EUNADICS early warning system dedicated to support aviation in case of crisis from natural airborne hazard and radionuclide cloud

The purpose of the EUNADICS prototype Early Warning System (EWS) is to proceed the combined use of harmonised data products from satellite, ground-based and in situ instruments to produce alerts of airborne hazard (volcanic, dust, smoke and radionuclide clouds), satisfying the requirement of ATM stakeholders 45 (www.eunadics.eu). The alert products developed by EUNADICS EWS (i.e. NRT observations, email notifications and NetCDF Alert data Products, called NCAP) have shown shows the significant interest in using https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c © Author(s) 2021. CC BY 4.0 License.

atmospheric transport modelling and by the VAACs, as described by Lechner et al. (2018). These Centres, as well as NMSs, are users of our system, which assures that all national and international downstream users continue getting their products through established and tested channels.
The EUNADICS consortium consists of 21 participating organisations from 12 different countries. It includes National Meteorological Services, Monitoring data providers, Operational Volcanologists, Small and Medium-170 sized Entreprises, a University institute, an Air Navigation Service provider and a Military Organisation.

Role of this work in EUNADICS
This study focusses on EUNADICS Early Warning System (EWS). This bloc/activity has a central role in 190 EUNADICS system as it acts as a trigger for the data integration of tailored observations in dispersion model forecasts, which can provide critical information in the resilience process for ATM decision-makers facing a crisis due to airborne hazard. As shown by the processing chain in Fig. 1, the EWS depends on inputs of observations, carried out by the data harvesting facility that acts as a primordial phase of the hosting platform of EUNADICS EWS. After a presentation of the requirements (i.e. users requirements, observational data inventories, review of 195 the input for the data integration and of the external reports), this study presents the concept, mechanism of EUNADICS EWS, with details about data products from satellite, ground-based and in situ platforms used to produce alerts. The pre-alerting mechanism and data provision is then described (with service description, performance verification, and cases studies). Finally, conclusion and future developments are presented.

Requirements for EUNADICS Early Warning System 200
After an overview of requirement reports (from users, data inventories, data integration and other sources) which cover the activity blocs addressed by EUNADICS, a summary of the requirements tackled by our EWS is presented.  A substantial report of user requirements has been established by EUNADICS partners. The main parties consulted was the VAACs, EUROCONTROL, ICAO, WMO, EASA, the airline companies, governmental institutions and primarily, the pilots and passengers. Table 1 presents a highlight of the three types of requirements identified from this report, i.e. Quality of information, System interoperability and Improvement of display.

Review of observational data inventories 210
Reviews of data inventories of satellite, ground-based and in situ products have been used to determine the most relevant products to be implemented in our EWS, according to user requirements presented in the previous section.
We identified two types of products, i.e. existing NRT (or proven NRT) products and tailored products in development. Key products, allowing NRT visualisation of natural airborne hazard and the implementation of them in EUNADICS EWS, have been investigated, showing added value to current existing public system (in the 215 late 2010s). Note that the observation of nuclide cloud is more sensitive and inaccessible to classic users as no data are public. After our review and the possibility with partners, a selection of observations has been determined with respect to the monitoring of European air space facing a crisis related to airborne hazard. Tables 1 and 2 show respectively the inventory of the selected products from satellite, in situ and ground-based instrumentation, to be considered by EUNADICS EWS. 220 For the implementation in EUNADICS EWS, we consider that each selected product can be characterised by four categories of information: the basic information (quantity, instrument/platform, responsible institute/provider, 225 units), the time-space resolution (temporal resolution, spatial/vertical resolution), the data availability information (spatial coverage for satellite and in situ instruments, temporal coverage for in situ instruments, Overpass time at the equator for satellite instruments, time delay for delivery, measurements schedule for ground-based instruments, processing level for satellite, data format, data volume, access, link to product overview, https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c Author(s) 2021. CC BY 4.0 License. dissemination/link to data), and the implementation information for data integration and decision-making 230 (alert/notification, visualisation system). Tables 2 and 3 present a subset of all the characterisation established for these products.

Requirements for data integration and harmonisation 235
Information gathered during the EUNADICS stakeholders workshop (Cologne, October 2017) and discussions with the VAACs at the AeroMetSci conference (Toulouse, November 2017), has brought a precise definition of the automated products needed from models that can be relevant for flight planning and management. A review of the inputs on data integration and harmonisation required by atmospheric transport and dispersion model, has been established to determine the most relevant product information/format/parameters to be implemented in our 240 EWS. The list of information/format/parameters is the following: satellite and ground-based Aerosol Optical Depth -AOD (from different platforms), satellite and ground-based lidar attenuated backscatter, ground-based measurements of particle matter, volcanic ash plume height (from radars, observatories), volcanic ash total column and plume-height from satellite, aircraft measurements of particle matter, lidar aerosol layer altitude, groundbased measurements of SO2, satellite SO2 column with average kernel and error estimate, aircraft measurements 245 of SO2, satellite SO2 plume height, SO2 profiles by ground-based spectrometers (observatories), nuclide concentrations from air sampler networks, and external dose rates from radiation monitoring networks. This list has been considered to obtain the inventories of observations in Tables 2 and 3.

