysical processes are now fundamentally linked to their controlling factors, i.e., the circulation. While in conventional climate models the convective parameterization is one of the main sources of uncertainties (and a popular tuning parameter), this role might be passed on to the microphysical parameterization in global SRMs. In this study, we use a global SRM with two different microphysical schemes. For each scheme we do several sensitivity runs, where in each run we vary one parameter of the applied microphysics scheme in its range of uncertainty. We find that the two microphysics schemes have distinct signatures, e.g., in how condensate is partitioned between ice and snow. In addition, perturbing single parameters of each scheme also affects condensate amounts and hence the heat budget of the tropics. Among the parameters tested, the model is particularly sensitive to the ice fall speed and the width of the raindrop size distribution, which both cause several 10s W/m2 variation in radiative fluxes. Overall, microphysical sensitivities in global SRMs are substantial and resemble inter-model differences such as in the DYAMOND ensemble. 

How to cite: Naumann, A. K. and Esch, M.: Microphysical sensitivities in global storm-resolving simulations, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-4433, https://doi.org/10.5194/egusphere-egu22-4433, 2022.

09:26–09:33
|
EGU22-5030
|
ECS
|
On-site presentation
Jérôme Neirynck et al.

Before an off-shore wind farm is built a thorough resource assessment of all available locations for the farm needs to be performed. Since the power extraction of a wind farm depends on the cube of the wind speed even the mesoscale variability in the wind speed plays an important role in the resource assessment of a wind farm. In order to study mesoscale systems that occur in the vicinity of off-shore wind farms we've set up a convection permitting simulation in COSMO-CLM for the Kattegat sea strait. The Kattegat is particularly interesting since it is an area which features a very irregularly shaped coastline and pronounced coastal effects. Centrally located in the Kattegat lays the 400 MW Anholt wind farm. Operational data of the Anholt wind farm and scatterometer data of the Kattegat are used to validate our simulation. A relatively good agreement between observations and the model output has been found. A variety of mesoscale systems has been identified, both in unstable (e.g. a downburst) as in stable (e.g. gravity waves) conditions. The wind speed variability on temporal scales and on spatial scales over the Kattegat has been investigated. The interactions of the Anholt wind farm with these systems have been investigated using the COSMO-CLM model which incorporates the Fitch wind farm parametrisation. This research is part of a larger project aiming at developing a fast and accurate resource planning and forecasting platform for off-shore wind farms. More information about this project can be found on freewind-project.eu.

How to cite: Neirynck, J., Stoffelen, A., Meyers, J., and van Lipzig, N.: Mesoscale weather systems and their interactions with windfarms: A study for the Kattegat., EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5030, https://doi.org/10.5194/egusphere-egu22-5030, 2022.

09:33–09:40
|
EGU22-7657
|
ECS
|
On-site presentation
Theresa Lang et al.

The dry subsidence regions of the tropics and subtropics play an important role in setting the Earth’s clear-sky climate sensitivity, as the clear-sky feedback in these regions is particularly sensitive to both the baseline relative humidity (RH) and small RH changes under warming. Therefore, it is crucial that climate models reliably simulate the RH and its response to warming in these regions. However, considerable inter-model differences in RH remain, also in global storm-resolving models, the newest generation of climate models with horizontal grid spacings sufficient to explicitly resolve deep convection. The goal of this study is to identify potential causes for these inter-model differences and understand the mechanisms behind it. For this we examine the effect of changes in different model parameters – including microphysical parameters and vertical grid spacing – on tropical free-tropospheric humidity in a global storm-resolving model, focusing on the dry subsidence regions. Back-trajectory calculations allow us to determine the characteristics of the last saturation points for dry tropical air masses as well as the magnitude of moisture sources and sinks during subsequent advection, and how both change in the sensitivity experiments. The trajectory analysis confirms that moisture gains and losses during advection play a secondary role in setting the RH distribution in tropical dry zones in the model, as suggested by earlier studies based on coarser models. This leaves changes in the points of last saturation, which are determined by the circulation and the temperature field, as the more likely driver of RH changes. Preliminary results from the sensitivity experiments indicate that particularly changes in the vertical grid spacing of the model can affect the RH in tropical subsidence regions. These RH changes are explained by changes in the temperature of the main outflow regions of deep convection in the upper troposphere, where most last saturation points are located. These results highlight the importance of circulation and temperature differences across global storm-resolving models in driving inter-model differences in RH.

