Navid Mahdizadeh Gharakhanlou
Projet
Utilisation de modèles d'apprentissage automatique et d'une approche de modélisation à base d'agents pour étudier les processus géospatiaux et socio-écologiquesResearch problem statements: promoting sustainability in the era of population growth requires the ability to maintain and improve food production to meet the growing demands of an increasing global population. Pollination services are essential for crop production, resulting in increased yields and the availability of diverse and nourishing food resources. Through its role in enhancing crop yield and quality, pollination contributes to the long-term stability of food sources and fosters sustainability in agriculture. Precise and timely predictions of crop yields, incorporating pollination services, are essential to maintain a stable food source that can meet the needs of a growing global population. Nonetheless, predicting crop yields is challenging due to various factors including unpredictable environmental conditions, crop growth variations, data quality concerns, climate change uncertainty, technological advancements, and human decision-making. Although existing crop yield prediction models can provide reasonably accurate estimates of actual yields, there is an ongoing quest for enhanced performance in this regard. The significance of pollination in the yield of crops relying on pollinators cannot be overstated. Although certain staple crops rely on wind pollination, a significant portion of fruit crops depend on both commercial beekeeping and wild pollinators. Investigations reveal that a significant portion of annual crop yield increases is due to commercial pollinators, primarily honeybees. However, the decline in managed honeybee population has led to a shortage in pollination services, potentially causing profound adverse consequences on both ecological and economic welfare. Conversely, the increasing cultivation of pollinator-dependent crops has outpaced the expansion of beehives, leading to a shortage of pollination services. Concerns regarding the decrease in honeybee populations have triggered alarm within the agricultural community due to its potential threat to the stability and productivity of pollinator-dependent crops. Addressing this issue necessitates modeling approaches capable of considering multiple factors affecting beehive mortality to accurately evaluate and mitigate its significant repercussions. The pollination process, which relies on decisions made by both farmers demanding pollination services and beekeepers offering these services, constitutes a complex adaptive system. Within this system, farmers determine whether to request pollination services, and their decisions are influenced by a variety of factors, including crop selection, land management practices, economic considerations, and environmental variables. These factors give rise to non-linear outcomes and emergent properties. Moreover, this system operates within a context of temporal dynamics, with changes occurring in farming practices and environmental conditions, which further add to its complexity. Similarly, beekeepers engaged in pollination services interact with farmers and the environment, creating dynamic feedback loops and unforeseeable consequences. Their ability to adapt and learn introduces yet more complexity, resulting in emerging properties that stem from these interactions. Environmental factors, changing conditions over time, spatial variations, and uncertainties all contribute to the complexity of this system. Effectively addressing these complexities requires the application of advanced spatial modeling techniques capable of capturing the decisions made by both farmers and beekeepers and the combined impact of these decisions on crop yields and beehive mortality. The recent surge in artificial intelligence (AI) is fueled by advancements in data-driven techniques such as machine learning (ML) and deep learning (DL), alongside the availability of big data and increased computing power. Simultaneously, geographic information systems (GIS) have transitioned into a domain of big data science, with readily accessible geospatial data enabling detailed analysis. Geospatial artificial intelligence (GeoAI), the integration of GIS and AI, leverages and extends AI to revolutionize geospatial problem-solving. On the other hand, the complex systems (CS) modeling approach encompasses complex nonlinear interactions among self-organizing components at various organizational levels. These interactions generate novel patterns that require analysis using complexity theory. GeoAI and CS modeling approaches allow us to depict and investigate geospatial phenomena by capturing the complex spatial relationships and interactions between various components. Through the design of various scenarios and the prediction of future trends, these modeling approaches provide researchers and policymakers with the tools to investigate strategies for addressing the decline of pollinators and enhancing crop productivity. This supports the sustainability of agriculture and brings advantages to both pollinator populations and food security. Main research objectives: This research aims to employ, develop, and integrate GeoAI and CS modeling approaches to enhance the prediction accuracy of the spatial models and develop more realistic spatial models to capture the complexity of real-world geographical phenomena. More specifically, this research aims to achieve four main objectives: 1. Crop mapping using various ML and DL models applied to multi-temporal satellite images. 