Enhancing Forecast of Adverse Weather Conditions through Advanced Modelling in Malawi

Solomon Dindi

Funding Area: Capacity building / Construcción de capacidad

Problem Statement: Accurate and comprehensive weather prediction are necessary to provide early warnings for Malawi which faces recurrent volatile weather conditions. Recent disasters occurrence includes cyclone Freddy in 2023, Cyclone Anna in 2022, and Cyclone Idai in 2019. The Tropical Cyclone Freddy that induced floods and mud slides in 15 districts in Southern region affecting and displacing 2,267,458 people, loss of over 679 people, 2178 injured persons and devastating damage to infrastructure including houses, roads, household property, schools, health facilities. Timely and accurate weather prediction is constrained by limited use of advanced model and high-resolution data by researchers and the Department of Climate Change and Meteorological services that requires high computing capacity. Malawi has the opportunity to optimize weather forecasting and climate variability as a High-Performance Computing (HPC) Facility was made available in 2022 in the country managed by the Malawi Research and Education Network (MAREN). The HPC is a huge open facility that would be harnessed and used by researchers and forecasters. The enablers for improved weather prediction in Malawi includes capacity for the researchers and forecasters to use the HPC with open-source models integrated with other physical models needed to develop a better understanding and record of wind, humidity, barometric pressure, precipitation and other weather measurements at a higher-resolution.

Proposed Activities: The project will serve to enhance the weather modelling using High-Performance Computing (HPC). The project activities will include: (i) Undertaking 3 – days training workshop which will convene researchers and model developers from University of Malawi and Malawi University of Science and Technology and participation of the forecaster and experts from the Department of Climate Change and Meteorological Services. The training workshop will be preceded by virtual course on computer languages (GOLANG/Python) for developing the model into the HPC. The workshop will cover computing processes including Performance, Portability and Productivity requirements of Weather and Climate models (3P’s); practice parallel programming implemented in weather and climate codes such as distributed and shared memory systems and coupling software in climate and weather applications. The workshop will be facilitated by MAREN together with HPC Experts and Climate Scientist from South Africa. (ii) Collaborations and works for the researchers and departments of climate change and Meteorological Services. The project for a period of 8 months after the training will work closely on ongoing application component using Malawi weather datasets and run simulations using advanced weather modelling software including operational Numerical Weather Prediction (NWP) model/system at high horizontal resolution. (iii) Concurrently, over the duration of the project, monitoring and support visits will be conducted by MAREN to the Universities and Malawi Department of Climate Change and Meteorological Services. The activity will include process of case studies and model evaluations to analyze past weather events and assess model performance and produce briefs and papers. (iv) The interactive and information dissemination for the project will include MAREN support to fostering collaborative platforms particularly mailing lists, wikis and Frequently Asked Questions (FAQs) which will be live and openly accessible. MAREN will also document the progress of the project and share articles and reports. It is expected that the project will contribute to building expertise and fostering collaboration in weather modelling through open access to High Performance Computer and open-source models benefiting both the research community and government agencies responsible for climate and meteorological services.

Openness: The project will exhibit openness through (a) Open Infrastructure – The project will ensure the use of the open-source, publicly accessible high-performance computing resources managed by MAREN in Malawi which will allow the researchers and government departments to access the same infrastructure for their work even after the project concludes and the use of open tools on the HPC including GOLANG and Python languages for modeling and the HPC already uses Linux (CentOS). The project will support utilization of open-source weather modelling software for easy access by the participants and also to encourages them to contribute back to the software's development and improvement (b) Engaging broader community – The project team will publicize the workshops academic institutions, meteorological associations, and online platforms including webinar and livestreams over the duration of the project for making the content accessible to a larger audience (c) Open sharing of project outputs – The project will ensure open sharing of materials including presentations over dedicated websites and Invest in Open Infrastructure (IOI) Platforms to allow for other interested individuals to benefit from knowledge sharing including making openly available public datasets, model outputs, the data and results to allow others to validate and build upon findings. The WiKi and FAQs will ensure that further information is accessible Openly.

