From patches to indicators: theory-driven metrics for drone-based monitoring of coastal marine environments

Nearshore environmental hazards often manifest as transient, spatially structured patches rather than uniform fields. Storms, runoff pulses, fronts, upwelling, and surface aggregation processes create features that can appear, drift, merge, fragment, and even disappear over hours to days. This patch-dynamics framing is well established in seascape ecology, yet it is rarely translated into operational, repeatable indicators of environmental status because routine observations typically lack the combined spatial detail and temporal frequency needed to resolve patch evolution. Here, we develop a general, drone-based approach that links patch theory to indicator design for three widely relevant coastal targets: (i) surface algal blooms expressed as surface accumulations or water-colour anomalies, (ii) plastic macro-litter in coastal areas, and (iii) beach wrack. We propose a theory-based indicator set for each target based on shared patch descriptors (extent, patch-size distribution, edge-to-area ratio, fragmentation/aggregation, spatial clustering, and persistence/turnover) derived from a harmonized data-collection for high-frequency aerial surveys and a common processing workflow that yields georeferenced mosaics and target-specific maps suitable for patch extraction and analysis. These indicators separate different properties of the potentially hazardous patches: how much (coverage and intensity) from how it is organized (patch structure), and how it changes (event dynamics). For blooms, indicators quantify the areal fraction and intensity of surface/color anomalies, along with patch coherence, boundary complexity, and short-term persistence. For plastic macro-litter, indicators combine abundance proxies with aggregation metrics that distinguish dispersed background from hotspot-driven risk. For wrack, indicators capture shoreline loading and continuity, emphasizing alongshore patch length scales and retention/clearance dynamics. Across targets, we provide an uncertainty-aware formulation that flags limits of applicability, supports local calibration where needed, and enables consistent comparison across systems. The result is a general drone-to-indicator framework that turns high-resolution, repeatable observations into interpretable, theory-grounded metrics for environmental assessment, modeling, and ecosystem-based management. Introduction Nearshore coastal zones are subjected to rapid, uneven change. Many management-relevant phenomena emerge as short-lived events, with possible forming, shifting, and dissipating over hours to days \cite{Levin_1992}. Moreover, their spatial structure is typically patchy rather than smoothly varying \cite{Wedding_2011}, which is important because impacts and responses are often triggered by where material accumulates (and how much), not by single-point quantities. Some of these patches correspond to acute environmental hazards and require timely assessment and decision-making \cite{Benveniste_2019}. Yet routine monitoring often struggles to capture the fine spatial scales and rapid turnover that are characteristic of the nearshore environment, especially when observations must be repeatable and comparable across time \cite{Bierman_2011}. Event-driven patch dynamics as a common framing A useful way to unify these challenges is to treat nearshore events as patch dynamics of material accumulation, because their potential impacts are driven by exposure, which depends on the intensity and duration of the accumulation and whether it overlaps with sensitive uses or assets \cite{Thrush_2021}. Storms, fronts, wind-driven surface convergence, runoff pulses, and shoreline transport can rapidly create, move, merge, and break apart these patches \cite{Cushman_2002}, resulting in dynamic exposure hotspots that demand timely, spatially targeted management actions \cite{McGarigal_2009,Wedding_2011}. Patch theory - as used across landscape and seascape ecology - emphasizes descriptors such as patch-size distributions, edge-to-area relationships, spatial clustering/autocorrelation, and persistence/turnover as system-level manifestations of heterogeneous patterns \cite{Gr_nbaum_2012}. The practical appeal is that these descriptors can serve as integrated representations of complex spatial fields, i.e., exactly what ecological indicators must do if they are to provide a basis for assessment and support decision-making \cite{Franco_2025}.This approach is relevant for a wide range of coastal and marine issues where impacts are driven by localized concentrations and sharp spatial contrasts \cite{Bierman_2011}. Examples include oil slicks and surface sheens that fragment and drift, harmful algal blooms (HABs) and floating macroalgal surface mats, turbidity and sediment plumes from runoff or dredging that expand and contract with currents, coastal erosion, hypoxic zones that appear episodically and persist as bounded water masses, and marine litter and beach wrack accumulations \cite{Benveniste_2019}. Similar patch structures occur in thermal-stress footprints during marine heatwaves, contamination hotspots, and spatial mosaics of coral bleaching, where coastal dynamics