Review of external reports
We proceeded a review of external reports (Zehner et al., 2010;ESA VAST user requirements, 2013;WMO 250 SCOPE Pilot Project Criteria, 2017;Inter Pilot Magazine, Issue 1, 2018) to determine the key development of the EWS and the most relevant product that would fulfil the requirement of the users. The same three types of requirements as in section 4.2 have been identified and are presented in Tab  -the improvement of the quality of information by optimising the risk assessment (using reliable and fast NRT observations, providing with level of confidence and errors bars if available, and implementing 260 tailored products), by the use of quality labels to increase the reliability of information, by providing point of contact and communicating about the data product version and availability with respect to data integration (points 1.a.i., 1.a.ii. and 1.b in Tab. 1, and all points of 1. in Tab. 4) -a contribution to the system interoperability by bundling the maximum of available information about selective detection of airborne hazard, by implementing compatible and homogenised information with 265 a global coverage, and by providing restricted access for key users to avoid misunderstanding of public users (points 2.c and 2.d in Tab. 1 and all points of 1. in Tab. 4).
-the improvement of the display of information by the visualising observations of hazard dispersion (point 3b in Tab. 1, and points 3.b and 3.e in Tab. 4).
Note that a documentation (with product characteristics, format, source of origin, algorithm, validation and 270 limitation of the selected data products by our EUNADICS EWS), has been provided by partners and the EUNADICS consortium (requirements 2.a and 2.e of Tab. 4). equipped with ground-based instruments for providing as much as possible fast and precise information concerning active volcanoes. Volcanic activity is monitored by observatories using a combination of fixed 280 instruments (e.g. seismometers, infrasound arrays, continuous GNSS stations, strainmeters, gas detectors, river monitors, radars) in addition to mobile instruments (e.g. mobile radars and lidars). Such instrumentation can stream live data from the field to the office, bringing critical information to the control cells. During episodes of volcanic unrest or eruption, observatories provide information to Civil Protection authorities, local populations and notifications to the VAACs. This procedure was established by ICAO, for providing crucial information 285 during volcanic eruptions to the aviation sector, and both IMO and INGV are using it. Volcanic Observatories provide notification of eruptive activity using the Volcano Observatory Notices for Aviation (VONA) messages that are issued according to the ICAO Doc.9766-AN/968 "Handbook on the International Airways Volcano Watch (IAVW)" (ICAO, 2012(ICAO, , 2014a(ICAO, , 2014bLechner et al., 2018). The VONA messages are aimed at dispatchers, pilots, and air-traffic controllers to inform them about volcanic unrest and eruptive activity that could produce 290 ash-cloud hazards. As an example, for Etna volcano, the VONA messages are sent by the Control Room of INGV-OE, which operates on a 24/7 basis, and they can be downloaded, together with other bulletins, reports, tremor graphs, images from video surveillance network, volcanic ash dispersal, etc. For the monitoring of Icelandic volcanoes, a network of UV spectrometer is used. This means that, for an event for which an increase of SO2 is deemed to be related to magma movement, an alert is issued by the VONA. The same is applied to the seismicity; 295 i.e. if there is a significant increase in seismicity (intense seismic swarms) or seismic tremors, this is outlined in the VONA and relayed through EWS. About regional monitoring, we can notice the Kamchatka Branch of Geophysical Survey (from the Russian Academy of Sciences) and the Kamchatka Volcano Eruption Response team (KVERT). A web interface allows to show the activity in this region (www.emsd.ru). Information about the plume height are retrieved in NRT using camera, and email notification are sent to the VAACs and researchers. 300 Concerning a specific support to aviation with a global coverage of possible volcanic emission, as far as we know, we can mention three existing EWS. The NOAA/CIMSS (US National Oceanic and Atmospheric Administration /Cooperative Institute for Meteorological Satellite Studies) VOLcanic Cloud Analysis Toolkit (VOLCAT) web site features NRT processing of many geostationary and low-earth orbit satellites covering much of the globe (https://volcano.ssec.wisc.edu). VOLCAT includes a collection of sensor agnostic algorithms for detecting, 305 tracking, and characterising volcanic ash and SO2 (e.g. Pavolonis et al., 2015aPavolonis et al., , 2015bPavolonis et al., 2018;Hyman and Pavolonis, 2020), and the products are utilised by many of the VAACs. The VOLCAT products are scheduled to achieve full operational status in NOAA in 2023. The alerting service consists of four categories of alerts: sudden changes in thermal output (hot spots), newly detected ash emissions, newly detected rapidly developing clouds near known volcanic vents, and newly detected SO2 emissions. Users can subscribe and 310 configure alert subscriptions using a web interface (https://volcano.ssec.wisc.edu/alert). Alerts are shown on an event dashboard. Access to the alerts and event dashboard is currently restricted to VAACs, MWOs, volcanic observatories, and research collaborators, as these are considered pre-decisional products. On the other hand, the SACS EWS is a highly successful system used by agencies worldwide (Brenot et al., 2014). This system, hosted by one of EUNADICS partners (BIRA; http://sacs.aeronomie.be), was initiated by the European Space Agency 315 aims at supporting the Volcanic Ash Advisory Centres, like Toulouse VAAC and London VAAC. NRT data of SO2 and volcanic ash are derived from hyperspectral sensors onboard polar orbiting satellite, in the ultravioletvisible (UV-vis) range with OMI, GOME-2B, GOME-2C, OMPS and TROPOMI, and in the infrared (IR) range https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c Author(s) 2021. CC BY 4.0 License.
with AIRS, IASI-A and IASI-B. The SACS multi-sensors system addresses automatically worldwide detection of volcanic plumes of SO2 and ash notifications, sending alert by email to interested parties 320 (https://sacs.aeronomie.be/alert). Finally, a continuous analysis and a systematic surveillance is in operation at the Free University of Brussels (ULB) in order to detect a possible anomalous threshold of SO2 caused by a volcanic eruption. This automatic system, based on IASI data (onboard MetOp-A & -B & -C), sends email alerts to the VAACs and key users when high SO2 levels are detected (http://cpm-ws4.ulb.ac.be/Alerts). Information about SO2 column density and layer height is provided. This IASI detection system provides automatically inputs of 325 SO2 (and ash) products to SACS system, aiming at providing NRT SO2 and ash measurements related to volcanic emissions.

Example of systems related to dust and sandstorms
The monitoring of extreme dust events is critical for aviation. Amongst existing system, we can mention the WMO ability of countries to deliver timely, quality sand and dust storm forecasts, observations, information and knowledge to users through an international partnership of research and operational communities. This service is divided into three regional centres. The WMO SDS-WAS Regional Centre for Northern Africa, Middle East and Europe (https://sds-was.aemet.es), is coordinated by a Regional Centre in Barcelona, Spain, and hosted by the 335 State Meteorological Agency (AEMET) and the Barcelona Supercomputing Centre (BSC). The WMO SDS-WAS Regional Centre for Asia (http://www.asdf-bj.net), is coordinated by a Regional Centre in Beijing, China, and hosted by the China Meteorological Administration (CMA). The WMO SDS-WAS Regional Centre for the Americas (http://sds-was.cimh.edu.bb), established in the USA with a possible regional centre hosted by the Caribbean Institute for Meteorology and Hydrology (CIMH) in Barbados, focuses on the health implications of 340 airborne dust. The prime objective of the three SDS-WAS regional centres is to lead the development and implementation in the region of a comprehensive system for mineral dust observation and forecast, with special emphasis on extreme sand and dust events. Theses observational systems aim to a continuous dust monitoring, validation and verification of forecast products and data assimilation into numerical models. SDS-WAS models used include ground observations (particulate matter measurements progressively becoming available in NRT, 345 indirect information from regular weather reports and remote-sensing retrievals from sun photometers or vertical profilers) and satellite products (single-band images, qualitative multi-band products designed to improve dust identification or quantitative retrievals). Currently, the WMO SDS-WAS Regional Centre for Northern Africa, Middle East and Europe provides a multi-model platform with analysis and +54 hours forecasts for 12 dispersion models (Nickovic et al., 2001;Woodward et al., 2001;Zakey et al., 2006;Benedetti, et al., 2009;Morcrette et al., 350 200;Colarco et al., 2010;Pérez et al., 2011;Haustein et al., 2012;Basart et al., 2012Basart et al., , 2020Lu et al., 2016). SDS-WAS contributes to the International Cooperative for Aerosol Prediction (ICAP), an unfunded international forum for aerosol forecast centres, remote sensing data providers, and lead systems developers, which coordinates the first global multi-model Ensemble for aerosol forecasts, as described in Sessions et al. (2015). The use of the https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c Author(s) 2021. CC BY 4.0 License. multi-model system is overall better than any individual model. Over specific regions, combining several models 355 leads to better forecasts than the best individual model even when number of ensemble members is small.

