5:15 PM - 6:45 PM
[ACG45-P04] Efficient Meteorological Environment Observation through UAV Route Optimization: Quantification of Importance and Algorithm Design
Keywords:Aircraft observation, Unmanned aerial vehicle (UAV), Route optimization, Weather forecast
The use of unmanned aerial vehicles (UAVs) for meteorological and atmospheric environmental observations offers a viable means to enhance the accuracy of weather and atmospheric environment predictions. For instance, accurate measurement of lower atmospheric water vapor levels is crucial for predicting phenomena such as linear precipitation zones. The dynamic observations enabled by UAVs can acquire impactful observational data for the improvement of forecast accuracy, harboring potential for meteorological forecast enhancement. Efficient observations necessitate the quantification of the importance of meteorological environmental phenomena and, based on this data, the optimization of aircraft routes to prioritize areas with significant meteorological environmental information. Concurrently, UAV-based meteorological environmental observations must consider constraints related to aircraft performance, safety aspects, and legal rules, necessitating efficient observations under these constraints.
This research undertakes the optimization considering the trade-offs between the maximization of observed meteorological environmental information and operational constraints or objectives, leading to the development of a new route selection algorithm. The aim is to evaluate the feasibility and practicality of the algorithm developed in this study by optimizing two objectives: meteorological environmental information quantity and flight distance.
For the quantification of meteorological information, we adopted a sparse measurement location optimization method to quantify the importance of meteorological environmental information. This approach addresses the problem of reconstructing high-dimensional physical quantities, such as the spatiotemporal fields of the meteorological environment, from limited observational data (in this case, measurements along UAV routes). To extract significant meteorological environmental structures from limited observations, we employ proper orthogonal demcomposition to derive characteristic modes, considering the degree to which these significant meteorological environmental structures can be extracted from limited observations as a measure of information quantity in meteorological environmental observations. The dataset used for applying eigen decomposition and calculating characteristic modes in this study is sourced from the Japan Meteorological Agency's Global Ensemble Prediction System (GEPS).
By implementing multi-objective optimization, we elucidate the trade-off between meteorological information quantity and analyze the characteristics of the resultant routes. From this trade-off relationship, we can identify routes with the highest information quantity (red point), shortest flight distance (green point), and those that represent a balance in the trade-off relationship (black, purple, and orange points). The characteristics of aircraft routes at each of these representative points are shown in Fig.1. It was confirmed that flight routes maximizing meteorological information quantity significantly vary with the available flight distance. Especially in cases of limited flight time, determining which locations to observe becomes critical. Given that many current UAVs have limited flight times, identifying and targeting meaningful locations hold considerable value.
This research undertakes the optimization considering the trade-offs between the maximization of observed meteorological environmental information and operational constraints or objectives, leading to the development of a new route selection algorithm. The aim is to evaluate the feasibility and practicality of the algorithm developed in this study by optimizing two objectives: meteorological environmental information quantity and flight distance.
For the quantification of meteorological information, we adopted a sparse measurement location optimization method to quantify the importance of meteorological environmental information. This approach addresses the problem of reconstructing high-dimensional physical quantities, such as the spatiotemporal fields of the meteorological environment, from limited observational data (in this case, measurements along UAV routes). To extract significant meteorological environmental structures from limited observations, we employ proper orthogonal demcomposition to derive characteristic modes, considering the degree to which these significant meteorological environmental structures can be extracted from limited observations as a measure of information quantity in meteorological environmental observations. The dataset used for applying eigen decomposition and calculating characteristic modes in this study is sourced from the Japan Meteorological Agency's Global Ensemble Prediction System (GEPS).
By implementing multi-objective optimization, we elucidate the trade-off between meteorological information quantity and analyze the characteristics of the resultant routes. From this trade-off relationship, we can identify routes with the highest information quantity (red point), shortest flight distance (green point), and those that represent a balance in the trade-off relationship (black, purple, and orange points). The characteristics of aircraft routes at each of these representative points are shown in Fig.1. It was confirmed that flight routes maximizing meteorological information quantity significantly vary with the available flight distance. Especially in cases of limited flight time, determining which locations to observe becomes critical. Given that many current UAVs have limited flight times, identifying and targeting meaningful locations hold considerable value.