How to cite: Lang, T., Naumann, A. K., Schmidt, H., and Buehler, S. A.: Understanding drivers of inter-model differences in tropical free-tropospheric humidity in global storm-resolving models , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7657, https://doi.org/10.5194/egusphere-egu22-7657, 2022.

09:40–09:47
|
EGU22-8575
|
ECS
|
|
On-site presentation
Theresa Kiszler et al.

In the context of the Transregional Collaborative Research Center on "Arctic Amplification: Climate Relevant Atmospheric and Surface Processes, and Feedback Mechanisms”, we challenge the ICOsahedral Non-hydrostatic modelling framework ICON by performing simulations in a complex Arctic environment. Our study aims at adding a significant reference for how well ICON can perform in the Arctic and give ideas on how to improve the performance related to the microphysical parameterizations.

With the ambition to resolve the clouds directly, we used ICON in the large-eddy mode (ICON-LEM), which enables the use of a 3D Smagorinsky turbulence scheme. We further applied a two-moment microphysics scheme. The setup consists of a circular domain with 600 m resolution centred around Ny-Ålesund (Svalbard) with approx. 100 km diameter. As forcing, hourly data from a 2.4 km ICON-NWP simulation covering a limited area around the archipelago of Svalbard was used. These NWP simulations were forced with the operational global ICON forecasts. Ny-Ålesund was chosen because of its intricate topography, heterogenic surfaces and availability of observational data for comparisons.

The setup was run semi-operationally for 24 h on a daily basis for several months and therefore we were able to create statistics based on an outstandingly large data set. Using the columnar output of Ny-Ålesund we compared it to a large variety of observations (e.g. liquid water path, wind and relative humidity). This evaluation showed an astonishingly high agreement between the measurements and the simulations. For instance, the orographically influenced flow, as well as seasonal and short-range changes in humidity, are captured. Certain aspects, such as the formation of liquid vs ice in clouds, need improvement. On the whole, we could show that ICON-LEM is a useful tool to study the Arctic atmosphere and its changing climate. Further, we can continue to get a better picture of possibilities to understand the microphysical processes and improve their representation in the model.

This work was supported by the DFG funded Transregio-project TR 172 “Arctic Amplification (AC)3“.

How to cite: Kiszler, T., Chellini, G., Ebell, K., Kneifel, S., and Schemann, V.: A performance baseline for the representation of clouds and humidity for cloud-resolving ICON-LEM simulations, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8575, https://doi.org/10.5194/egusphere-egu22-8575, 2022.

09:47–09:54
|
EGU22-9043
|
ECS
|
|
Virtual presentation
Bernard Alan Racoma et al.

In this study, we examine the effect of the Cordillera Mountain Range (CMR) in Luzon, Philippines on Tropical Cyclone (TC) precipitation. Using the Weather Research and Forecasting model, we simulated multiple TC events with three different terrain profiles: control, reduced CMR, and enhanced CMR. We find that for most of the TC cases overland precipitation increases as mountain height increases. To further understand the interaction between TC precipitation and the mountain range, we examine the effects of relevant dynamical fields, including mountain slope, incoming perpendicular wind speed, and the moist Froude Number (Fw). We highlight that TC precipitation is strongly and positively correlated with the product of approaching wind speeds and mountain slope. It is hypothesized that stronger winds along steeper mountain slopes translate to vertical motion which in turn causes higher amounts of precipitation, especially during TC events. In contrast,  the linear relationships with other variables are less clear. It is also worth noting that a significant weakening of TCs may cause less rainfall overland, which is an indirect effect of the mountain range on TC precipitation. Understanding the interactions between TCs and mountain ranges may help in regional quantitative precipitation forecasting efforts in the mountainous regions of the Philippines.

How to cite: Racoma, B. A., Holloway, C., Schiemann, R., Feng, X., and Bagtasa, G.: The Effect of Topography on Tropical Cyclone Precipitation in the Philippines, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9043, https://doi.org/10.5194/egusphere-egu22-9043, 2022.