frequently produce strong small-scale variability \cite{Wedding_2011}. Translating this framing to indicators requires observations that resolve patch coverage, edges/contact, and (when possible) change through time \cite{Forman_1995}. In the nearshore zone, satellite remote sensing is often too constrained by pixel size and revisit timing, and it is further challenged by mixed land–water pixels and rapidly changing optical conditions, which can blur small patches and bias edge-based metrics \cite{Benveniste_2019}. This motivates the use of drones as the primary instrument to support patch-statistic indicators in coastal zones managed by regional authorities. Drones: an observing system for patch statistics Drones provide a practical way to observe small- to medium-scale patch dynamics driven by regional events \cite{Chapapr_a_2022,Karahan_2025,Andriolo_2022,Escobar_S_nchez_2022,Pan_2021,Izar_2025,Sabaliauskait__2024,Cohen_2023}. Compared with many traditional field approaches, drone surveys can be deployed quickly, repeated frequently, and produce spatially continuous, georeferenced imagery that supports direct mapping of extent and structure \cite{Anderson_2013}. Environmental monitoring benefits from methods that can capture both short-term variability and longer-term patterns \cite{Levin_1992}, ideally using standardized schemes to enable comparison across surveys and sites \cite{Karahan_2025}. Drone-based observation can complement other monitoring components by filling the 'in-between' space: the spatial detail and temporal responsiveness needed for event-scale assessments. At the same time, drone data are only as useful as the definitions, processing rules, and quality assurance (QA) that turn imagery into reproducible products. The central methodological requirement is therefore not simply high resolution, but a workflow that produces (i) consistent maps, (ii) uncertainty-aware summaries, and (iii) indicators that remain meaningful when the scene changes (illumination, water optics, shoreline complexity) and when targets differ in appearance \cite{Mathews_2023}. Three targets, one observational problemThis paper focuses on three nearshore targets that frequently trigger monitoring and management actions: Algal bloom surface manifestations, including surface accumulations and visible changes in water color, can range from dense surface expressions to more diffuse discolorations, and their optical appearance can differ substantially across bloom types and conditions. Remote sensing work on HABs highlights that bloom observation can be hampered by coarse resolution and by the nearshore environment itself \cite{Martinez_Vicente_2020}, where phytoplankton distributions can vary strongly over short time scales and display small-scale patchiness associated with fronts and sub-mesoscale features \cite{Gernez_2023}. Plastic macrolitter in coastal areas is often concentrated along shorelines or in accumulation zones after transport events. Deposition is typically event-driven (e.g., storms, high-water episodes, shifting currents and winds), producing alongshore hotspots separated by cleaner stretches and rapid changes between surveys. From an observational standpoint using drones \cite{Escobar_S_nchez_2022,Andriolo_2022} , the key challenge is not only estimating total load, but mapping where litter concentrates at actionable scales (e.g., per shoreline segment) and how hotspots emerge, persist, and shift. Beach wrack, i.e., deposited organic material (often macroalgae/seagrass mixtures) forming strandline features that can expand, fragment, and shift rapidly with winds and waves. Beach wrack can be a nuisance because when large amounts wash ashore over a short period (often linked to eutrophication), it accumulates along the strandline, then degrades and can produce strong odours, unattractive mats, and practical problems for beach access, recreation, and clean-up \cite{Hyndes_2022}. Wrack can occur as continuous bands or as patchy deposits, with geometry that changes quickly as material is redistributed, buried, or removed \cite{Pan_2021}. Monitoring, therefore, benefits from spatially explicit metrics that capture not only cover or shoreline contact, but also patch structure (band continuity vs fragmentation) and persistence, because these characteristics strongly influence nuisance levels, access constraints, and collection needs \cite{Izar_2025,Sabaliauskait__2024}. We treat these targets together because, from a measurement perspective, they represent the same class of problem: event-driven, spatially patchy features that require rapid, repeatable, site-scale mapping and analysis of the patch overlap with areas of human activities that are sensitive to exposure to these targets (e.g., aquaculture farms and recreational areas). This commonality motivates a unified indicator logic grounded in patch theory, rather than three unrelated sets of ad hoc metrics. In addition, shoreline monitoring for plastics and beach wrack can often be integrated within the same drone surveys and mapping workflow, and wrack patterns may even provide useful contextual information for interpreting (and, in some cases, proxying) plastic accumulation in the strandline zone \cite{Izar_2025,Sabaliauskait__2024}.