Graph Percolation as Decision Threshold for Risk Management in Cross-Country Thermal Soaring

John J. Bird · 2025

Read Graph Percolation as Decision Threshold for Risk Management in Cross-Country Thermal Soaring by John J. Bird (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-16945.txt.

Introduction

John Bird presents a 2025 paper on risk in glider flight, where pilots ride rising air to travel long distances. A sailplane with no engine must find the next thermal before its height runs out. Deciding when a climb is too weak to leave is the core gamble of cross-country soaring.

Bird frames the sky as a graph: thermals are nodes, and glides between them are edges. Graph percolation, the point where the network connects end to end, becomes the safety threshold. Flight logs from human pilots showed they almost never fly below that threshold. Pilots even accept weaker climbs to keep the network connected, revealing a preferred minimum link count. The theory turns gut feel into a measurable rule for instructors and flight computers.

Catalog record

CREATORJohn J. Bird
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-16945

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Graph Percolation as Decision Threshold for Risk Management in Cross-Country Thermal Soaring John J. Bird Abstract Long range flight by fixed-wing aircraft without propulsion systems can be accomplished by “soaring” – exploiting randomly located updrafts to gain altitude which is expended in gliding flight. As the location of updrafts is uncertain and cannot be determined except through in situ observation, aircraft exploiting this energy source are at risk of failing to find a subsequent updraft. Determining when an updraft must be exploited to continue flight is essential to managing risk and optimizing speed. Graph percolation offers a theoretical explanation for this risk, and a framework for evaluating it using information available to the operator of a soaring aircraft in flight. The utility of graph percolation as a risk measure is examined by analyzing flight logs from human soaring pilots. This analysis indicates that in sport soaring pilots rarely operate in a condition which does not satisfy graph percolation, identifies an apparent desired minimum node degree, and shows that pilots accept reduced climb rates in order to maintain percolation. I Introduction In thermal soaring an unpowered aircraft sustains flight by circling within updrafts called “thermals” where the vertical air motion, or “lift,” exceeds the aircraft’s unpowered descent rate. The aircraft thus climbs, gaining energy which is stored as altitude then expended in a glide to reach the next thermal or an objective point [ 1 ] as seen in figure 1 . Soaring is practiced by birds to provide economical locomotion, by humans seeking to maximize speed as a competitive sport [ 1 ] [ 2 ] , and by Uncrewed Aircraft Systems (UAS) to extend range and endurance [ 3 ] [ 4 ] [ 5 ] [ 6 ] . Classic thermal cross-country soaring theory focuses on optimizing the speed between lift sources of known strength, under the assumption that there is adequate altitude to reach the next thermal [ 1 ] . Under these assumptions, “speed-to-fly theory” can be derived which clearly defines the optimal speed in response to random air motion and expected thermal strength for a given aircraft. As the strength of expected lift increases a faster speed is optimal – the greater descent rate which occurs at higher speeds is compensated by more rapid climbs to regain energy. The expected next thermal strength is often referred to at the “MacCready” number after an instrument developed by Paul MacCready which provided an instantaneous indication of the optimal speed to fly [ 7 ] . The stochastic nature of thermal formation poses a challenge to this model however. Thermals are not uniformly spaced or of uniform strength, so a pilot may occasionally have to exploit weaker thermals in order to continue flight as failure to find a thermal results in an “outlanding” in which the aircraft must land in an improvised landing area (e.g. an agricultural field). In a competition an outlanding results in a poor score and can effectively place a pilot out of contention for top rankings. In the soaring community the interaction of sporting risk and speed is acknowledged in a few ways – pilots often speak of a “working band” of altitude. When the altitude is maintained within this band the mean speed is relatively fast [ 2 ] . Pilots may also speak of “shifting gears” when encountering difficult conditions or missing a climb, into a mindset where they will accept weaker climbs to “reconnect” with good conditions [ 8 ] . Efforts to formalize risk management have been made, most notably by Fukada [ 9 ] and Cochrane [ 10 ] who studied the impact that speed has on expected competition score when lift sources are not assured. These analyses show that as a pilot’s altitude decreases they should slow down and accept lower climb rates to hedge against not having enough energy to find the next thermal. This can be expressed in the MacCready setting that the pilot chooses to guide their speed to fly. Fig. 1: Cross-country soaring requires exploiting a sequence of updrafts to gain energy which is expended in gliding flight. Updrafts are stochastically distributed and cannot be reliably detected except in situ, putting an aircraft at risk of failing to find additional energy sources before reaching the ground. This use of the MacCready setting is intuitive to soaring pilots, but does not directly address risk in a theoretically rigorous way. An alternative is to model cross country soaring as a random graph traversal. A flight takes place between nodes (thermals) which are connected by edges if one node can be reached from another (the thermals are within gliding range). Because of the stochastic nature of thermal formation the locations of the nodes (and thus their interconnections) are random. The possibility of completing a sequence of waypoints (a “cross-country task”) can be stated in terms of the existence of a path which connects a sequence of nodes that traverse the task. While not directly answering this question, continuum percolation allows properties of paths on random graphs to be determined [ 11 ] , and indeed this has been applied to thermal soaring in the past [ 12 ] . Percolation suggests that once enough nodes are reachable from a given node that the probability of continued cross-country flight suddenly approaches 1 (though not necessarily along the desired task). Indeed, Depenbusch verified this behavior in Monte Carlo simulation and used percolation analysis to derive constraints on aircraft performance as a function of thermal spacing [ 12 ] . The present work takes percolation analysis a step further. By considering a slightly different percolation problem, operational (rather than design) bounds on soaring performance can be obtained. This manuscript briefly reviews continuum percolation as applied to cross-country soaring, suggests a new percolation condition, and applies a post hoc percolation analysis to flights completed by human pilots to assess the utility of percolation as a risk measure. The results suggest that percolation offers an effective means to assess soaring risk and a path to unify risk management and speed to fly. This has the potential to improve cross-country soaring practiced as a sport by human pilots, and also to provide a principled basis for autonomous decision making when soaring is used to defray energetic costs for UAS. II Graph Percolation and Soaring Continuum percolation is a property of random geometric graphs, which are composed of nodes or vertices which are randomly located and connected to each other if some condition is satisfied. Figure 2 illustrates a thermal graph \mathcal{G} : vertices \mathcal{V} are located at thermals whose locations are generated by a Poisson process, and connected with an edge \mathcal{E} if a thermal is within gliding range of the previous thermal. The graph is colored by traversable clusters (every thermal in a traversable cluster can be reached from at least one other thermal in the cluster). Note that in general edges do not need to be two-way connected, but for simplicity we will consider symmetric edges (equivalent to constant thermal height and no wind). Fig. 2: A collection of thermals which forms a random geometric graph whose connection degree is three-quarters of the critical threshold. The colored area surrounding each thermal

Read the complete text