Article 2/3
Once the foundations have been laid, the central question remains: what could quantum computing actually be used for in GIS?
Rather than speculating in abstract terms, this article explores the domains where the challenges faced by GIS overlap with the types of problems for which quantum computing is being investigated.Spatial networks, graph analysis, multi-criteria optimization, remote sensing, and territorial simulation: this second part examines, case by case, the potential applications of quantum computing in geomatics, clearly distinguishing between credible perspectives and what still belongs, for now, to exploratory research.
Spatial Network Optimization
Networks as a cornerstone of GIS analysis
Spatial networks are among the most foundational application domains in GIS. Road networks, public transport systems, water, electricity, and telecommunication networks are central to many analyses: routing, accessibility, flow management, and resilience to disruptions.
In their simplest form, these analyses rely on well-known algorithms (shortest path, minimum spanning tree). However, as soon as additional constraints are introduced — capacities, variable costs, temporal dynamics, multiple priorities — the nature of the problem fundamentally changes.
When the shortest path is no longer enough
In operational GIS, the questions being asked are rarely as simple as “what is the shortest path?”. They are more likely to involve:
- finding optimal routes for entire vehicle fleets,
- simultaneously integrating distance, time, cost, and CO₂ emissions,
- accounting for regulatory constraints, time windows, or limited capacities,
- optimizing networks subject to disruptions or future scenarios.
These problems often correspond to complex variants of network optimization (routing, flow, coverage problems), which are known for their high combinatorial complexity. The number of possible solutions increases very rapidly with network size.
Limits of current classical approaches
Modern GIS rely on several strategies to cope with this complexity:
- specialized heuristics,
- genetic or evolutionary algorithms,
- decomposition into sub-problems,
- parallel or distributed computing.
While effective and robust, these approaches have well-known limitations:
- results depend strongly on parameter choices,
- difficulty in guaranteeing overall solution quality,
- computation times sometimes incompatible with near–real-time use,
- partial exploration of the solution space.
Potential contributions of quantum computing to network problems
Quantum computing — particularly optimization-oriented approaches — offers a different way of tackling these problems. The goal is not to explicitly compute every possible route, but to formulate the network and its constraints as a global problem, where each possible configuration corresponds to a system state.
Within this framework, quantum methods could:
- explore a large number of configurations simultaneously,
- converge toward low global “cost” solutions,
- rapidly produce high-quality approximate solutions for complex networks.
For GIS, this opens interesting perspectives for:
- multi-criteria optimization on large networks,
- territorial-scale logistics planning,
- disruption and resilience scenario analysis (construction works, incidents, natural hazards).
A complementary tool, not a replacement
It is essential to stress that these approaches are not meant to replace the classical algorithms embedded in GIS. Their potential value lies instead in hybrid processing chains, where:
- the GIS prepares and structures the network,
- quantum computing addresses a critical optimization sub-problem,
- the GIS analyzes, visualizes, and interprets the results.
In this perspective, quantum computing becomes a specialized accelerator, used only when problem complexity exceeds the reasonable limits of classical computation.
Geographic Graph Analysis
Graphs at the heart of spatial analysis
In GIS, many phenomena can be modeled as graphs: transportation networks, hydrological systems, ecological connectivity, functional relationships between territories, or socio-spatial interactions. In these graphs, nodes represent spatial entities, while edges encode relationships, often weighted and directed.
Graph analysis goes beyond cartographic representation to explore deeper spatial structures: network organization, hierarchies, dependencies, vulnerabilities, and diffusion capacities.
Increasingly expensive analyses at scale
GIS tools already provide many indicators derived from graph theory: centrality, connectivity, accessibility, proximity or fragmentation indices. These measures are well mastered for networks of moderate size.
Difficulties arise, however, when:
- graphs become massive (millions of nodes or edges),
- relationships are multi-level or time-evolving,
- multiple graphs must be analyzed or compared simultaneously,
- analysis criteria are numerous and sometimes contradictory.
In such cases, computation times explode and classical methods reach their limits, even with substantial computing infrastructures.
Community detection and emerging spatial structures
One key area of graph analysis is community detection, i.e. identifying groups of nodes that are strongly connected to each other. In geography, these communities may correspond to:
- mobility basins,
- functional areas,
- ecological sub-networks,
- urban or socio-economic clusters.
Detecting such structures relies on complex optimization problems, often formulated as the maximization or minimization of global graph functions. These formulations naturally lend themselves to quantum-inspired approaches capable of efficiently exploring large configuration spaces.
Centrality, vulnerability, and network resilience
Centrality analysis helps identify critical nodes or edges in a network: congestion points, bottlenecks, or strategic elements for territorial resilience. When multiple indicators are combined (flows, capacity, redundancy, disruption scenarios), the analysis quickly becomes multidimensional.
In a quantum context, these problems can be approached as global optimization or extreme-configuration search tasks, potentially enabling:
- broader exploration of scenarios,
- better integration of interactions between criteria,
- relevant results even when the solution space is very large.
Comparing and evolving spatial graphs
GIS increasingly focus on network evolution over time: urban growth, mobility changes, infrastructure transformations. Comparing successive graphs or identifying recurring patterns in evolving graphs is computationally demanding.
Quantum approaches could eventually facilitate:
- rapid comparison of complex structures,
- detection of significant changes,
- exploration of possible evolutionary trajectories.
A still exploratory perspective
As with network optimization, quantum applications to geographic graph analysis remain largely experimental. Potential gains depend heavily on the ability to correctly translate GIS problems into formulations compatible with quantum models.
Nevertheless, the conceptual alignment between graph analysis, global optimization, and quantum computing makes this domain one of the most promising for future experimentation in geomatics.