Fri, 27 May, 10:20–11:50

Chairperson: Samuel Hatfield

10:20–10:27
|
EGU22-9110
|
On-site presentation
Hauke Schmidt and Sebastian Rast

In global atmospheric modeling the importance of an appropriate ratio of vertical to horizontal model resolution has been emphasized earlier. Theoretical considerations for appropriate ratios have been based, e.g., on quasi-geostrophic considerations for large-scale flows and the dissipation conditions for gravity waves. In limited-area convection-permitting simulations it has been shown that in particular the simulation of shallow cloud layers depends on the vertical model resolution. 
A recent focus in global climate modeling is to increase horizontal resolutions down to a few kilometers grid spacing in order to resolve processes like convection that need to be parameterized at coarser resolutions. In these simulations, often the vertical model resolutions haven’t been changed much in comparison to traditional approaches. Questions like the following may arise: Is this appropriate? How strongly does the climate at storm-resolving horizontal scales depend on vertical resolution? Can convergence of the simulated climate be expected at a certain vertical resolution? Is it useful to invest in further increases of horizontal resolution without a refinement of the vertical grid?
To start answering these questions we have performed simulations with the ICON global atmospheric model at a horizontal resolution of 5 km with three different vertical grids comprising 55, 110, and 190 layers and corresponding vertical resolution in the troposphere of 400, 200, and 100 m, respectively, for a period of 6 weeks.  Here we will show the dependence of selected climate parameters, including the global energy budget, on the vertical resolution. 

How to cite: Schmidt, H. and Rast, S.: The dependence of the climate simulated in a global storm-resolving model on its vertical resolution, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9110, https://doi.org/10.5194/egusphere-egu22-9110, 2022.

10:27–10:34
|
EGU22-10757
|
ECS
|
Virtual presentation
Thomas Rackow et al.

We give an overview of the global coupled storm-resolving simulations performed so far with IFS-NEMO and IFS-FESOM2 for the H2020 Next Generation Earth Modelling Systems (NextGEMS) project. The project aims to build a new generation of eddy- and storm-resolving global coupled Earth System Models. Such models will constitute the substrate for prototype digital twins of Earth as envisioned in the EU’s ambitious Destination Earth project.

NextGEMS relies on several model development cycles, in which the models are run and improved based on feedback from the analysis of successive runs. In an initial set of storm-resolving coupled simulations, the models were integrated for 75 days, starting in January 2020. ECMWF’s Integrated Forecasting System (IFS) has been run at 9km and 4km global spatial resolution. The runs at 9km were performed with the deep convection parametrization, while at 4km, the IFS was run with and without the deep convection parametrization. So far, the underlying ocean models NEMO and FESOM2 were run on an eddy-permitting 0.25° resolution grid in a single-executable configuration with IFS. Based on the analysis by project partners during a Hackathon organised in October, several key issues were identified both in the runs with IFS, and in those run with the second storm-resolving coupled model developed in NextGEMS, ICON.

We will describe the model improvements made to IFS-NEMO/FESOM based on the lessons learned from the first runs, which will be included for the second round of simulations. These mainly consist in vastly improved conservation properties of the coupled model systems in terms of water and energy balance, which are crucial for longer climate integrations, and in a much more realistic representation of the snow and surface drag. The second round of NextGEMS simulations will also target eddy-resolving resolution in large parts of the global ocean (better than 8km) to resolve mesoscale eddies and leads in sea ice. This is thanks to a refactored FESOM2 ocean model code that allows for efficient coupled simulations in the single-executable context with IFS via hybrid parallelization with MPI and OpenMP.

How to cite: Rackow, T., Becker, T., Pedruzo Bagazgoitia, X., Sandu, I., Zampieri, L., and Ziemen, F. and the ECMWF-AWI Team: Storm-resolving simulations with IFS-NEMO/FESOM in the NextGEMS project, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10757, https://doi.org/10.5194/egusphere-egu22-10757, 2022.

