The Planet Fusion Surface Reflectance (PF-SR) product is a daily, 3 m, gap-free, analysis ready data product in four spectral bands (blue, green red, NIR). It is generated using a rigorous methodology to enhance, harmonize, inter-calibrate, and fuse optical data from PlanetScope, Landsat-8, Sentinel-2, MODIS, and VIIRS. An integral part of this methodology is gap-filling to ensure a spatially complete and temporally continuous product, irrespective of the actual acquisition coverage and cloud environment. This gap-filling subprocess is informed by cloud masked (but not yet gap-filled) PF-SR data from both before and, if available, after the prediction/gap-fill date. Both spatial and temporal interpolation techniques are used to provide an informed and weighted estimate of the gap-filled pixel values. For larger gaps or when data after the prediction date is limited, vegetation growth trajectories from past growing seasons for the given area are matched with current trajectories/phenologies and used to guide predictions. The uncertainty of the gap-filled pixel values is closely related to the temporal gap between actual observation data. In this work, we explore the fusion of Sentinel-1 SAR data with the PF-SR product during the gap-filling subprocess. The main goal of incorporating SAR data into this process is to decrease the temporal gap between actual observation data and improve/validate the uncertainty estimates of gap-filled pixel values. As SAR data is unhindered by cloud cover or smoke, exploiting the SAR data in the gap-filling process also serves as a means of capturing outlier events (e.g., early harvest, flooding, fires, etc.) during extended periods of obscured or missing optical data. We present our initial results for the fusion of Sentinel-1 SAR data with PF-SR data. With a focus on agriculture, we investigate the interoperability of Sentinel-1 with PF-SR, showing the synergies that exist between these different data sources. We discuss the importance of data quality when it comes to establishing phenological relationships and patterns between SAR and optical data. We then show that exploiting the temporal cadence of Sentinel-1 data and fusing it into the gap-filling process can improve the data quality and confidence estimates of PL-SR products. This improves the capability of PF-SR products to provide a more complete and accurate panorama of phenology in the agricultural domain.

Hydrological variability is widely recognized as a key ecological factor in river ecosystems (Jacobson and Jacobson, 2013). As a result, these ecosystems have been classified according to their hydrological characteristics, but these classifications exhibit biases for intermittent or ephemeral hydrological regimes. Ephemeral rivers, characteristic of arid and semi-arid areas, are numerous in the world but only few information about them is available. An ephemeral river is defined by an absence of water flow interrupted by flooding (Jacobson and Jacobson, 2013). The Kuiseb river is one of the twelve ephemeral rivers of Namibia, in South West of Africa. It originates in a semi-arid high plateau (~1,700 m, with 200-300 mm/year of rainfall), and crosses the hyper arid Namib desert (~10-50 mm/year of rainfall, Benito et al., 2010). The Kuiseb River is the border between the South composed of sand dunes and the North with gravel plains. The rainy season, from November to April, can be at the origin of flash floods, with an average duration of few hours to few days. Floods in ephemeral rivers are the major source of water for humans and ecosystems, it is therefore necessary to study them to better understand them (Grodek, 2007; Benito et al., 2011). They present very complex longitudinal, lateral, vertical and temporal gradients. Following a flood, water quickly infiltrates the substrate and then provides the necessary humidity for vegetation and recharging of aquifers. In addition, infiltrated water provides the necessary resources for humans and wildlife. Thus, understanding floods and associated losses is crucial for better management of water resources in arid zones. The study of water flow events, including magnitude, frequency and duration, are essential hydrological components (Jacobson and Jacobson, 2013). However, quantification of water loss and flood routing is limited because very few in situ stations are located in arid environments, and few flooding occurs. Space remote sensing offers the possibility of studying these extreme phenomena in arid environments, thanks to a spatial and temporal resolution adapted to these phenomena. This study aims at monitoring the postflood behavior of the Kuiseb River using Sentinel-1-2-3 missions. We focus on the exceptional flood of January 2021, with a total rainfall three times greater than the average over the last years, which followed the rainy month of December 2020. Such a rare event occurs every ten years, last ones happened in 2011 and in 2000. Sentinel-2 multispectral images were used to produce Normalized Difference Vegetation Index (NDVI) and Normalized Difference Moisture Index (NDMI) maps at a spatial resolution of 10 m. First results show a high spatial variability, with a rapid growth of the vegetation after the flood in the riverbed. Sentinel-3 radar altimetry data shows an increase of the backscattering coefficient due to to the flood, and last over time. Finally, Sentinel-1 data were used to study the radar backscattering amplitude over the study site and shows variations, specially in northern tributaries of the Kuiseb river. Finally, Sentinel-1 SAR phase data from interferograms will be used to track changes in soil moisture due to aquifer level dynamics, and then the riverbed drought over time.


