Coupled Sun-to-Earth models represent a key part of the future development of space weather forecasting. With respect to predicting the state of the thermosphere and ionosphere, there has been a recent paradigm shift; it is now clear that any self-respecting model of this region needs to include some representation of forcing from the lower atmosphere, as well as solar and geomagnetic forcing. Here we assess existing modeling capability and set out a road map for the important next steps needed to ensure further advances. These steps include a model verification strategy, analysis of the impact of nonhydrostatic dynamical cores, and a cost-benefit analysis of model chemistry for weather and climate applications.Plain Language Summary Numerical models that comprehensively simulate the region between the Sun and the Earth represent a key part of the future development of space weather forecasting. With respect to predicting the Earth's upper atmosphere, there has been a recent paradigm shift; it is now clear that any self-respecting model of this region needs to include some representation of impacts from below (the lower atmosphere) as well as from above (solar variability and the effects of solar wind fluctuations). Here we assess existing modeling capability and set out a road map for the important next steps needed to ensure further advances. These steps include a strategy for checking the accuracy of the models, an analysis of the impact of methods chosen to represent upper atmosphere dynamics, and an assessment of the relative benefits of comprehensive (but expensive) and simplified (but inexpensive) model representations of upper atmosphere chemistry. Key Points:• We have reached a paradigm shift, where any self-respecting space weather model of the upper atmosphere now needs to have some representation of the lower atmosphere • Further model developments are required in several key areas, including dynamical cores and the improved representation of gravity waves • A road map of future actions is presented to ensure good progress continues to be made; this includes the development of a multi-model verification strategy

Space weather driven atmospheric density variations affect low Earth orbit (LEO) satellites during all phases of their operational lifetime. Rocket launches, re-entry events and space debris are also similarly affected. A better understanding of space weather processes and their impact on atmospheric density is thus critical for satellite operations as well as for safety issues. The Horizon 2020 project Space Weather Atmosphere Model and Indices (SWAMI) project, which started in January 2018, aims to enhance this understanding by: Developing improved neutral atmosphere and thermosphere models, and combining these models to produce a new whole atmosphere model. Developing new geomagnetic activity indices with higher time cadence to enable better representation of thermospheric variability in the models, and improving the forecast of these indices. The project stands out by providing an integrated approach to the satellite neutral environment, in which the main space weather drivers are addressed together with model improvement. The outcomes of SWAMI will provide a pathway to improved space weather services as the project will not only address the science issues, but also the transition of models into operational services. The project aims to develop a unique new whole atmosphere model, by extending and blending the Unified Model (UM), which is the Met Office weather and climate model, and the Drag Temperature Model (DTM), which is a semi-empirical model which covers the 120–1500 km altitude range. A user-focused operational tool for satellite applications shall be developed based on this. In addition, improved geomagnetic indices shall be developed and shall be used in the UM and DTM for enhanced nowcast and forecast capability. In this paper, we report on progress with SWAMI to date. The UM has been extended from its original upper boundary of 85 km to run stably and accurately with a 135 km lid. Developments to the UM radiation scheme to enable accurate performance in the mesosphere and lower thermosphere are described. These include addition of non-local thermodynamic equilibrium effects and extension to include the far ultraviolet and extreme ultraviolet. DTM has been re-developed using a more accurate neutral density observation database than has been used in the past. In addition, we describe an algorithm to develop a new version of DTM driven by geomagnetic indices with a 60 minute cadence (denoted Hp60) rather than 3-hourly Kp indices (and corresponding ap indices). The development of the Hp60 index, and the Hp30 and Hp90 indices, which are similar to Hp60 but with 30 minute and 90 minute cadences, respectively, is described, as is the development and testing of neural network and other machine learning methods applied to the forecast of geomagnetic indices.

Ray-tracing techniques have been used to investigate numerical effects on the propagation of acoustic and gravity waves in a non-hydrostatic dynamical core discretized using an Arakawa C-grid horizontal staggering of variables and a Charney-Phillips vertical staggering of variables with a semi-implicit timestepping scheme. The space discretization places limits on resolvable wavenumbers, and redirects the group velocity and the propagation of wave energy towards the vertical. The time discretization slows the wave propagation while maintaining the group velocity direction. Wave amplitudes grow exponentially with height due to the decrease in the background density, which can cause instabilities in whole-atmosphere models. Although molecular viscosity effectively damps the exponential growth of waves above about 150 km, additional numerical damping might be needed to prevent instabilities in the lowermost thermosphere. These results are relevant to the Met Office Unified Model, and provide insight into how the stability of the model may be improved as the model's upper boundary is raised into the thermosphere.

A coupled Sun-to-Earth model is the goal for accurate forecasting of space weather. A key component of such a model is a whole atmosphere model – a general circulation model extending from the ground into the upper atmosphere – since it is now known that the lower atmosphere also drives variability and space weather in the upper atmosphere, in addition to solar variability. This objective motivates the stable extension of The Met Office’s Unified Model (UM) into the Mesosphere and Lower Thermosphere (MLT), acting as a first step towards a whole atmosphere model. At the time of performing this research, radiation and chemistry schemes that are appropriate for use in the MLT had not yet been implemented. Furthermore, attempts to run the model with existing parameterizations and a raised upper boundary led to an unstable model with inaccurate solutions. Here, this instability is examined and narrowed down to the model’s radiation scheme – its assumption of Local Thermodynamic Equilibrium (LTE) is broken in the MLT. We subsequently address this issue by relaxation to a climatological temperature profile in this region. This provides a stable extended UM which can be used as a developmental tool for further examination of the model performance. The standard vertical resolution used in the UM above 70 km is too coarse (approx. 5 km) to represent waves that are important for MLT circulation. We build on the success of the nudging implementation by testing the model at an improved vertical resolution. Initial attempts to address this problem with a 3 km vertical resolution and a 100 km lid were successful, but on increasing the resolution to 1.5 km the model becomes unstable due to large horizontal and vertical wind velocities. Increasing the vertical damping coefficient, which damps vertical velocities near the upper boundary, allows a successful year long climatology to be produced with these model settings. With the goal of a whole atmosphere model we also experiment with an increased upper boundary height. Increasing the upper model boundary to 120 and 135 km also leads to stable simulations. However, a 3 km resolution must be used and it is necessary to further increase the vertical damping coefficient. This is highly promising initial work to raise the UM into the MLT, and paves the way for the development of a whole atmosphere model.

Ray tracing techniques have been used to investigate numerical effects on the propagation of acoustic waves in a non-hydrostatic dynamical core discretised using an Arakawa C-grid horizontal staggering of variables (Arakawa & Lamb 1977) and a Charney-Phillips vertical staggering of variables (Charney & Phillips 1953) with a semi-implicit timestepping scheme. It is found that the space discretisation places limits on resolvable wavenumbers and redirects the group velocity of waves towards the vertical. Wave amplitudes grow exponentially with height due to the decrease in the background density, which can cause instabilities in whole-atmosphere models. However, the inclusion of molecular viscosity and diffusion acts to damp the exponential growth of waves above about 150 km. This study aims to demonstrate the extent to which numerical wave propagation causes instabilities at high altitudes in atmosphere models, and how processes that damp the waves can improve these model’s stability.