**1. Introduction**

Persistently active volcanoes are an ever-present threat to populations and infrastructure exposed to primary and secondary volcanic hazards on their flanks. Continuous, ground-based monitoring by local or regional volcano observatories plays an essential role in hazard mitigation in such locations. When ground-based techniques are impractical or too hazardous, Earth-orbiting satellites can be a valuable tool for volcano monitoring. However, satellite remote sensing techniques can be limited by temporal and/or spatial resolution, particularly in rapidly evolving situations.

Satellite remote sensing (or Earth observation, EO) has played an increasingly important role in volcano surveillance over the past few decades, concurrent with technological advances in satellite sensors [1–5]. The principle applications of EO data in volcanology include volcanic gas and ash monitoring [6–8], monitoring of heat fluxes [9–13], optical measurements [14–16], mapping of ground deformation [17,18], geological mapping [19,20] and geological hazard assessment [21–23]. The main factors limiting the efficacy of EO data for volcano monitoring are the temporal and spatial resolution of the measurements, which are seldom optimized simultaneously in satellite instruments. Temporal resolution is arguably the most critical, since volcanic activity remains largely unpredictable and can produce dynamic phenomena that can be difficult to track and analyze in real-time during rapidly evolving crisis situations. Several volcanic eruptions in 2018 demonstrated the need for rapid response, including: Kilauea (Hawaii) in May 2018; Agung and Anak Krakatau (Indonesia) in November–December 2018; Mayon (Philippines); Piton de la Fournaise (Reuniòn Island); and Mount Etna (Italy) [24].

Commonly-used pushbroom visible-infrared satellite sensors, such as those deployed on the USGS Landsat or European Sentinel-2 satellites, have intermediate-high spatial (~10–20 m), but low temporal (days to weeks) resolution (Table 1). Since 2008, with the open access of Landsat archives, "time series" images have been used to analyze natural processes. High acquisition frequency allows for a more complete understanding of geological dynamics [25]. Here, we explore some volcanological applications of a new EO paradigm that can provide observations with both high spatial and temporal resolution: constellations of small satellites or 'CubeSats'. The use of a constellation of CubeSats provides a higher cadence than is possible with the use of a single satellite, whilst also providing high spatial resolution by minimizing swath width. One of the pioneers in this field is Planet Labs Inc. [26], which has operated an expanding constellation of over 100 CubeSats (informally called 'Doves') since 2014. The Planet constellation provides three or four band (visible to near-infrared) PlanetScope (PS) imagery of the entire land surface of the Earth with ~3 m spatial resolution and a temporal cadence of ~1–72 h (Table 1). Although the temporal resolution of current CubeSat constellations is lower than that provided by geostationary satellites, their spatial resolution is several orders of magnitude higher, with a pixel size of ~3 m compared with pixel sizes of 0.25–2 km for geostationary imagers. In some cases, consecutive data acquisition by Planet Doves in the same 'flock' (or overlapping flocks) can provide multiple images of the same region within a few minutes. These characteristics, along with an academic open data access policy, make PS data appealing for monitoring of volcanic activity or other dynamic geophysical phenomena resulting in surface change (e.g., landslides, earthquakes).


**Table 1.** Characteristics of optical satellite instruments currently used for volcano monitoring.

We use Fuego volcano (Guatemala; Figure 1) as the target of this initial study. Fuego is a persistently active volcano. It has produced several paroxysmal eruptive events in the last three years, providing a rich opportunity to analyze PS imagery. Paroxysmal eruptions at Fuego follow extended periods of lower-level activity, and typically begin with continuous lava fountains and lava flows, followed by tephra fallout and pyroclastic density currents (PDCs), which present the greatest threat to the surrounding areas. Due to its persistent activity and subsequent remobilization of deposits by rainfall, the morphology of Fuego and the barrancas (valleys) that radiate from its summit is in a constant state of flux, both in the near-vent region and on the lower flanks.

We focus here on activity at Fuego in February 2018. We assess the utility of optical PS data for monitoring activity within Fuego's active vent region and for mapping eruption deposits using the Normalized Difference Vegetation Index (NDVI) [27,28] and visual analysis. Of particular interest is whether PS data can assist with 'rapid response' mapping of eruption deposits that could be eroded or removed by rainfall soon after emplacement, since the volume and extent of such deposits can provide critical information on eruption magnitude and on the threat of secondary volcanic hazards, such as lahars. Monitoring and mapping of such ephemeral deposits is also important for probabilistic mapping of volcanic hazards around Fuego.

**Figure 1.** Location of the Fuego-Acatenango Complex. (**a**) Location of Fuego volcano in the Central American Volcanic Arc. (**b**) Detail of the main barrancas channels surrounding the Fuego edifice.
