AIS-Informed and Land-Aware Eco-Routing for Maritime Decision Support Using Machine Learning and Metocean Data

Geospatial sensing and data analytics can support maritime risk and safety management by transforming vessel-movement and environmental observations into transparent evidence for route assessment. Within this context, reducing maritime fuel consumption requires routing methods that exploit historical vessel-movement data while representing geographical and metocean conditions explicitly. This study contributes an AIS-informed, land-aware eco-routing framework that integrates geospatial data processing, machine-learning-based fuel estimation, and weather-screening analysis to generate minimum-predicted-fuel routes and examine their exposure to environmental conditions. Minimum-predicted-fuel routing is formulated as a shortest-path problem and solved using Dijkstra's algorithm. The search operates on an undirected graph constructed from historical AIS-derived segments and augmented with short stitching links screened using a global land mask. Edge costs are estimated by a Random Forest fuel-consumption model using vessel speed and draft together with current wave height and wind speed. The framework is evaluated through a paired comparison with a historical AIS observation sequence and a controlled nine-combination factorial weather-screening experiment on two routes using frozen execution-time weather snapshots. In the paired comparison, the generated route had 25.54% lower model-estimated fuel consumption and 11.65% shorter reconstructed travel time, but 9.31% greater distance and higher distance-weighted mean wave height and wind speed. This baseline difference coincided with higher assigned generated-route speeds and was sensitive to the fuel-model training-speed domain, so it is treated as an illustrative result. Weather screening either disconnected the graph or left the fuel-minimising path unchanged. It also left every feasible path on the shorter route unchanged. These findings illustrate how geospatial sensing and data analytics can expose interactions among fuel efficiency, metocean exposure, graph connectivity, and missing-weather assumptions, while supporting transparent maritime risk and safety decision-making. By combining AIS sensor observations, weather data, and historical vessel paths, the proposed approach identifies fuel-optimal candidate routes within a geographically and environmentally screened network and provides evidence for safety-oriented route assessment.