Considerations when Developing HPC Weather and Climate Models for Scientists in Developing Countries

Great strides have been made in the advancement of weather and climate models for use in both forecasting as well as research and development, particularly in the ability to do regional climate modeling. This allows for scientists across the globe to forecast systematic changes in their region’s weather due to changing climate. This is especially useful for countries and regions that are highly impacted by climate change because high performance computer (HPC) models can help shape long-range planning in climate policy and decision-making. However, certain regions of the globe that are highly impacted by climate change may also be classified as developing countries, thus have limited access to computing resources and cannot fully utilize the benefits of most HPC models. Model developers can be considerate of these needs if they are aware of the financial and computational restraints some regions face when studying the impacts climate change for their region.

Developing countries are at the crossroads of juggling advances in technology, changing climate, economic development, environmental concerns, and the health and well-fair of its people. The countries struggle to keep up with global advances that can and should guide their countries policies and practices.  Additionally, developing countries often have limited access to technology capable of running HPC models (Mesquita et al. 2011). The array of types of Internet access (modem, ISDN, T1 line, G1-5), the type and generation of computer used, and location of access points pose significant challenges for individuals in these developing countries. Access to multiple processing computers and supercomputers is extremely rare and costly, which makes running any HPC model not feasible for many scientists.

It is easy to take for granted the idea that those who have an interest and need for HPC models have access to them. However, scientists in these developing countries who want to study advanced weather models attempt to find ways to access this technology through: computing resources (including the Internet) at work, if relatively wealthy they may have a personal computer with or without Internet access, or use a smartphone which may provide the most affordable computing power and Internet access.

Additionally, those who do have access to HPC model products have limited experience and knowledge about the model set up, parameters, and limitations. Thus, the quality of the conclusions and decisions drawn from these products may be diminished; which is a phenomenon that we call the “Warner effect” (Walton et al. 2016).

This presentation will provide results from a survey of participants (in both developed and developing countries from around the globe) about their daily access to computing resources. Participants come from two open-access, free online courses at M2lab.org, which utilizes an educational version of the WRF model (e-WRF; Mesquita 2013) for training novice weather modelers and for use in small-scale pilot studies (Walton et al. 2016). The presentation will include reports of participants computing technology, where they access this technology, who owns this technology, and the speed of their access. Implications for the overall findings will be discussed. The e-wrf model will also be discussed as an example of one approach to helping those with limited computing and Internet resources access state-of-the-art weather and climate models.

References:

Mesquita, M. D. (2013). WRF for Educational Purposes. Available at m2lab.org.

Mesquita, M. D., Veldore, V., Yarker, M. B., & Lamadrid, A. (2011). Long-Term E-Capacity Building (LEAD): A New Approach for Climate Science Research. Conference Of the Parties (COP) Publication. New Dehli: TERI Press.

Walton, P. J., Yarker, M. B., Mesquita, M. d., & Otto, F. (2016). Helping to make sense of regional climate modeling: Professional development for scientists and decision makers anytime, anywhere. Bulletin of the American Meteorological Society. DOI:10.1175/BAMS-D-14-00111.1