Example of systems related to smoke from wildfires and biomass burning
NASA provides a global EWS related to fire detection from its Fire Information for Resource Management System (FIRMS). It provides Fire Radiative Power (FRP) from Low Earth Orbit (LEO) satellites sensors, i.e. MODIS instruments onboard the Terra and Aqua satellites (Kaufman et al., 1998;Giglio et al., 2003;Justice et al., 2011) 360 and VIIRS sensor onboard Suomi-NPP (Csiszar et al., 2014). FIRMS focus and objectives include providing quality resources for fire data on demand, working with end users to enhance critical applications assisting global organisations in fire analysis efforts, delivering effective data presentation and management (https://firms.modaps.eosdis.nasa.gov/alerts). On the other hand, CAMS has developed a monitoring system, which provides observations of fire detection and forecasts of smoke dispersion. By using NRT observations of 365 the location and intensity of active wildfires, i.e. FRP product based on SEVIRI (Roberts et al., 2015) from the EUMETSAT LSA-SAF (http://landsaf.meteo.pt), CAMS estimate the emissions of aerosols and pollutants. This is done through its Global Fire Assimilation System (GFAS). This allows active fires to be monitored and their estimated emissions to be used in the CAMS forecasts to predict the transport of the resulting smoke in the atmosphere (https://atmosphere.copernicus.eu/fire-monitoring). The forecasts are used in air quality apps, to help 370 people limit their exposure to pollution, and by policymakers and local authorities to manage the impact of fires.

The Copernicus Emergency Management Service (EMS) has developed the European Forest Fire Information
System (EFFIS; http://effis.jrc.ec.europa.eu; see EFFIS, 2018). This system supports the services in charge of the protection of forests against fires in the EU and neighbour countries and provides the European Commission services and the European Parliament with updated and reliable information on wildland fires in Europe. 375

Example of systems related to radionuclide clouds
This fourth type of system addresses the monitoring of nuclear accidents and radionuclide plumes. The development of such system is quite sensitive and the dissemination of information is confidential. Most European dose-rate results are recorded at the European Radiological Data Exchange Platform (EURDEP) web site (https://eurdep.jrc.ec.europa.eu) but accessing the site and downloading data requires agreements. A collaboration 380 with the Joint Research Centre of the European Commission (JRC) is required to establish NRT or archive access to data (including historical data). Individual countries can provide their own data (i.e. providing gamma dose rates, including spectrometric, and activity concentrations in air). For selected case studies and research, airborne activity concentration measurements of radionuclide-bound aerosols may be provided by selected laboratories.
EMERCON (Emergency Convention) messages are also produced by the International Atomic Energy Agency 385 (IAEA) through the WMO RSMCs. An EMERCON message is a descriptor referring to the official system for issuing and receiving notifications, information exchange and assistance provision through the IAEA's Emergency Response Centre in the event of a nuclear or radiological emergency. The ICAO system allows the issuance of SIGMET (SIGnificant METeorological Information message) for radioactivity , from the ground to  The impact of the SO2 exposure and the sulphur damage to engines has affected hundreds of flights in the last 400 decade. Sulphidation mechanisms can cause damage to the engine with solid diffusion process or corrosionfatigue. A flight through a volcanic plume and the exposure to SO2 is a problem for passengers and aviation stakeholders as it is a threat to the safety and health, and it requires turbine maintenance. The detection of SO2 from satellite is straightforward (see Fig. 1), and of great interest for aviation. Generally, wAhen SO2 clouds reach the free troposphere up to the lower stratosphere, it is a good indicator of volcanic activity. Height satellite sensors 405 are considered by EUNADICS EWS for the SO2 detection, that is retrieved by OMI, GOME2-B, OMPS and TROPOMI in the UV-vis (Yang et al., 2007;Rix et al., 2009Rix et al., , 2012Li et al., 2017;Theys, 2017Theys, , 2019, and AIRS, IASI-A and IASI-B in the IR range (Prata and Bernardo, 2007;Clarisse et al., , 2012Clerbaux et al., 2009).
Details about the detection and the limitation of these products can be find in Brenot et al. (2014).

Satellite IR sensors considered by EUNADICS (IASI-A & -B)
can measure the SO2 layer height (SO2LH) using optimal estimation algorithms . The SO2LH retrieval is very fast, with an accuracy of 1-2 km, which can be obtained even for low SO2 column (under 1 DU). SO2LH results are obtained for estimates between 3 and 21 km, with low performance for heavily saturated plumes. Figure

