ABSTRACT:
Climate variability and change remains one of the most significant challenges for Africa in the 21st century. It threatens to undo the development gains made over the last two decades and could potentially reduce the region’s ability to attain the Sustainable Development Goals by 2030. Action is required to bolster the region’s capacity to urgently address the debilitating effects climate variability and change has on agricultural production, water resources and energy as well as the propagation of conflicts. Deployment of artificial intelligence (AI) in other regions of the world, especially in the Global North, has proven critical because of its ability to increase adaptation and mitigation capacity and improve the precision of evidence-based decision making.
This project will implement and manage an AI and climate change innovation research network. This network will support development and scaling of responsible AI innovations for climate action in sub-Saharan Africa, capacity building of African innovators and researchers, and the increased contribution of African research to international AI policy and practice through its integration into a global network on machine learning and AI. The hub will fund and support 10 climate-change innovation research projects as well as 10 master’s students focused on the intersection of machine learning and climate change.
This project is part of the four-year Artificial Intelligence for Development Africa (AI4D Africa) partnership between the Swedish International Development Cooperation Agency and IDRC to support the policy, innovations and expanded leadership that will spur responsible AI development in Africa. This network is one of five innovation research networks that will be funded through the program. The other networks are involved in work on education, gender and inclusion, agriculture and food security, and sexual and reproductive health.
LOCATION
Ghana, West African Science Service Centre on Climate Change and Adapted Land Use
TOTAL FUNDING
CA$ 1,158,100
PRINCIPAL INVESTIGATOR:
- Kone Daouda
SUBPROJECTS
| Project 1 — Leveraging AI for near real-time cattle counting and Farming system indexing using UAV Videos and images for estimation of GHG emissions (LAIRG) | |
| Principal Investigator: Assoc. Prof. Eng. Isa Kabenge, Makerere University | Country: Uganda |
| Project Summary: This research project will use remotely acquired UAV images and video to develop artificially intelligent algorithms that detect and classify cattle management systems and quantify cattle green house gas emissions in near real-time within Mubende District in the Central Region of Uganada. technologies that strengthen resilience and adaptive capacity to climate change. This work will improve the reliability of National Climate Change Department reports and, in so doing, help to support evidence-based decision making strategies for climate change mitigation in the lifestock sector and will empower knowledge and skills shared among stakeholders and farmers. | |
| Project 2 — Use of Smart Technology to Predict Climate Change Effects on Terrestrial Plants Diversity and Conservation for Sustainable Livelihood in North-Rift Region, Kenya | |
| Principal Investigator: Dr. John Wanjala Makokha, Kibabii University | Country: Kenya |
| Project Summary: This study aims to utilizing SMART technology to analyze climatic trends and develop predictive models for climate change effects, particularly focusing on terrestrial plant diversity and conservation in the North Rift region of Kenya. It encompasses eight counties, targeting a multi-stakeholder approach to assess the impacts of climate change on both the environment and the livelihoods of pastoral communities. Key objectives include analyzing climatic patterns, evaluating the effects of climate change on plant diversity, assessing land use practices on vegetation cover, determining the influence of soil variables on vegetation and species diversity, and ultimately developing mitigation measures for sustainable livelihoods. The project’s outputs are expected to include the establishment of county-specific climate change laws, policies, and climate change units, furthering Kenya’s commitment to climate action and contributing to sustainable development goals, vision 2030, and increased GDP. | |
| Project 3 — Modeling land productivity and crop yields under changing climate and land use management using Artificial intelligence in Lake Kyoga basin, Uganda | |
| Principal Investigator: Dr. Catherine Mulinde Kafeero, Makerere University | Country: Uganda |
| Project Summary: This project seeks to contribute to improved understanding of climate change adaptation and crop production through assessment of land productivity and crop production of different farm-based land management under changing climate in Kyoga basin. It will also build Makerere University student capacity in crop productivity and climate change adaptation assessment using artificial intelligence. The study spans three agroecological zones and employs extensive field surveys, geospatial analysis, and artificial intelligence techniques. Two Master of Science students will be trained in crop productivity and climate change adaptation, and the project will yield valuable outputs, including research theses, journal articles, a machine learning-based model for crop yield prediction, and a policy brief. The primary objective is to enhance understanding of climate change adaptation and crop production, ultimately contributing to sustainable agriculture in the Kyoga basin. | |