10:34–10:41
|
EGU22-11478
|
ECS
|
On-site presentation
Karol Corko and Ulrike Burkhardt

The high-resolution DYAMOND simulations resolve much of the cloud relevant dynamics and cause a large improvement in the structure and diurnal cycle of clouds and precipitation. Nevertheless, from DYAMOND simulations we know that cloud properties can vary significantly even in high-resolution simulations. We focus on evaluating and if possible constraining ice cloud processes in the tropics, an area that should particularly benefit from the increased resolution because deep convection is resolved and controls the tropical upper tropospheric water budget. We analyse not only the horizontal distribution of IWP but also the cloud phase and cloud vertical structure as they are crucial to Earth’s radiation budget.

When comparing the high-resolution global simulations performed within the DYAMOND project among each other and with passive remote sensing data and ERA5 reanalysis we find that the horizontal distribution of ice water path (IWP) varies significantly. In order to understand better those differences, we analysed the connection between the simulated vertical velocity and the total IWP, and the water path of the individual hydrometeors. While the PDF of tropical vertical velocity simulated by the different models is quite similar, the total ice water path connected with those vertical velocities varies strongly. In most models, high vertical velocities are connected with significantly higher IWP than liquid water path (LWP) except in the ICON simulations which simulates similarly large increases in IWP and LWP. Most models simulate large increases in larger ice hydrometeors for large vertical velocities while FV3 simulates also large increases in ice water connected with deep convection. Differences in cloud phase e.g. when comparing NICAM and ICON simulations are connected with different vertical distributions of the condensate with NICAM IWC reaching higher atmospheric levels than the ICON IWC. We attempt to constrain the vertical distribution using active remote sensing data. 

How to cite: Corko, K. and Burkhardt, U.: Impact of tropical convection on upper tropospheric cirrus in high resolution DYAMOND simulations, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11478, https://doi.org/10.5194/egusphere-egu22-11478, 2022.

10:41–10:48
|
EGU22-12292
|
ECS
|
Gaston Irrmann et al.

AGRIF (Adaptive Grid Refinement In Fortran) is a package for the integration of full adaptive mesh refinement features within a multidimensional finite difference model. This library is used in ocean models like NEMO (Nucleus for European Modelling of the Ocean) to offer the possibility to run multiple levels and 2-way nested embedded zooms. Within the ESiWACE2 project, AGRIF performance have been addressed toward high resolution simulations. 
First, a simple AGRIF configuration within NEMO has been set up to simplify benchmarking, profiling and testing new optimizations ideas. We selected on purpose a configuration with small MPI sudomains to mimic simulations running on high numbers of core. Second, a profiling analysis has let us identify an important overhead. Indeed, on a zoom with a refinement of a factor 3 in both latitude and longitude covering 1/9 of the simulated domain an overhead of 46% has been observed compared with the theoretical performance. The correction of land points used in the interpolation on the zoom has been found to be a major bottlenecks. Third, we implemented an optimization concerning the correction on land point limiting as much as possible the computations and taking advantage of the specificity of each interpolation. This adjustment provided us with a reduction of 25% of the time to solution in the aforementioned configuration. For future work, we identified numerous optimizations including further optimizations of the correction of land points.

How to cite: Irrmann, G., Masson, S., Debreu, L., Guibert, D., and Raffin, E.: Optimizations of Multiscale Simulation with AGRIF, towards Exascale Applications, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12292, https://doi.org/10.5194/egusphere-egu22-12292, 2022.

10:48–10:55
|
EGU22-12876
Rossella Ferretti et al.

On the morning of 10 July 2019, an intrusion of relatively cold and dry air, over the Adriatic Sea, through a "bora jet", gave rise to a frontal structure at the ground, which moved rapidly from the Northern to the Southern Adriatic. The intense thermal gradient (together with a high positive sea surface temperature anomaly), the interaction of the jet with the complex topography of Apennines  and the coastal boundary, generated a storm structure that moved parallel to the central Italy coast. In particular, between 8UTC and 12UTC, a supercell developed along the coast to the north of Pescara city (middle Adriatic), producing rainfall that reached 130 mm in 3 hours, and a violent hailstorm (estimated diameter greater than 10 cm). 