Challenges: Carrying out the proposed work on enhancing capacity on weather modelling using high-performance may encounter challenges which includes: diversity of background of the targeted researchers and the representatives from the Department of Climate Change and Meteorological Services particularly on the level of understanding of computer programming and weather models. The project will ensure that content is tailor-made to suit the needs of all the participants. The project will integrate a bridging training for the basics of computing, programmes and HPC architecture prior to the modelling training. Secondly, the need for continuous follow-up and support to ascertain the use of the HPC for weather modelling which has been placed as central support for the duration of the project with availability of dedicated team. Lastly, technical challenges relating to hardware failures, software compatibility issues causing interruptions in the training and the processes. Through the project, MAREN Networks and Infrastructure experts will be available to provide technical backstops on the use of the HPC. MAREN will also leverage on the support from Centre for High Performance Computing of South Africa for additional support and facilities.

Neglectedness: There is limited support and funding allocation towards development and accelerating use of High-Performance Computing in Malawi including the unique application of HPC in weather modelling. More generally, HPC facilities elsewhere are owned and managed by commercial entities such as AMAZON and those managed by the National Research and Education Networks. The Stem-Trek provides support to individual scholars for technical knowledge around operations of the HPC. This application to the IOI is MAREN first attempt to solicit resources that would significantly help to ensure that the available HPC that MAREN manages is put into essential use in addressing major challenges affecting Malawi such as Adverse Weather and Climate Conditions. It opens up an opportunity to exploring further use of the HPC as well as catalyzing collaboration and research.

Success: MAREN will ensure that the project includes assessments and evaluations on the progress of the implementation of the project. MAREN expects to derive positive results from the project which will be demonstrated by key performance indicators and set targets which will form monitoring and evaluation of the project. Some of the measures include: (i) completion of bridging training on Linux, Introduction to Programming and Basics of GOLANG/Python (ii) a training workshop on weather modeling with HPC (iii) creation of a Special Interest Group (SIG) on weather and climate modeling (iv) creation of a WIKI and FAQ related to weather/climate modeling with HPC (v) Increase in the number of users of the HPC from 0 to at least 30 users over the project lifespan

Total Budget: US$24957

Affiliations: The proposal is affiliated with the Malawi Research and Education Network (MAREN), the National Research and Education Network (NREN) for Malawi and which is recognized as such by the Government of Malawi.

LMIE Carveout: The proposed Project fits LMIE CarveOut considering the working location for MAREN, which is Malawi where the project will be implemented. Most LMIE currently do not own HPC facilities and Malawi is privileged to have the facility. The users of the facilities under the project will be predominantly researchers from Malawi and government of Malawi through the Department of Climate Change and Meteorological Services.

Team Skills: MAREN project team that will support the implementation of the project includes experts with diverse skills in the areas of high performance computing, data analysis, research, network infrastructure, project management, finance and administration. (a) The team will be led by Solomon Dindi - He has over 17 years’ experience coordinating and deploying infrastructure and systems. He was instrumental in the acquisition of the HPC for use by Malawi’s researchers. Now his interest is to ensure the facility is used as per the intention. (b) Alex Chipalamwazani, Network Infrastructure Manger has over 7 years’ experience working on infrastructure both from user perspective and manufacturer perspective. He has been key in the redeployment of the HPC and the design of the cooling system for the facility at its current location. (c) Mwiza Mhango is an engineer with over 7 years’ experience. He plays a pivotal role in HPC infrastructure support from the day of deployment to-date. Has acquired a lot of skills in administration of HPC through trainings attended in the US in 2022. (d) Grace Dzoole is a seasoned finance professional with over 20 years’ experience in both private and public sectors, including audit. An MBA holder and a Chartered Global Management Accountant. She has previously managed the financial resources of over USD 2 million revenue firm. (e) Bryan Johnston - key member of the Centre for High Performance Computing (CHPC - South Africa) coordinating the trainings.

How Did You Hear About This Call: Word of mouth (e.g. conversations and emails from IOI staff, friends, colleagues, etc.) / Boca a boca (por ejemplo, conversaciones y correos electrónicos del personal del IOI, amigos, colegas, etc.)

Submission Number: 115