Selective detection of ash
The volcanic ash detection from satellite is far from trivial but is essential for aviation, as it can cause severe 420 damage to turbine engines (Clarkson et al., 2016). Differential absorption by ash between two channels, e.g. by using of the brightness temperature difference (BTD) between the 11 and 12 μm channels for SLSTR instrument, can be used to detect volcanic plume, as illustrated in Fig. 4. BTD results from SLSTR are expressed in K. A negative value of -2K issue an alert for a data granule (3 granules with data for a time duration of about 3 minutes are shown in blue in Fig. 4). 425  The BTD technique can yield to false detection in the presence of absorbing aerosols other than ash (e.g., desert 430 dust). However, using SLSTR sensors (onboard Sentinel-3A & -3B), combining automated aerosol detection, the algorithm of Virtanen et al. (2014) can provide ash/dust height estimate (Fig. 5). This algorithm is based on matching top of the atmosphere reflectances and brightness temperatures of the nadir, and 55 forward views, and using the resulting parallax to obtain the height estimate (stereoscopic technique).
The absorption signatures of ash can be discriminated from other types of absorbing aerosols using hyperspectral 435 thermal infrared sensors (Clarisse et al., 2010). The IASI-A&B and AIRS ash index products used in EUNADICS EWS are based on a three-step process (Clarisse et al., 2013), which provides ash detection with levels of confidence. Four colour-codes are used to defined the level of confidence of pixel ash detection. High level detection appears in red (ash is almost certainly present with less than 1% of false alerts; this issues an alert), and Medium level in orange (ash detection with high confidence nevertheless no alert is issued). Low level is in yellow 440 (mineral aerosol signature observed in spectra and proximity to a high confidence detection), meaning the volcanic ash detection is highly probable but false detections can still occasionally occur. The unknown detection appears in white (no ash detected by the algorithm). Figure 6 illustrated a selective detection of ash from AIRS few hours after the start of Raikoke eruption. For the dedicated EU area, a last ash detection from geostationary satellite is considered by EUNADICS EWS, i.e. the ash mask, column load (in kg/m²) and top height (in km) products from SEVIRI (onboard MSG). The VADUGS (Volcanic Ash Detection Utilising Geostationary Satellites) algorithm used to retrieve these products, is based on a backpropagation neural network which combined BTD technique with mask of high clouds and 450 atmospheric model look-up tables for a broad range of particle concentrations for different ash types in various layers (Kox, 2012;Kox et al., 2013;Graf et al., 2015).    Clarisse et al. (2020) have recently presents a new algorithm for estimating atmospheric dust optical depths (dust AOD) and associated retrieval uncertainties. This AOD retrieval, specific to dust, are based on the calculation of 470 a dust index and on a neural network trained with synthetic IASI spectra. It has an inherent high sensitivity to dust and efficiently discriminates mineral dust from other aerosols, as show in Fig. 8 (see Clarisse et al., 2013Clarisse et al., , 2020. EUNADICS EWS considers dust flag and dust AOD from IASI (onboard MetOp-A, MetOP-B and MetOp-C).
As an example, Saharan dust AOD is observed by IASI-B on 30 June 2020 (Fig. 9). This approach essentially relies on the collocation (or near collocation) of a selective product with a non-selective product to trigger (and propagate) alerts. Figure 10 illustrates the triggered approach applied by EUNADICS EWS to obtain detection of dust from the combined Dark Target and Deep Blue AOD products from MODIS-Terra (Levy, R., Hsu, C., et al., 2016) using selective detection of dust from IASI-A (colour code red for dust flag = 1).
This figure shows all the AOD pixel values from MODIS (for a data granule measuring between 11:20 and 11:25 480 UTC on 15 October 2017). A threshold value is used to select the AOD from MODIS (threshold of 0.7) for testing geographic matching (critical distance of 200 km) with alert pixels (from IASI-A). These alert data pixels have been obtained from a previous AOD alert from IASI-A, and are stored by EUNADICS EWS as active dust alert pixels (defining the pool of dust alerts). Each pixel of this pool is characterised by a position and a time. An alert pixel of the dust pool stays active during 12 hours. If a match is obtained between selected MODIS AOD pixels 485 and the alert pixels from the dust pool, the MODIS AOD pixel is determined as an alert pixels (which join the dust pool). A completion of the dust cloud from MODIS AOD is operated looking at the neighbouring pixels (with AOD value over 0.5) of all the newly obtained dust alerts (completed cloud in Fig. 10). On 15 October 2017, the Hurricane Ophelia reached Ireland (downgraded to storm) and during its northward movement brought desert dust particles and smoke particles from raging fires over the Iberian Peninsula in Europe (Osborne et al., 2019). During this twin event, OMPS instrument measured high values of AAI (between 12:50 and 14:40 UTC) due to dust from Africa's Saharan desert and smoke from wildfires in Spain and Portugal. Figure  495 11 illustrates the triggered approach applied to obtain selective detection of dust and smoke cloud from OMPS AAI using, respectively, the pools of dust alert (fed by IASI-A and MODIS-Terra alerts) and smoke alert. The active pool related to smoke is fed by the NRT fire localisation from NASA-FIRMS, i.e thermal anomaly and FRP data from VIIRS sensor, in the present case. EUNADICS EWS identifies smoke alerts from OMPS on 15 October 2017 (Fig. 11). These alert pixels remain active for 24 hours in our system. On 16 October, the smoke pool trigger new OMPS smoke alerts, which are observed over France and UK (Fig. 12). In the same time, the system isolates dust alerts from OMPS, which are located over the Saharan desert and the Atlantic Ocean. Using both alert pools (dust and smoke), the displacement 505 of the two hazardous clouds (on the 16th and the 17th of October 2017), is shown in Fig. 12. On 17 October, the smoke plume and the respective OMPS alert pixels are retrieved over northern Europe. In the same time, dust alert pixels are still observed over the Saharan desert and the Atlantic Ocean. A tailored alerting product from the European Aerosol Research Lidar Network (EARLINET; https://www.earlinet.org) has been developed during the EUNADICS project (Papagiannopoulos et al., 2020). It has been designed for NRT EWS applications. Using a stand-alone lidar-based method for detecting airborne hazards (volcanic ash and desert dust; no discrimination), this product is based on the EARLINET Single Calculus 520 Chain (version 5.1), and provides temporally high-resolved, calibrated attenuated backscatter and volume depolarisation ratio (at 532 nm), and cloud mask. The vertical resolution is 7.5 m, and the temporal resolution is 30 s. From these calibrated data, further particle-like products (particle-like backscatter and depolarisation ratio) can be retrieved that act as the basis of the tailored product. The final product (to be used by EUNADICS EWS) is the aviation alert for desert dust/volcanic ash with a three colour-codes. This alert product uses particle mass 525 concentrations (pmc) based on backscatter coefficient thresholds. High level detection appears in red (almost certain detection of ash or dust aerosol with pmc ≥ 4 mg/m³), Medium level in orange (4 mg/m³ > pmc ≥ 2 mg/m³).
Low level is in yellow (2 mg/m³ > pmc ≥ 0.2 mg/m³). An example of alert from EARLINET is shown in Fig. 13. In addition to the NRT distribution of its observation data in NetCDF format, the E-PROFILE network from 540 EUMETNET provides an interactive geographical overview of observations. Quicklooks and interactive plots of aerosols and clouds from the network of Automatic Lidars and Ceilometers (ALC, currently 345 units in continuous operation) as well as of wind profile observations from radar wind profilers (40 units in continuous operation) and from precipitation radars (96 units) can freely be accessed at https://e-profile.eu (Haefele et al., 2016). An example is given in Fig. 14    The IMO volcanic observatory has recently improved the VESPA (Volcanic Eruptive Source Parameter Assessment) system. In case of an on-going eruption, information about the plume height and mass eruption rate from weather radars (C-band, X-band mobile; http://brunnur.vedur.is/radar/vespa), is provided to EUNADICS EWS, which ingests such information (links to quicklooks of the top height of volcanic plume; i.e., 2D images and time-series). An example of the ash plume detected by a mobile X-band radar, during the Grimsvötn 2011 585 eruption, is displayed in Fig. 16. Data file describing the plume height are also available in NRT, providing the height of the highest point where a significant radar reflection is detected within 10 km distance of the volcano, and the height of the next radar elevation angle above volcano, where plume was not detected. Arason et al. (2011) and Petersen et al. (2012) present more details about the radar and webcam products. 590