2. Utilizing two distinct modeling approaches, ML and statistical, to (i) evaluate and predict beehive mortality considering various environmental, ecological, and climatic conditions influencing honeybee colony losses, and (ii) assess the influence of these factors on beehive mortality. 3. Investigating the potential of various ensemble ML algorithms to predict crop yields by incorporating a variety of geospatial variables such as climate data, vegetation indices from satellite imagery, soil properties, and honeybee census data into the crop yield prediction process. 4. Developing CS models integrated with ML simulations of crop yield prediction and assessment of beehive mortality to investigate the impacts of farmers’ and beekeepers’ decisions regarding pollination services on crop yield and beehive mortality. Contribution to science: This research offers several primary contributions to science: 1. It sheds light on the performance of ML and DL models in crop mapping, highlighting their strengths and limitations. Additionally, it produces precise and up-to-date crop maps, offering valuable insights for modeling systems and guiding policymakers’ decisions. 2. This research through assessing beehive mortality offers insights into the complex relationships among environmental factors, ecology, climate, and beehive health, aiding in the conservation of honeybee populations vital for crop pollination and food security. Accordingly, it promotes sustainable agriculture, environmental conservation, and advancements in monitoring and beehive management. 3. The research contributes by providing crop yield prediction models, pinpointing the most influential factors affecting crop yields, and offering insights into the potential effects of various climate change scenarios on crop production. These findings can assist farmers in predicting crop yields and devising effective strategies to maximize agricultural productivity. 4. This research enhances comprehension regarding how various decisions made by farmers and beekeepers affect both beehive mortality and crop productivity across varying situations, without necessitating real-world experimentation, which often consumes time and resources. This research contributes to providing flexible decision-making frameworks applicable to agriculture and beekeeping. Contribution to SDGs: Bearing in mind the 17 Sustainable Development Goals (SDGs) adopted by the UN in the 2030 Agenda for Sustainable Development, this research will aim to contribute to the SDGs 1 (End poverty), 2 (End hunger, achieve food security and improved nutrition and promote sustainable agriculture), 3 (Ensure healthy lives and promote well-being), 8 (Promote sustained, inclusive and sustainable economic growth), 12 (Ensure sustainable consumption and production patterns), 13 (Take urgent action to combat climate change and its impacts).
Mots-clés
machine learning, Geographic information systems (GIS), deep learning, GeoAI, complex systems, Spatial modelingPublications
1- Flood susceptible prediction through the use of geospatial variables and machine learning methodsMahdizadeh Gharakhanlou, Navid, Liliana Perez
2023 Journal of Hydrology
2- Urban search and rescue (USAR) simulation in earthquake environments using queuing theory: estimating the appropriate number of rescue teams
Hooshangi, Navid, Navid Mahdizadeh Gharakhanlou, Seyyed Reza Ghaffari-Razin
2022 International Journal of Disaster Resilience in the Built Environment
3- Mapping Crop Types for Beekeepers Using Sentinel-2 Satellite Image Time Series: Five Essential Crops in the Pollination Services
Mahdizadeh Gharakhanlou, Navid, Liliana Perez, Nico Coallier
2024 Remote Sensing
4- From data to harvest: Leveraging ensemble machine learning for enhanced crop yield predictions across Canada amidst climate change
Mahdizadeh Gharakhanlou, Navid, Liliana Perez
2024 Science of The Total Environment
5- Evaluating environmental, weather, and management influences for sustainable beekeeping in California and Quebec: Enhancing beehive survival predictions
Mahdizadeh Gharakhanlou, Navid, Liliana Perez, Evan Henry
2025 Journal of Environmental Management
6- Evaluation of potential sites in Iran to localize solar farms using a GIS-based Fermatean Fuzzy TOPSIS
Hooshangi, Navid, Navid Mahdizadeh Gharakhanlou, Seyyed Reza Ghaffari Razin
2023 Journal of Cleaner Production
7- Geocomputational Approach to Simulate and Understand the Spatial Dynamics of COVID-19 Spread in the City of Montreal, QC, Canada
Mahdizadeh Gharakhanlou, Navid, Liliana Perez
2022 ISPRS International Journal of Geo-Information
8- Spatial Prediction of Current and Future Flood Susceptibility: Examining the Implications of Changing Climates on Flood Susceptibility Using Machine Learning Models
Mahdizadeh Gharakhanlou, Navid, Liliana Perez
2022 Entropy
9- Dynamic simulation of fire propagation in forests and rangelands using a GIS-based cellular automata model
Gharakhanlou, Navid Mahdizadeh, Navid Hooshangi
2021 International Journal of Wildland Fire
10- A Spatial Agent-Based Model to Assess the Spread of Malaria in Relation to Anti-Malaria Interventions in Southeast Iran
Gharakhanlou, Navid Mahdizadeh, Navid Hooshangi, Marco Helbich
2020 ISPRS International Journal of Geo-Information
11- Spatio-temporal simulation of the novel coronavirus (COVID-19) outbreak using the agent-based modeling approach (case study: Urmia, Iran)
Mahdizadeh Gharakhanlou, Navid, Navid Hooshangi
2020 Informatics in Medicine Unlocked
12- Developing an agent-based model for simulating the dynamic spread of Plasmodium vivax malaria: A case study of Sarbaz, Iran
Mahdizadeh Gharakhanlou, Navid, Mohammad Saadi Mesgari, Navid Hooshangi
2019 Ecological Informatics