References:
Benito, G., Rohde, R., Seely, M., Külls, C., Dahan, O., Enzel, Y., ... & Roberts, C, 2010. Management of alluvial aquifers in two southern African ephemeral rivers: implications for IWRM. Water Resources Management, 24(4), 641-667.
Jacobson, P. J., & Jacobson, K. M., 2013. Hydrologic controls of physical and ecological processes in Namib Desert ephemeral rivers: Implications for conservation and management. Journal of Arid Environments, 93, 80-93.
Benito, G., Thorndycraft, V.R., Rico, M.T., Sanchez-Moya, Y., Sopena, A., Botero, B.A., Machado, M.J., Davis, M., Perez-Gonzalez, A., 2011. Hydrological response of a dryland ephemeral river to southern African climatic variability during the last millennium. Quat. Res. 75, 471e482.
Grodek, T., Enzel, Y., Benito, G., Porat, N., Jacoby, Y., Dahan, O., ... & Seely, M., 2007. The contribution of paleoflood hydrology to flood recharge estimations along hyperarid channels, Kuiseb River, Namibia. In XVII INQUA congress, Cairns, Australia.

1. Introduction
Nowadays, the Copernicus Sentinel missions together provide a combination of high-resolution observation capabilities in different bands that few years ago were not yet available. If used in a coordinated way, their data can upgrade the scale and spatial and temporal resolution of investigations of small-sized volcanoes, whose eruptions can be violent and potentially hazardous for local communities.
With its peculiar type of activity and limited extent, the Stromboli volcano in southern Italy offers an ideal test case to demonstrate that the TROPOspheric Monitoring Instrument (TROPOMI) onboard the Copernicus Sentinel-5 Precursor (Sentinel-5P) satellite has the suitable spatial resolution and sensitivity to carry out local-scale sulfur dioxide (SO2) monitoring of relatively small-size, nearly point-wise volcanic sources and – in combination with Sentinel-2 multispectral observations of the situation on the ground – distinguish periods of different activity intensity.

2. Data and methods
The main input dataset of this study is the whole record of Sentinel-5P TROPOMI Level 2 SO2 geophysical products from UV sensor data acquired over Stromboli, since the beginning of the mission routine operations, after its 6-month-long commissioning phase. In particular, the TROPOMI time series for Stromboli starts on 6 May 2018, and the analyzed data cover three full years of observations, until 31 May 2021.
The latter were processed with purposely adapted Python scripts, according to a methodological workflow encompassing the extraction of total SO2 Vertical Column Density (VCD) at given coordinates (including conditional VCD for three different hypothetical peaks at 0–1, 7 and 15 km), as well as filtering by quality in compliance with the Sentinel-5P Validation Team’s recommendations. The total SO2 VCD time series for the main crater and across different averaging windows (3 x 3, 5 x 5 and 4 x 2) were compared to prove the correctness of the adopted spatial sampling criterion. As described in detail in Cofano et al. (2021), an approach for detecting SO2 VCD peaks at the volcano was trialed, and the detections were compared with the level of SO2 flux measured at ground-based instrumentation. To this scope, bulletin data were obtained from the FLux Automatic MEasurements (FLAME) and ROCcette site (ROC) Stations (Delle Donne et al., 2017; Randazzo et al., 2005). These records permitted to distinguish periods of intense activity from periods with low emissions and tremors, and were used to make a qualitative comparison between the quantity of SO2 emitted on the ground and that observed from satellite.
SO2 time series analysis was then complemented with information provided by contextual Sentinel-2 multispectral (in the visible, near and short-wave infrared) and Suomi NPP VIIRS observations. The aim was to correctly interpret SO2 total VCD peaks when they either (i) coincide with medium to very high SO2 emissions as measured in situ and known from volcanological observatory bulletins, or (ii) occur outside periods of significant emissions despite signs of activity visible in Sentinel-2 data. Finally, SO2 VCD peaks in the time series were further investigated through daily time lapses during the paroxysms in July–August 2019, major explosions in August 2020 and a more recent period of activity in May 2021. Hourly wind records from ECMWF Reanalysis v5 (ERA5) data and World Meteorological Organization (WMO) stations were used to identify local wind direction and SO2 plume drift during the time lapses.