Nuclear hazard and EU network
Several sources of radioactivity monitoring data are available, and to obtain robust, harmonised and real-time data is a challenge for the nuclear events. Although not validated Gamma Dose Rate (GDR) radiological monitoring 595 data from most European countries is available in NRT from the EURDEP system (European Radiological Data Exchange Platform, https://eurdep.jrc.ec.europa.eu), the usage of such data inside an automatic notification system common in all the European domain is not done. However, EUNADICS, with view of a future operationalisation, requires access to real-time well established data and data channels. Providing NRT availability and accessibility to such data is complex. The monitoring information is collected from automatic surveillance systems in 39 600 European countries. Without a radiological event, these data provide information on the background radioactivity and its variability. However if a nuclear accident with gamma emitting radionuclides occurs (e.g. anomalous GRD values over 0.5 µSv/h recorded), the network must be able to capture the existence and the geographical distribution/evolution of the event within the limitations of the network.  The approach taken to design and implement a notification system for nuclear events has largely relied on already existing capacities of the two main contributors to the current work, i.e. ZAMG and STUK. By making use of their national and international mandates and capacities, STUK and ZAM design the interfaces from monitoring data to notification and posterior alerting have been designed. As illustrated in Fig. 17, the current approach is 610 based on both EURDEP data and the EMERCON (Emergency Convention) messages produced by IAEA through the WMO Regional Specialised Meteorological Centres (RSMCs). Both data sources are ultimately released by STUK after proceeding the filtering of the EURDEP data implemented and collecting the EMERCON information (from ZAMG). STUK finally provides the information to the automatic alerting system developed in EUNADICS. 615 Following its national responsibilities, STUK has implemented an elevated GRD alerting software based on measurements available in the EURDEP system. It consists of separate Python scripts that handle data collection, alerting, and database maintenance functions. This work at national level has been further extended to fit the 620 purpose of EUNADICS EWS. The alerting condition is checked in 10 minutes intervals. A nuclear threat is The detailed information link leads to a HTML webpage containing the map with zoom and pan features (Fig.   19). The map contains current real online measurements of the EURDEP network. However, and as a note of caution, currently in the EURDEP system, most European countries do not share validated radiological monitoring 635 data in NRT and gaps may appear per country or time window. In case of GDR alert, an alert data product is created by EUNADICS-AV EWS, in a homogenised format (see next section 4.4.3).

Overview of EUNADICS EWS
The EUNADICS EWS is built on the SACS system (Brenot et al., 2014), which provides NRT satellite data products (SO2 columns, ash and aerosol index) to aviation stakeholders in the forms of maps and email 640 notifications. The SACS system and its web interface (https://sacs.aeronomie.be) is dedicated to volcanic hazard.
The development achieved by EUNADICS concerned an upgrade of SACS system (volcanic emission), and the extension to other airborne hazard (dust, smoke and radionuclide clouds), with creation of news alert products.
EUNADICS EWS is a prototype multi-hazard system which has expanded SACS system and create new functionalities (based on existing mechanisms of SACS) by using: (1) key satellite products from IASI/MetOp-645 A&-B&-C, SEVIRI/MSG, TROPOMI/Sentinel-5 Precursor, SLSTR/Sentinel-3A&B and MODIS-Terra & -Aqua sensors, (2) ground measurements from EU networks (EARLINET and E-PROFILE) and regional networks from  The development of pre-alerting mechanism is based on email notifications (volcanic and radionuclide cloud hazards only) and the creation of homogenised tailored data alert products and alert maps, as illustrated in Fig.  655 20. Initially, SACS system was built integrating NRT data products (SO2 columns and aerosols/ash indexes) from 7 polar-orbiting instruments (OMI, OMPS, GOME-2A, GOME-2B, AIRS, IASI-A and IASI-B) in a single monitoring and alerting system for volcanic eruptions. This system considers only satellite instruments on board polar orbiting satellites. Figure 20 shows the new additional data information ingested by EUNADICS EWS.
NRT products from satellite, ground-based and in-situ platforms/instruments are provided by EUNADICS 660 partners and external data sources (Tab. 2 and 3), and transferred to the EWS. The blue block in Fig. 20 indicates the multi-source input data (Ash, SO2, dust, radioactive and smoke cloud observations). The structure and the choice of the data products considered in the EUNADICS EWS relies on the user requirements and risk assessments presented in section 2. The automated EWS (including routine data products) applies its own mechanisms to create NRT images and to issue alerts. This represents an extension of the SACS system (Brenot 665 et al., 2014) to other hazards and/or instruments, taking into account inputs from existing systems. Innovative Alert products are created using data from 13 sensors onboard polar orbiting satellites, 1 geostationary imager (SEVIRI onboard MSG9), EU networks (EARLINET, E-PROFILE).  The implementation of alert notifications and data products requires a two steps approach. First, a specific establishment of warning criteria for the different sensors (satellite, ground-based and in situ) and for the different types of alerts (i.e. issued from the detection of volcanic, sand/dust storms, smoke from fire and radionuclide plumes) is required. A particular attention is given to the avoidance of false notifications (e.g. due to noise or 680 retrieval failures) or overly frequent/redundant notifications (caused by highly dispersed plumes). As a baseline, the ash/SO2 alerts criteria of some of the satellite products used in SACS (see Brenot et al. 2014), and considered by EUNADICS EWS, are summarised in Tab. 5. This includes the name of the quantity products, the type of instruments and the platform for satellite sensors, the criteria of alerts and the limitation, and the availability and access of data (Off-line or NRT). The other products and criteria used by EUNADICS EWS are also listed in Tab. 685 https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c Author(s) 2021. CC BY 4.0 License. 5. The second step is to combine the information from all the products in one multi-sensors system. The EWS relies on pre-defined geographical regions, and notifies the start of an event to parties of interest (volcanic and radionuclides clouds only) as soon as a new airborne hazard plume is detected. If within a period of 24 hours, a plume is detected again in the same region (for the same quantity product) no new notification is generated (to avoid sending redundant information). An illustration of this two steps approach of EUNADICS multi-sensors 690 EWS is presented in Fig. 21. The description of the successive processes, related the first (definition of alert) and the second step (combination of all data and avoidance of redundant information), is the following: 695