In this work, the frontal dynamics and the genesis of the thunderstorm are studied using the numerical system COAWST. Local polarimetric radar observations are also used to check the consistency of the simulations in the mature phase of the supercell. Numerical experiments are performed using a 1 km grid over central Italy, initialized using the ECMWF IFS analysis/forecasts. The sensitivity study investigates the role of the orography, the sea surface temperature (SST) and the coupling between ocean and atmosphere. Orography tests include simulations where the relevant peaks of the Apennine range  (such as Gran Sasso and Picentini) are removed as well as cases where their peaks are modified compared to their real values. In terms of SST, we employ, using an uncoupled approach, the ECMWF SST dataset, the MFS-CMEMS Copernicus dataset at 4 km, 0.01°C Satellite SST, and we investigate the role of the SST anomaly (adding +1°C and +2°C to the real field). The role of the ocean-atmosphere interaction is tested using the COAWST numerical model using an ocean model numerical grid at 1 km resolution over the whole Adriatic Sea. 

The preliminary results show that the topography and in particular  the interaction with the peaks of the Apennine range plays a fundamental role in the dynamics of the cold pool that trigger the convective system. Also, the SST anomaly is found to play an important role in the development of the supercell. In particular, we observed that the simulations forced with MFS-CMEMS SST and the COAWST model runs produce a very realistic SST, in terms of spatial and temporal distribution, but colder by about 1.5 °C in absolute value if compared to observed satellite data. This difference generates lower heat fluxes, less evaporation, weaker precipitations and smaller hail than using warmer SSTs. 

How to cite: Ferretti, R., Mazzarella, V., Marzano, F., Miglietta, M. M., Picciotti, E., Montopoli, M., Baldini, L., Vulpiani, G., Tiesi, A., Mazzà, S., and Ricchi, A.: Analysis of the development mechanisms of a large-hail storm event, on the Adriatic Sea using an atmosphere-ocean coupled model (COAWST), EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12876, https://doi.org/10.5194/egusphere-egu22-12876, 2022.

10:55–11:02
|
EGU22-12956
|
On-site presentation
Lloyd Treinish et al.

Numerical simulations at cloud-resolving scales have becoming practical for both research and operational applications due to advances in computing technology.  However, deploying such capabilities beyond a limited scale (e.g., extended metropolitan region, large watershed) typically remains out of reach due to the computational cost and the complexity of the systems to support such work.  Yet such capabilities are needed to address the local impacts of precipitation events that can impact much broader areas.  In particular, convective storms driven by monsoons remain unresolved by current numerical weather prediction systems applied to sub-Saharan Africa.  To address this problem, the African Rainfall Project (ARP) was initiated to deploy the community Weather Research Forecast (WRF) model across this region at 1x1 km horizontal resolution on the World Community Grid (WCG).  WRF is configured to capture a diversity of geographic conditions in the region with appropriate boundary layer, land surface and cloud microphysics and parameterizations in addition to high vertical and temporal resolution.  WCG provides a fully distributed computational environment that crowd-sources unused computing power from volunteers’ devices and donates it to scientific projects.  As such, all computations must be embarrassingly parallel, which creates a challenge for models like WRF.  Hence, each instance of WRF must operate serially on a volunteer’s device.  To address the regional-scale simulations, sub-Saharan Africa is decomposed into individual 52 by 52 km domains at 1x1km as the third nest in two-way telescoping grids with common centroids.  The outer domains are at 3 and 9 km resolution, respectively with the same vertical resolution.  Each 48-hour simulation is done as a cold-start forced by reanalysis with output saved every 15 minutes.  The collection of these simulations will cover at least one year to capture seasonal variations.  Since there is no operational imperative, the ability of typical volunteer’s system to compute each simulation in several hours is practical.  Scaling is achieved with many thousands of systems being deployed simultaneously.  With this decomposition, over 35000 overlapping domains cover the region.  During post-processing, the individual simulations are stitched together to create a consistent, single output for over for the period of study.  Although the focus is precipitation, the simulations provide additional standard output for 2m temperature and 10m horizontal wind velocity, for example.  We will report on the results to date and validation in comparison to in situ (e.g., from TAHMO, www.tahmo.org) and remotely sensed observations as well as conventional WRF deployments for a large computational domain covering a small subset of the region.

How to cite: Treinish, L., van de Giesen, N., and Le Coz, C.: Meso-gamma-scale numerical weather simulations for sub-Saharan Africa via grid-based, distributed computing, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12956, https://doi.org/10.5194/egusphere-egu22-12956, 2022.