New and emergent computing architectures and software engineering practices provide an opportunity for environmental models to be deployed more efficiently and democratically. In this paper we aim to capture the software engineering practices of environmental scientists, highlight opportunities for software engineering and work towards developing a domain specific language for the configuration and deployment of environmental models. We hold a series of interviews with environmental scientists involved in developing and deploying computer based environmental models about the approach taken in engineering models, and describe a case study in deploying an environmental model (WRF: Weather Research Forecasting) on a cloud architecture. From these studies we find a number of opportunities for A) software engineering methods and tools such as Domain Specific Languages to play a role in abstracting from underlying computing complexity, and for B) new architectures to increase efficiency and availability of deployment. Together, we propose they will allow scientists to concentrate on fundamental science rather than specifics of the underlying computing.
53 pags., 19 figs., 1 tab. ; Our understanding of the processes that control the burden and budget of tropospheric ozone has changed dramatically over the last 60 years. Models are the key tools used to understand these changes, and these underscore that there are many processes important in controlling the tropospheric ozone budget. In this critical review, we assess our evolving understanding of these processes, both physical and chemical. We review model simulations from the International Global Atmospheric Chemistry Atmospheric Chemistry and Climate Model Intercomparison Project and Chemistry Climate Modelling Initiative to assess the changes in the tropospheric ozone burden and its budget from 1850 to 2010. Analysis of these data indicates that there has been significant growth in the ozone burden from 1850 to 2000 (approximately 43 ± 9%) but smaller growth between 1960 and 2000 (approximately 16 ± 10%) and that the models simulate burdens of ozone well within recent satellite estimates. The Chemistry Climate Modelling Initiative model ozone budgets indicate that the net chemical production of ozone in the troposphere plateaued in the 1990s and has not changed since then inspite of increases in the burden. There has been a shift in net ozone production in the troposphere being greatest in the northern mid and high latitudes to the northern tropics, driven by the regional evolution of precursor emissions. An analysis of the evolution of tropospheric ozone through the 21st century, as simulated by Climate Model Intercomparison Project Phase 5 models, reveals a large source of uncertainty associated with models themselves (i.e., in the way that they simulate the chemical and physical processes that control tropospheric ozone). This structural uncertainty is greatest in the near term (two to three decades), but emissions scenarios dominate uncertainty in the longer term (2050¿2100) evolution of tropospheric ozone. This intrinsic model uncertainty prevents robust predictions of near-term changes in the tropospheric ozone burden, and we review how progress can be made to reduce this limitation. ; ATA and PTG would like to acknowledge support from National Centre for Atmospheric Science. YE would like to acknowledge support from the National Science Foundation Atmospheric and Geospace Sciences awards # 1900795 and 1929368. TW acknowledges support from the Hong Kong Research Grants Council (T24-504/17-N) and the National Natural Science Foundation of China (91844301). ASL thanks European Executive Agency under the European Union's Horizon 2020 Research Innovation programme (Project "ERC-2016-COG 726349 CLIMAHAL"). RH is supported by an NERC Independent Research Fellowship (NE/N014375/1). YMS is supported by an NERC PhD studentship. Part of this research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration.