Mechanism EWS 675
1) The first step is the analysis of the data using the alert criteria defined in Tab. 5. This takes place as soon as the EWS harvests new observations. After the detection of airborne emission (from a natural hazard or a nuclear source) and the start of an event, there are potentially multiple warnings generated by the system. For this reason, it has been decided to consider a set of world regions of 30° by 30° plus two polar regions poleward of 75° in latitude. Each 700 region is associated to a name/number for a total of 62 regions (see Fig. 9 of Brenot et al., 2014). As soon as a notification is issued, the related region is flagged 'ON'. 2) The second step is to check the "warning status" of this region. If there is no on-going notification for this region (meaning no notification since 24 hours), the warning status can possibly become ON. The system compares the time of observations with the processing time. 705 If the delay is less than 8 hours, a notification is issued. Then, alert data products are created and archived. If the "warning status" is already ON (i.e. on-going event determined using existing pool of active alert pixels; see section 4.1.4), there is no notification issued, but an update of the event is operated in the alert data archive.
This set-up enables to provide timely information to the users and also to avoid issuing too many notifications 710 (maximum one notification per region and per 24 hours).
When an alert is issued by the system, the first step is to check whether this represents a new event (Start of Natural airborne Hazard -SNH) or it is linked (LNK) to an on-going event. At this stage, a characterisation of the source of the hazard can be obtained a) directly from the alerting process (e.g., the name and the height of the plume of a volcanic eruption, as provided by a VONA message, or the name and location of nuclear central facing 715 an incident), b) or by the "mother alert". In fact, if the new alert is linked to a previous event (i.e., proximity of the new alert pixels with previous alert pixels of an on-going event), the source of the hazard can be transferred.
To obtain the LNK status, a proximity, in space and time (between new alert pixels and previous ones), is required to link two alerts together; see the description alert products in section 4. The proximity setting is specific to each type of event (volcano, dust, smoke or nuclear) and the observational technique of each instrument. Generally, 720 the distance criteria range between 500 km to 1000 km (respectively for a time-threshold of 14h or 26h). This depends on the mean revisiting time of each region, i.e., the frequency of observations for the same region. Note that sometimes the source cannot be determined and is considered as "unknown".
EUNADICS-AV EWS is based on the detection of volcanic ash/SO2, sand/dust storms, smoke plumes and nuclear accidents. If new alert pixels drive the start of a new event (SNH status), our EWS creates a log file and an internal 725 specific notification. This triggers the sending of an email to stakeholders or public users (currently only in the case of volcanic hazard), or simply the sending of an email to the management of EUNADICS EWS (prototype status/check). A notification is associated to the creation of a dedicated webpage related to the event. If a new alert is linked to an on-going event, a confirmation of alert is established with creations of new mass, alert products and an update the alert webpage (see section 4.6.2). A new detection of alert pixels (status SNH or LNK) is 730 associated to the implementation of an event, which includes the reference number (date of the first alert) associated to alert products and links to quicklooks from EARLINET, E-PROFILE or other observations from the volcanic observatories. The collection of NetCDF alert data products (NCAP) and the associated data directory of an event type (ASH, SO2, DST, SMK or NUC, respectively for ash, SO2, dust, smoke and radionuclide clouds) is created, with the objective of triggering dispersion model. Access to EUNADICS partners and key users is 735 assured via ftp or https connection. https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c Author(s) 2021. CC BY 4.0 License.

Alert products
For each alert/event issued by EUNADICS EWS, the associated event type is created/updated, and the alert archive is completed. Figure 22 shows the three kinds of alerting production related to an event. The diagram in Fig. 1 shows the links between the blocs of activities of EUNADICS and the red arrows characterise the specific EWS processing chain (see also Fig. 20), from observation retrievals, data harvesting, triggering with the implementation of alerts, to the delivery of alert data products for modelling integration and 745 ATM stakeholders.

NRT observations
NRT observations (from satellite, ground-based and in situ platforms/instruments are provided by EUNADICS partners, external data sources or thanks to agreements; see Tab. 2 and 3). A prototype data portal has been implemented in a demonstration phase of EUNADICS project (see the exercise study of Hirtl et al., 2020a). In 750 case of a future operationalisation of EUNADICS activity (TRL higher than 5), all the NRT observations will be visible on the EUNADICS data portal.
The Routine data products, based on NRT products from 8 satellite hyperspectral sensors (i.e. OMI, OMPS, GOME-2B, GOME-2C, TROPOMI in the UV-vis, and AIRS, IASI-A, IASI-B in the IR range) related to the detection of volcanic eruptions, sandstorms or smoke from wildfires, and can be consulted and monitored through 755 the SACS/EUNADICS web interface. The currently operational SACS website (https://sacs.aeronomie.be) is a self-contained system that allows the consultation of NRT satellite data and provided alerts to subscribed users in case of detection of elevated amounts or concentration of volcanic emissions. Within EUNADICS project, the development of a new SACS interface has started. This work in progress is based on modern visualisation methods and handling of geophysical data. It is currently in development phase and allows to monitor user-selected satellite 760 sensors and products in the form of zoomable/pannable maps, using GeoTiff, GeoServer and Web Map Service (WMS) facilities for serving geographical data. At the moment, all the NRT observations linked to EUNADICS EWS relies on the current SACS web interface. Figure 23 shows NRT observations of a volcanic burst from a paroxysm at Etna on 28 February 2021). Note that currently, the NRT images from GOME-2 correspond to GOME-2B images for SO2 and cloud cover, and to a composite images of GOME-2B and GOME-2C for the 765 Aerosol Absorbing Index images (AAI).  Currently email notifications from EUNADICS EWS for public or governmental users take only place after volcanic and radionuclide cloud hazards detection. Data products are collected by EUNADICS data harvesting 780 facility and transferred in NRT for analysis by our EWS. The automated EWS applies specific mechanisms to issue selective detection (extension of the SACS system to other kinds of alerts and instruments) but also take into account inputs from existing systems, like NASA-FIRMS, VONA messages and EURDEP/EMERCON messages. In case of the detection of a natural airborne hazard in a specific SACS region (see section 4.5), a notification is sent internally to EUNADICS partners (one notification per affected region). In case of exceptional 785 SO2/ash concentration detected, a notification is sent to stakeholders (email) with relevant information (e.g., time, position and highest value detected) and a link to a dedicated webpage (see Fig. 24). The SACS system has currently about 300 users (from the VAACs, NMS, scientific institutions, airlines, pilots, other ATM institutions, and other public users).
On this webpage, images of volcanic observations (e.g., ash, SO2 vertical column, SO2 layer height if available 790 for the instrument in alert) are shown. Additional links to other images are provided (i.e., links to interpolated plot and google earth file), as illustrated in Fig. 25  source. Note that if two SACS regions are affected by an eruption, the same SO2 mass can be provided in two successive SO2 notifications related to 2 different regions (however the max values are different). Figure 25 shows an illustration of the tailored visualisation (of an SO2 alert from TROPOMI) available from the alert webpage created by EUNADICS EWS. The link to this webpage is provided in the email SO2 notification (Fig. 24).