3. Results and discussion
The SO2 time series after Quality Assurance and further filtering includes 589 observations, thus providing robust and sufficiently continuous records of volcano degassing at Stromboli.
Our spatial sampling tests suggest that in the case of volcanoes of “limited” areal extension, i.e., comparable to the pixel size of TROPOMI data, such as Stromboli, it is generally suitable to sample the single pixel and consider its value with respect to that found for larger averaging windows.
The time series of total SO2 column density for the hypothetical profile at the 0–1 km peak altitude was the profile used for reference for weak eruptions and degassing, as in those cases the bulk of the SO2 emitted by Stromboli is expected to be within the first kilometer of the atmosphere (i.e., ground-level plume). During major explosions and paroxysms, higher plume altitudes should be accounted for (e.g. ~9 km observed after the main explosion on 3 July 2019). On days of moderate to very high activity at the volcano, the data showed very clearly a series of significant total VCD at the main crater.
We also found that SO2 total VCD peaks as captured by TROPOMI data can serve as reliable proxies of volcanic activity and, as such, can be used to investigate other volcanoes when information about events is scarce or absent. However, it is not always trivial to associate SO2 total VCD peaks with moderate to very high SO2 emissions recorded by ground-sensor data, and reported in volcanological observatories bulletins. In this regard, the inspection of contextual Sentinel-2 multispectral observations in the visible, near and short-wave infrared (as well as Suomi NPP VIIRS data) provides an effective means to refine the interpretation of SO2 VCD peaks when they occur outside known periods of significant emissions. Although some limitations may occasionally constrain the analysis (i.e., cloud coverage that may hinder the visibility; time lag between the acquisition date of Sentinel-2 vs. TROPOMI data), this multi-sensor data approach adheres to the holistic concept of multi-band and multi-mission observations that are behind the whole Copernicus Earth Observation Programme.
Finally, the multi-temporal analysis of daily time lapses of SO2 VCD during the paroxysms that occurred in July–August 2019, major explosions in August 2020 and a more recent period of activity in May 2021 demonstrated that the proposed approach was successful in showing the SO2 degassing associated with these events, and warning whenever the total SO2 column density values at Stromboli may be overestimated due to clustering with the plume of the Mount Etna volcano. This geographical parameter, alongside wind direction, has to be accounted for whenever the nearly point-wise volcanic source under investigation is located in proximity to other SO2 emission sources, either natural (e.g., Mount Etna) or anthropogenic, that may interfere.

4. Conclusions
In the current context of an increasing body of literature on the use of Sentinel-5P data for volcanic studies, our presentation aims to deepen the discussion on the practical technical issues involved in the handling and post-processing of these geophysical data that are yet to become of common and standard use across the scientific community interested in volcanological applications, and how they can be used in a synergistic way with Sentinel-2 multispectral imagery to investigate eruptions, paroxysms and other events of potential concern for safety, at small-size Strombolian volcanoes. Based on the evidence gathered at Stromboli from 2018 to 2021, we propose practical recommendations for further implementation in similar volcanic environments. The proposed analysis approach is successful in showing the SO2 degassing associated with these events, and warning whenever the SO2 VCD at Stromboli may be overestimated due to clustering with the plume of the Mount Etna volcano.