| Project 4 — Assessing impact of climate change or variability on the emergence of new plant diseases and development of innovative AI mobile application for control | |
| Principal Investigator: Dr. Yinka-Banjo, Chika Ogochukwu, University of Lagos | Country: Nigeria |
| Project Summary: The study focuses on the escalating problem of Climate Variability-induced Emergence of Infectious Diseases (EID) affecting cultivated plants, particularly Taro crops in West Africa, threatened by Taro Leaf Blight (TLB) caused by Phytophthora colocasiae. This project aims to assess the impact of climate change on new plant diseases like Phytophthora colocasiae and develop an AI mobile application to evaluate biological control through fungal endophytes. Object detection and classification algorithms build on the the Tensorflow and Keras libraries to train a pre-trained Single Shot Detector (SSD) Mobilenet model machine learning library to identify early-stage TLB identification. The resulting AI model will be deployed on Android smartphones for smallholder farmers to detect TLB early, reducing outbreaks and associated costs. The study also intends to evaluate the impact of climate change on Phytophthora colocasiae and establish fungal endophytes for TLB control, fostering farmer adoption of this technology. | |
| Project 5 — Towards sustainable carbon-neutrality and climate-resilient green growth development: Using a 4-tier Approach, IoT, AI, Mobile App and Drone Techniques for Early-Prediction and Control of Deforestation, Climate Change and their Notable Effects in Tanzania. | |
| Principal Investigator: Dr. Ing. Judith Leo, The Nelson Mandela African Institution of Science and Engineering (NM-AIST) | Country: Tanzania |
| Project Summary: This project seeks to address the complex environmental challenges arising from rapid population growth, deforestation, and climate variability in Tanzania and across Sub-Saharan Africa. With a focus on responsible AI, inclusivity, and climate resilience, it aims to combat deforestation, reduce climate change’s adverse effects, and promote sustainable development. The project’s holistic approach integrates AI, IoT, mobile apps, and drone technology, encompassing data collection, analysis, prediction, and policy development. It also emphasizes gender inclusivity, capacity building, and regional collaboration, fostering a community of experts and leaders. Research topics encompass greenhouse gas emissions, climate hazards, climate impact assessment, responsible AI, climate change, biodiversity, climate-smart agriculture, and smart cities development. The project’s outcomes aim to bridge the gap between research and practical application, ultimately enhancing sustainable carbon neutrality and climate-resilient economic growth and development in Africa. | |
| Project 6 — RAGA: An Artificial Intelligence Based System for Predicting Groundwater Availability | |
| Principal Investigator: Dr Cyril Dziedzorm Boateng, Kwame Nkrumah University of Science and Technology (KNUST) Prof D. D. Wemegah, Kwame Nkrumah University of Science and Technology (KNUST), Dr Marian Osei, Kwame Nkrumah University of Science and Technology (KNUST) | Country: Ghana |
| Project Summary: This project aims to develop an innovative web-based artificial intelligence (AI) driven open-source framework to predict groundwater availability in Ghana using Groundwater Levels (GWLs) as a proxy. We will call this project RAGA: Rapid Assessment of Groundwater Availability. This project will shift the paradigm of groundwater monitoring from a static process to a dynamic process to allow for the adaptation of resilient water management systems in response to climate change and variability. This study intends to achieve its aim by building a database of spatio-temporal hydrological, geological and physiographical, climate and groundwater level (GWL) variables for Ghana; developing AI algorithms and workflows for integration of varied data sources and prediction of groundwater availability and developing an open-source web-based application for rapid groundwater availability assessment to be used by stakeholders and the general populace. | |
| Project 7 — Bridging critical gaps in relative humidity data to enhance climate science and services in Ethiopia: The case of Awash River Basin | |
| Principal Investigator: Dr. Mekonnen Adnew Degefu, Debre Markos University Dr. Abebe Belay Adege, Debre Makos University | Country: Ethiopia |
| Project Summary: The objective of the proposed research project is to enhance climate services by filling critical gap in relative humidity data for the National Meteorological Agency of Ethiopia. The plan is to produce representative high quality gridded relative humidity data by merging better quality satellite and reanalysis RH data of the Awash River Basin from global open sources with quality control in-situ records. We plan to first evaluate the quality of global data products, then produce new datasets by merging the relatively better RH data with quality control in-situ records. We propose to move research into-use by applying and testing the new data product for early drought detection ability and for human and livestock comfortability indices. We plan on aiding implementation through our existing multidiscipline expert networks (climate scientist, data expert, and expert for artificial intelligence) and partnership with National Meteorological Agency of Ethiopia. | |
| Project 8 — Artificial Intelligence and Machine Learning for Modelling Climate Change, Landscape Dynamics, and Improving Uptake of Renewable Energy Technologies for Environmental Care in the Congo Basin | |