805
Notification information is also shown on this page. A link to a tailored SO2 alert image is provided. This image contains key information (in the legend), and shows the SO2 images with the time of alert (near the max value) 810 and the name/position of the identified volcano source (shown by a red triangle). There links to Google Earth images (SO2, AAI and CCF) are indicated. We can see precisely the SO2 plume, the possible ash aerosol and the cloud cover situation. This last information about the cloud cover is essential to know if the SO2 observation is optimal. Indeed, the vertical SO2 columns measured by UV-vis satellite sounders only considers the SO2 amount above the cloud and underestimate the possible concentration inside or under the cloud (see the limitations 815 presented by Brenot et al., 2014). It is also important to mention the role of the cloud in the AAI estimates. cloud bow, viewing zenith angle dependence, sunglint, and unexplained increase in AAI values at extreme viewing and solar geometries). In the case study shown in Fig. 25, the high AAI values are not due to the sunglit effect. We can see that the delimitation of the AAI cloud correspond precisely to the occurrence of high CCF 820 values (Google earth facility is quite convenient for this kind of investigation). This leads us to think that the ash cloud (observed by AAI in clear sky condition and highlighted in Fig. 25) is probably not observed under the cloud/aerosol structure (area shown with high CCF). The AAI pattern observed here reduces probably the real size of the ash cloud at low altitude (under the cloud cover). Note that the sunglit effect is systematically flagged (and avoided) in the NRT TROPOMI AAI images available from SACS website. However, the AAI Google Earth 825 images do not use this flag. This is a way to see the whole scene (eventually with artefact AAI high values, but also eventually with natural airborne aerosol observations dismissed by the sunglint flag).
If a nuclear accident takes place, EUNADICS EWS send a notification to authorities (restricted dissemination of the information). An example of EMERCON message is shown in section 4.3 (Fig. 18).

Data file transfer 830
The data integration in dispersion models is essential in the resilience process and the decision-making after a crisis in aviation related to airborne hazard. It is also critical for the ATM stakeholders to receive homogenised and easy-to-read data to have a fast and clear view of the scene during such a crisis Sivčev, 2011, 2012). This is why one of EUNADICS objectives was the implementation of alert data products (with metadata, key information about the alert, flag and gridded data), allowing a good dissemination of information. 835 The data file transfer established by EUNADICS EWS consists in the creation of Alert Products in a homogenised, standardised, format (NetCDF), so called NCAP file. The NetCDF format has been chosen because this is a very common and convenient format (easy access), with relevant metadata information for users. Routine alert product is operated using a Python script that handle data collection, alerting, and database maintenance functions.

840
The content of the alert products is the following: links (SACS images, SACS notification, quicklooks from EU or regional network) More details about the description of the content of the NCAP files is presented in Appendix 1. Figure 26 illustrates an example of NCAP (dust from OMPS). An snapshot of an overview of a NCAP fiel using 855 hdfviewer tool is shown. Arrows in black show associated the data field with images (all pixels, dust alerts pixels, extended dust could, and the contours identified with mean values). See section 4.1.4 for more detailed about this triggered detection of dust using OMPS AAI product. The inventory of the NCAP implementation in EUNADICS EWS is presented in Table 6.

Performance verification, conclusion and future developments
Within the system definition and design of EUNADICS EWS, a review of system requirements has been 875 considered (see section 2). As part of our EWS, an assessment of the NRT capability of the system has been undertaken for all the products implemented in EUNADICS system (monitoring and alerting production from satellite and ground-based instruments related to the detection and situational awareness of natural, i.e. sandstorms, volcanic eruption and wildfires). The performance of the alerting approach developed within EWS is illustrated in Fig. 27   which shows the feasibility of EUNADICS prototype service. EUNADICS EWS has developed a concept for starting V3 validation (i.e. pre-industrial development & integration). With regard to the alert products developed, a verification of requirements has be performed and a verification of EUNADICS EWS performance obtained 895 (Fig. 27). A validation of the global concept of EUNADICS and its potential benefits, has been demonstrated during the EUNADICS exercise (Hirtl et al., 2020a), showing the benefits in a limited framework.
The development achieved in EUNADICS EWS shows the significant interest in using selective detection of natural airborne hazards from polar orbiting satellite. The combination of several sensors inside a single global system demonstrates the advantage of using a triggered approach to obtain selective detection from observations, 900 which cannot initially discriminate the different aerosol types. Satellite products from hyperspectral UV and IR sensors (e.g. TROPOMI, IASI and SEVIRI) and retrievals from ground-based networks (e.g. EARLINET, E-PROFILE and the regional network from volcanic observatories), are combined by our system to create tailored alert products (e.g. selective ash detection, SO2 column and plume height, dust cloud and smoke from wildfires), with identification and traceability of extreme events. 905 To conclude, EUNADICS EWS has development new tailored alert products for aviation, i.e. NRT observations, notification and the implementation of NetCDF Alert data Products (NCAP). EUNADICS EWS achievements concern:  the improvement of the NRT discrimination between volcanic ash and other aerosols (dust or smoke) or meteorological clouds 910  the NRT retrievals of plume heights (ash and SO2)  the NRT retrievals of volcanic ash mass loadings with use of SEVIRI onboard a geostationary platform  the use of polar orbiting NRT measurements with better spatial resolution  the use of key measurements from the ground-based network, in particular lidars and ceilometers measurements, as well as near-source parameters from volcanic observatories 915 Only the aspect of EUNADICS related to our early warnings system is presented in this study. The better characterisation of the source obtained by EUNADICS EWS is complementary and beneficial for other developments of EUNADICS consortium. With a demonstration exercise, Hirlt et al. (2020a) shows the interest of EUNADICS system in the route optimisation of the European air space during a volcanic eruption of Etna. 920 EUNADICS activity about the assimilation, forecasts and inverse modelling, and the characterisation and the impact of the source term in dispersion modelling, is presented by Hirtl et al. (2020b) and the two recently submitted studies from Plu et al. (2021aPlu et al. ( , 2021b. EUNADICS consortium will now target an operationalisation of its activity, in the frame of SESAR H2020, with 925 the objective of completing the TRL 6 (demonstration in a relevant environment). EUNADICS EWS passes with success the performance verification. Concerning future plans with regard to natural airborne hazard, collaborations are on-going with key stakeholders in charge of proceeding data integration in dispersion model and providing advisory for aviation (i.e. VAACs, NMS), but also in collaboration with WMO SDS-WAS, in the frame of InDUST COST action (https://cost-indust.eu) with the use of EUNADICS dust alert products.  (Hyman and Pavolonis, 2020;Corradini et al., 2020;Hedelt et al., 2019). In addition, the development of EUNADICS EWS is also used and contributes to a recent SESAR H2020 project, which has the objective to upgrade the EUNADICS prototype EWS with other hazard to aviation. In addition to natural airborne hazard (volcano, dust and smoke), the ALARM project (multi-hAsard monitoring and earLy wARning system; https://alarm-project.eu) 940 plans to develop early warning and NCAP files with respect to Space Weather, Severe Weather and Environmental hotpots risk to aviation. This new activity of EUNADICS/ALARM EWS might join the ARISTOTLE consortium basis (http://aristotle.ingv.it) in the future.
Concerning the operationalisation of EUNADICS with regard to nuclear accident, the European network of experts, called Ring of Five (Ro5), will be approached also to become part of their mailing distribution system 945 that is used whenever one of the Ro5 laboratories detect something anomalous in their measurement networks.
Although in this case, unlike with the EURDEP data, the data will not be harmonised, it can be used as a triggering system or, at least, as an awareness system potentially for such events for which the gamma dose rate monitoring may not provide useful information (very far away sources or for radionuclides that are pure beta emitters, for instance). We can clearly see the interest of EUNADICS consortium to proceed a future relay of radiological data 950 (gamma dose rate and radionuclides concentrations in ground-level air), to create early warnings using homogenised critical dataset, to be used to trigger data assimilation / inverse modelling for source term estimate.