Spatial Modeling and Multi-Criteria Optimization
Multi-criteria optimization: ubiquitous in GIS
Spatial modeling very often involves decision-making under constraints. In territorial planning, environmental management, or risk assessment, GIS are used to answer questions such as: where? how many? according to which criteria?
Public facility location, infrastructure siting, regulatory zoning, site prioritization, or spatial resource allocation almost always rely on multi-criteria optimization. Each decision must integrate sometimes conflicting dimensions: economic costs, accessibility, environmental impacts, territorial equity, and social acceptability.
From layer overlay to combinatorial explosion
GIS have long offered multi-criteria analysis tools: weighted overlays, aggregated scores, spatial masks. These methods are effective for exploratory analyses or relatively simple decisions.
However, when the problem involves:
- selecting multiple sites simultaneously,
- interactions between sites (competition, complementarity),
- global constraints (budget, territorial coverage),
- multiple scenarios,
the overlay logic reaches its limits. The problem becomes combinatorial: each possible combination of sites represents a distinct solution, and their number grows exponentially with territory size and the number of criteria.
Limits of classical spatial optimization approaches
To address this complexity, GIS rely on:
- greedy algorithms,
- heuristic or metaheuristic methods,
- iterative or evolutionary approaches.
These methods produce satisfactory solutions but raise several issues:
- sensitivity to parameters and weightings,
- difficulty in guaranteeing global optimality,
- partial exploration of the solution space,
- high computational costs for realistic problems.
In public decision-making contexts, these limitations can affect robustness and transparency.
Potential contributions of quantum computing
Quantum computing — particularly optimization-oriented approaches — offers a different way of addressing these problems. The goal is not to explicitly test all possible combinations, but to represent the problem as a whole, where each spatial configuration corresponds to a possible system state.
Within this framework, quantum approaches could:
- explore many configurations simultaneously,
- better integrate complex global constraints,
- rapidly provide multiple near-optimal solutions,
- facilitate the analysis of trade-offs between criteria.
For GIS, this opens the door to models that no longer seek a single “best” solution, but rather a set of relevant solutions, helping to inform decision-making instead of freezing it.
Toward enriched decision support
A particularly interesting aspect is the potential of quantum computing for scenario exploration. Rather than recalculating an optimization for each hypothesis (weights, constraints, objectives), quantum methods could facilitate sensitivity analysis of solutions.
In a GIS processing chain, quantum computing would then act as an exploration engine:
- the GIS structures the data and constraints,
- quantum computing explores the solution space,
- the GIS presents the results in cartographic and analytical form.
Still an experimental perspective
As with networks and graphs, these applications remain at an exploratory stage. Translating a real-world spatial problem into a formulation compatible with quantum computing remains a major conceptual and methodological challenge.
Nevertheless, multi-criteria spatial optimization is one of the domains where the potential gains are most evident, as it concentrates precisely the difficulties — combinatorics, global constraints, trade-offs — that quantum computing is designed to address.
Remote Sensing and Geospatial Image Processing
Exploding data volumes and image complexity
Remote sensing is now one of the main drivers of GIS growth. Multispectral and hyperspectral satellite imagery, Earth observation time series, LiDAR and radar data provide a fine-grained, continuous view of territories.
This richness comes with major challenges:
- very large data volumes,
- high dimensionality (spectral bands, dates, resolutions),
- heterogeneous sensors and acquisition conditions,
- the need for rapid processing in near-operational contexts.
Processing these data relies on increasingly complex algorithms, often at the limits of intensive computing.
Classification, segmentation, and pattern recognition
Key remote sensing tasks include:
- supervised and unsupervised classification,
- image segmentation,
- change detection,
- recognition of spatial structures or objects.
These tasks involve exploring very large feature spaces, where each pixel or object is described by many variables. Even with powerful machine learning methods, computation time and resource requirements can become significant, especially for large-scale or multi-temporal analyses.
Potential contributions of quantum machine learning
One active research direction concerns quantum machine learning, which aims to exploit qubit properties to improve certain stages of machine learning. In remote sensing, such approaches could eventually:
- accelerate parts of the classification process,
- improve class separation in high-dimensional spaces,
- facilitate detection of complex or rare patterns,
- reduce training data requirements for certain problems.
For GIS, the goal would not be to replace existing processing chains, but to enhance their capabilities for difficult cases: spectrally similar classes, heterogeneous environments, noisy or incomplete data.
Time series analysis and change detection
The growing number of Earth observation satellites now enables fine-grained temporal analyses: land-use dynamics, vegetation monitoring, rapid urban or environmental changes.
These analyses require comparing large numbers of images over time, using computationally expensive methods. Quantum computing could offer perspectives for:
- efficiently comparing spectro-temporal signatures,
- identifying significant transitions or breakpoints,
- rapidly exploring multiple evolution scenarios.
Between theoretical potential and operational reality
Caution remains necessary. Quantum applications in remote sensing are still largely theoretical or limited to demonstrators on small datasets. Current processing chains based on high-performance computing and GPUs remain far more efficient and mature.
However, remote sensing represents a privileged experimentation field for quantum computing, as it combines precisely the key ingredients: massive data, high dimensionality, pattern recognition, and uncertainty.
In the medium to long term, hybrid architectures — combining classical preprocessing with specialized quantum modules — could enrich GIS analytical capabilities in this domain.
The potential applications presented in this second part outline a broad and stimulating landscape: networks, graphs, multi-criteria optimization, remote sensing, and spatial simulation. Above all, they show that some recurring GIS challenges — combinatorics, global constraints, scenario exploration — naturally resonate with quantum computing paradigms.
These perspectives must nevertheless be confronted with the reality of current technologies. The third and final part of this series adopts a more critical and factual stance, returning to existing experiments, observed limitations, and realistic horizons for integrating quantum computing into GIS practice.