References

Cofano, A., Cigna, F., Amato, L. S., de Cumis, M. S., & Tapete, D. (2021). Exploiting Sentinel-5P TROPOMI and Ground Sensor Data for the Detection of Volcanic SO2 Plumes and Activity in 2018–2021 at Stromboli, Italy. Sensors 2021, Vol. 21, Page 6991, 21(21), 6991. https://doi.org/10.3390/S21216991
Delle Donne, D., Tamburello, G., Aiuppa, A., Bitetto, M., Lacanna, G., D’Aleo, R., & Ripepe, M. (2017). Exploring the explosive-effusive transition using permanent ultraviolet cameras. Journal of Geophysical Research: Solid Earth, 122(6), 4377–4394. https://doi.org/10.1002/2017JB014027
Randazzo, D. A., Caltabiano, T., Salerno, G. G., Murè, F., Bruno, N., Longo, V., Spina, A. la, & Burton, M. R. (2005). Rapporto sullo sviluppo delle reti FLAME Etna e Stromboli, per la misura del flusso SO2, durante il periodo 2005 – 2009. https://www.earth-prints.org/handle/2122/5509

Air Quality (AQ) in Belgium is driven by complex processes covering a wide range of temporal and spatial scales, from point-like emissions of primary pollutants to intercontinental transport over an area of highly inhomogeneous land use. Therefore, the responsibility for AQ regulations is distributed among different levels of public authority, from international and federal to regional and local. Informed policymaking in terms of air quality regulation and sustainable energy consumption requires tailored monitoring of air quality and of the impact of past and future regulations taken.

AQ monitoring in Belgium has hitherto been relying mostly on in-situ measurements of surface concentration (mainly NOx and PM2.5), with geographical gaps between observations filled in with numerical modelling. However, today Belgium is mapped on a daily (or better) basis at unprecedented resolution by the new Copernicus Sentinel series of satellite sounders on sun-synchronous Low Earth Orbits (LEO), and soon also from a geostationary vantage point (GEO). Of particular interest are the observations of molecular atmospheric pollutants (NO2, CO, and CH4) by the Sentinel-5 Precursor (S5P, LEO) and soon by the Sentinel-4 (GEO) and Sentinel-5 (LEO) series, of particulate atmospheric matter (PM, observed as aerosol optical depth, AOD) by S5P and Sentinel-3, and of proxies for pollution sources such as fire radiative power (FRP) from Sentinel-3 and the land cover classification that can be obtained from Sentinel-2 (S2GLC).

We present here synergistic applications of these Sentinel data to monitor AQ in Belgium, developed in the framework of the Belgian federal research project LEGO-BEL-AQ (2020-2023, https://lego-bel-aq.aeronomie.be) funded by BELSPO. First, 1km-resolution (downscaled) maps of tropospheric NO2 (from the operational S5P and upcoming S4) are generated under various aggregation criteria. Their consistency but also particularities with respect to in-situ surface NOx measurements (i.e., the primary proxies for AQ) are assessed. To arrive at a more complete picture of Belgian AQ, in terms of atmospheric composition and attribution to potential sources, these results are cross-correlated with the AOD (particulate matter proxies), fire (FRP) and land cover (S2GLC) data sets. Case studies are elaborated on the capital (Brussels), key harbours (Antwerp and Ghent), industrial sites near Liège, and a wildfire (Groot Schietveld). As such, this work demonstrates the potential role of the Sentinel constellation in AQ monitoring and in supporting AQ policies, and its complementarity to the well-established surface monitoring programmes.

The 2017 launch of the Sentinel-5p mission marked the start of a new era of air quality earth observation using UV/VIS hyperspectral observations. The unprecedented spatial resolution of Sentinel-5p combined with its improved detectors and instrument calibration resulted in a single-satellite-pixel quality that allows for day-to-day monitoring several tropospheric trace gas and aerosol plumes.