Appendix 1: Description of the NCAP files
All the data pixels provided in a NCAP file corresponds to all the data information relevant for data integration (for example from IASI and TROPOMI satellite sensors information about the uncertainty of measurements is 955 provided, e.g. SO2 VCD for 8 altitudes from IASI and the averaging kernel from TROPOMI; see Clarisse et al., 2011, and Theys et al., 2017. For the lidar data, all the data observations correspond to the back scattered coefficient of the whole profile (the same for te alert is a flag-value along the profile with low, medium and high alert status). For the radionuclide data, it corresponds to all the gamma radiation data from EURDEP network. 960 About the level of severity (LOW, HIGH), for an SO2 alert the HIGH level is obtained if the mass loading is over 5 kt. For other satellite alerts, we have defined the level of severity considering the area affected by alert pixels (if this area is over 100000 km², this brings a HIGH level). For the EARLINET, the criteria is based on the number of high alert pixels. If this number is over 10, the level is HIGH. The criteria for the level of severity is an arbitrary choice of our system, which can be easily changed if we find this is not appropriate. For example for the SO2, 965 email notification are only sent if the level of severity/notification is HIGH. Initially, we choose a threshold of 10 kt (for a plume high altitude assumed at 15 km). This threshold has been moved to 5kt. For nuclear incident, if an EMERCON message is sent, this is automatically HIGH, there is no LOW level.
The extended plume of hazard is determined by using a lower threshold applied for the neighbour pixels of an alert (see section 4.1.4). For example, if the threshold of an SO2 alert for a sensor, is 2 DU. We can extend the 970 https://doi.org/10.5194/nhess-2021-105 Preprint. Discussion started: 4 May 2021 c Author(s) 2021. CC BY 4.0 License. plume with the neighbouring pixels with lower SO2 values (e.g. up to 0.5 DU). Detailed of the plume extension is provide in the metadata.
About the information of source of the airborne hazard. For gamma radiation, this information is generally provided in the EMERCON message. For alert related to volcanic activity, VONA messages are ingested by EUNADICS EWS, and used for determining the name of the erupting volcano, source of the detected volcanic 975 emission. We also use information from Volcano Discovery (https://www.volcanodiscovery.com) and MIROVA systems (https://www.mirovaweb.it) to highlight the activity of world wild volcanoes. If there is no active volcano provided by VONA messages (or other system) close to a volcanic plume detected, our system tries to determine if there is a most likely candidate volcano which can be identified as the source of the emission. If the ID (identity) of a source is successful (60% of the highest values are located close to the same volcano), our NCAP file provides 980 all the information about this volcano, i.e. its name, latitude, longitude, elevation, type (e.g. stratovolcano), country, rock (e.g. basaltic / andesite basaltic). This information comes from the Global Volcanism Program (https://volcano.si.edu), and a link to GVP webpage of the identified volcano is provide in the NCAP file. The ID of the source is sometimes wrong (generally when the plume is far from the source), and we still need to investigate this aspect to avoid as much as possible ID errors. The use of a constellation of several satellites is a way to avoid 985 this problem. Indeed, the pool of active alert pixel is defined by a volcano ID and should be able to keep the good ID, even for a plume detected far from its source (what we call the traceability of an event, in time and space).
For the other natural hazard, we plan to use information of the nature of the neighbouring ground using ESA-CCI Land Cover information. This is not yet considered by EUNADICS EWS.
In case of an on-going airborne hazard event affecting the European air space, EUNADICS EWS creates link to 990 quicklook images in the NCAP files (e.g. links to E-PROFILE, with quicklooks of aerosol, cloud and wind observations from Automatic Lidars and Ceilometers; see section 4.2 for more details about the available observations from EU and regional ground-based networks).
The generic name of a NetCDF Alert product (NCAP) is the following: 995 XXX_YYYYMMDDHHMM_yyyymmddhhmm_ZZZ_SENSOR.nc -LNK for an alert product linked to a previous SNH alert product -END for an alert product ending an event (file is empty; this is issued of no more alert products are linked 26 hours after the last LNK products) SENSOR (from 3 to 11 digits) to refer to the name of sensors (or ground-based network) used for issuing the alert. 1020