| Principal Investigator: Prof. Engr. Derek Ajesam Asoh, National Higher Polytechnic Institute (NAHPI), The University of Bamenda | Country: Cameroon |
| Project Summary: This study seeks to use AI and ML techniques to model climate change, and landscape dynamics; predict renewable energy potentials, and deploy related technologies in the Congo Basin under current and future climate change conditions. The main methods to be used include spatial modelling with artificial neural networks/deep neural networks, random forest supervised classification, cellular automata (CA) land cover/use change simulations, trend analysis and dynamics with TIMESAT, and field setup and operationalization of a solar power plant in rural Cameroon. The main outputs of the project include spatial trends in landscape dynamics – climate change nexus, land cover/use changes, projections and deforestation trends, daily/seasonal dynamics of surface water resources (e.g., dams and reservoirs), and impacts on hydropower provision, long-term trends in solar radiation and simulated suitable solar power sites, and an operational solar power plant in rural Cameroon. The project will have far-reaching impacts on the decision-making processes pertaining to energy provision in the Congo Basin as it will provide community-driven clean energy models that provide smart climate solutions at the local level. Capacity building and training of the local community stakeholders with an emphasis on gender equity will be the adopted strategy to ensure proper acceptance, use, maintenance, and management of the energy farm for environmental care. | |
| Project 9 — Modeling Grid Electricity Demand using Artificial Intelligence | |
| Principal Investigator: Prof Olusanya Elisa OLUBUSOYE; University of Ibadan | Country: Nigeria |
| Project Summary: This project proposes a machine learning solution to profiling electricity demand in African countries as a means to mitigate climate change. This study will adopt a mixed approach employing simulation and predictive algorithms on existing data which is to be statistically compared with real-time data. Furthermore, parameters to be adopted in the study entails Installed Capacity, Actual Generated Capacity, GDP Per Capita in PPP Annual Peak Load, Annual Base Load, Annual Electricity Consumption, Temperature and other Country Specific Variables includes (e.g. frequency of Grid Collapse) and foreign exchange rate. These parameters have been gathered over the past 10 years and the model would allow for input from real-time data as the study is hinged on a time- series approach. Modelling and simulations would be done with the use of supervised machine learning processes. Well-trained models would inform the true and optimal nature of grid demand and supply required for Africa’s energy transition. This would encourage energy optimization as related to both decarbonization and load cost energy system modelling. | |
| Project 10 — Climate change and land use land cover dynamics impacts on hydropower generation and consequences on electricity supply in West Africa using Ensemble machine learning | |
| Principal Investigator: Dr. Salomon Obahoundje, Université Félix Houphouët Boigny | Country: Côte d’Ivoire |
| Project Summary: This project aims to contribute to sustainably manage and plan the hydropower generation in West Africa under climate uncertainties and LULC dynamics using machine learning. This project will be focused on four dams (pilot sites) in Togo (Nangbeto), Côte d’Ivoire (Taabo), Burkina Faso (Bagré) and Senegal (Manantali), located in different climates and will aim to build communities of practices with HPG stakeholders (dam managers, local policy makers, representative of civil society, association of women, etc) to address the challenges, synergies and trade- off in the climate land energy water nexus in WA for a sustainable management and planning of HPG. During the implementation, the project will foster sharing and common learning between the 4 case studies. | |
| Project 11 — Development of an artificial intelligence assisted framework for assessing the vulnerability to climate change of mangrove ecosystems in West Africa: application on Benin coastal mangroves | |
| Principal Investigator: Adandé Belarmain FANDOHAN, Université Nationale d’Agriculture | Country: République du Bénin |
| Project Summary: This study aims to develop an artificial intelligence-assisted dynamic framework for assessing the vulnerability to climate change of West Africa’s mangrove ecosystems. Specifically, it aims to determine meaningful indicators for the vulnerability assessment of mangrove ecosystems of West Africa; to build an integrated model for mangroves’ vulnerability assessment; and to assess the vulnerability of Benin mangrove ecosystems using the built model. Thus, supervised and unsupervised learning algorithms of machine learning methods will be combined to select meaningful parameters to develop a new dynamic framework for the vulnerability assessment of mangrove ecosystems in West Africa. Parameters will be selected considering eco-surplus and deficit, degree of hydrological alteration, hydro- period, remote sensing, and regression coupled TSI. Particle Swarm Optimization (PSO), Support Vector Machine (SVM), Artificial Neural Network (ANN) techniques, Radial Basis Function (RBF), M5P and bagging algorithms will be integrated to model the vulnerability state of mangroves; and the Receiver Operating Characteristics (ROC) method and the sensitivity analysis will be used for models’ validation. Results will be disseminated to stakeholders through awareness-raising programs, workshops, and training. | |