Since then, many publications have explored this potential and have shown that Sentinel-5p indeed can be used to observe fine scale details of emission source. The main focus has been on anthropogenic emissions associated with air quality and methane in urbanized regions and affluence countries and the global impact of wide-spread massive wildfires that have occurred during the past few years.

The use of Sentinel-5p data for studying the aspects of much smaller and/or seasonal small scale localize fires and characterization of their emissions in relation to fire information derived from other Sentinel missions has not drawn much attention yet.

Here, we explore the capacity of Sentinel-5p for monitoring small scale localized fires in conjunction with fire information derived from the Sentinel-1, -2, and –3 missions. Such fires are often associated with agricultural activity as well as deforestation activities in subtropical and tropical regions. The emissions of these fires have an important contribution to the global carbon budget, but the magnitude of their emissions – and thus associated air quality – very much depends on fire characteristics such as fuel load, vegetation type, and soil moisture. How these fire characteristics and fuel characteristics interact and how emissions of trace gas and aerosols as well as the partitioning between emissions of various trace gases and aerosol types vary as a function of underlying fuel and soil characteristics is largely unexplored from an observational point of view.

We show that Sentinel-5p emissions can be used to build a large and data rich small scale localized fire-characteristics database that can be used for extensive and elaborate statistical analysis of their fire characteristics and dependencies. Understanding and quantifying these relations will further help bridge the gap between observations and model simulations. The latter often still are of insufficient spatial resolution to resolve fires at the spatial scales where they occur, nor are they capable of capturing the large variability of fire types and emissions and their dependency on fuel and soil characteristics.

Recent advances in satellite-based monitoring methods have enabled the detection and quantification of methane (CH4) point emissions. In particular, the combination of TROPOMI regional-scale data with high spatial resolution data from hyperspectral (e.g. PRISMA, high detection sensitivity with 30m pixel resolution) and multispectral (e.g. Sentinel-2, high temporal resolution and global coverage with 20m pixel resolution) optical missions creates an ideal framework to detect, quantify and monitor individual CH4 emissions of highly emitting point sources.
In this contribution, we will illustrate the use of satellite synergy to analyse the emissions from two important hot spots of industrial methane emissions. These two areas are the west coast of Turkmenistan and Algeria's O&G fields.
In Turkmenistan, we have found, during January 2017-November 2020, 29 individual CH4 sources that are directly linked to the O&G sector, of which 86% had been identified as inactive flares that vent gas. In the study period, the Sentinel-2 satellite has detected a total of 944 CH4 plumes with fluxes >1800kg/h, and the ZY1 and PRISMA hyperspectral satellites have quantified 25 plumes with emissions ranging from 1.400 ± 400 kg/h to 19.600 ± 8.100 kg/h. Moreover, the emission monitoring had shown a substantial increase in the number of methane plume detections in 2020 concerning previous years. Furthermore, according to VIIRS data, the use of flaring in the country has been decreasing significantly in recent years. This suggests a causal relationship between the decrease in flaring and the increase in venting. On the other hand, the historical Landsat record shows that gas vent emissions in this region have been occurring sporadically since at least the late 80' s.
Based on the Turkmenistan experience, we have extended the methodology to Algeria, a country with a large O&G extraction activity and favourable surface characteristics for satellite emission detection. To define the emitter search area, we have focused first on locations where TROPOMI detects (or has ever detected) super-emissions, and second on fields where, according to VIIRS data, flaring installations show anomalous or intermittent behaviour (flaring intensity peaks and troughs or periods of inactivity). Using the historical Landsat record, we have made a temporal tracking of the emission sources in order to better understand the problem behind Algeria's emissions.
These studies show that the synergy between the different satellites has been able to precisely reveal the origin of large anthropogenic CH4 emitters, which have the potential to be rapidly fixed, and monitored over time. They have also shown the outstanding and strong emissions from the O&G sector, which have remained unnoticed for decades.