OBSERVER: How Copernicus Sentinels monitor the snow and ice extent over EuropeCSO Tanya Walker
Thu, 18/03/2021 - 11:14
On the 11th of January 2021 Storm Filomena hit Spain, blanketing almost half of the country with a layer of snow measuring up to 40 cm. The Madrid region was particularly impacted with a shutdown of economic activities for several days due to the very low temperatures (close to -10 °C) prohibiting the snow from melting over this period. This exceptional snowfall was observed from space by Sentinel-2 and was processed to quantitative information by the recently launched High Resolution Snow and Ice monitoring product (HR-S&I) by the Copernicus Land Monitoring Service (CLMS). This enabled the snow cover to be monitored with a very high spatial resolution for several days providing accurate information on its evolution. These high-quality observations are available throughout Europe in near real time (NRT). They aim at improving time critical applications such as weather forecasting, avalanche bulletins and precise hydrological models.
Five days after Storm Filomena, this image shows snow covering most of the Madrid metropolitan area. Snow cover fraction maps on 11 and 18 January from CLMS (credit: © European Union, Copernicus Land Monitoring Service 2021, European Environment Agency (EEA))
Snow and ice extent: essential climate variables
Snow cover and lake ice extent have both been specified by the Global Observing System for Climate (GCOS) as part of the 50 essential climate variables (ECVs) to be monitored by satellite remote sensing. They are relevant parameters for a wide range of applications due to their strong influence on the water cycle and surface energy fluxes, both for private sector and scientific institutions.
Snow cover and ice extent monitoring has growing socio-economic interests related to weather forecast, climate change and hydrology but also to road maintenance and winter tourism. Many stakeholders are also interested in the observation and modelling of water stocks in the form of snow for hydropower generation and water management. Snow and ice maps also offer significant benefits in the environmental field, for forestry and biodiversity conservation. They are essential to national and local institutions for the assessment of natural hazards and managing the associated risks, such as avalanches, ice jams that disturb inland navigation as well as severe flooding due to ice occurrences on the hydrological network.
The importance of the Sentinel-2 constellation
CLMS already provides a set of medium to low resolution Snow and Ice monitoring products for the northern hemisphere. With almost daily coverages, it addresses crucial requirements in climatology, meteorology, and large-scale water management applications. To broaden EO based Snow and Ice applications down to regional and local scales, the CLMS service has now extended its Snow and Ice monitoring capabilities using the Sentinel-2 constellation. The Sentinel-2 constellation is made up of two identical satellites (Sentinel-2A and Sentinel-2B). Together, they orbit Earth at an altitude of 786 km but 180° apart, a configuration which optimises coverage and revisit times. They freely provide high quality optical images of the very same location every five days at the equator under cloud-free conditions which results in 2-3 days at European latitudes. Due to the overlap between Sentinel-2A and 2B swaths from adjacent orbits, this revisit frequency increases under different viewing conditions. Each satellite carries a camera that images Earth’s surface in 13 spectral bands. These instruments are characterised by high spatial resolution (10m for visible, near-infrared bands and 20m for red edge and 60m shortwave infrared bands) which allow unprecedented observations capabilities. Once clouds are masked out, it is possible to detect the presence of snow and ice over lands and inland waters.
Introducing Europe's eye on snow and ice cover
After intensive consultations with experts and users, the Pan-European High-Resolution Snow & Ice products (HR-S&I) have been operationally available to users since July 2020. The HR S&I time series reaches back to September 2016 and is continuously updated with new images in near real time (NRT), that is, not later than 12 hours after sensing in nominal circumstances.
Observing and monitoring snow cover
For optical remote-sensing, the detection of snow over land mainly relies on accurate correction for the atmosphere and identification of clouds. The HR-S&I monitoring service is based on pre-existing Research & Development algorithms and products conducted by the French laboratory CESBIO (Centre d'Etudes Spatiales de la Biosphère) and supported by CNES (French National Centre for Space Studies) for snow products.
After the atmospheric correction and the cloud masking, the algorithm detects the presence of snow pixels within each Sentinel-2 image (level L2A) and then further computes the snow cover fraction per pixel. The final product is a quantified Fractional Snow Cover (FSC) map that provides for each 20m x 20m pixel the fraction (0% – 100%) of its surface covered by snow at the top of the canopy (FSC-TOC). Further the FSC on ground (FSC-OG) is computed, where the snow laying below the forest canopy is estimated using the CLMS product High Resolution Layer Tree Cover Density.
Early snowfall in the French Pyrenées captured by Sentinel-2 - 27 September 2020 - FSC product (credit: © European Union, Copernicus Land Monitoring Service 2021, European Environment Agency (EEA), S.Gascoin)
The next product available as part of the HR-S&I monitoring product is the Persistent Snow Area (PSA). PSA is derived from the FSC information over a hydrological year and provides the extent of persistent snow cover, i.e. the area where snow is continuously present throughout a hydrological year. The PSA layer provides, at a very high spatial detail, information which is important for research, climatology, and water management in a context of growing pressure on the hydrologic resources.
Persistent snow area is assessed from the fraction snow cover information - here we compare the yearly snow extent in the Alps, near the Bernina range for the hydrological years 2017 and 2020. Black polygons represent glacier outlines from the Randolph Glacier Inventory (credit: © European Union, Copernicus Land Monitoring Service 2021, European Environment Agency (EEA), Copernicus Climate Change Service)
Observing and monitoring ice extent over rivers and lakes
Besides snow, the HR S&I project also implements other Sentinel-2 based products which address ice phenomena over the entire hydrological network of Europe. The HR S&I products are implemented by the Polish company Astri-Polska which were already providing national ice monitoring services. The first Ice product is the River and Lake Ice Extent (RLIE), it enables users and public authorities to monitor at the spatial detail of 20m x 20m and in NRT the ice cover across Europe’s rivers, lakes and channels, providing important input data for ice breaking activities to keep waterways viable or anticipate the risks for ice jams. HR S&I products are also meant to address and monitor the viability of frozen rivers and lakes in Nordic countries, particularly Finland.
Ice episode near the Wloclawek hydropower plant (Poland - 10 February 2017) (credit: © European Union, Copernicus Land Monitoring Service 2021, European Environment Agency (EEA))
The RLIE product highlights river and lakes pixels that are covered with snow-covered or snow-free ice (clear blue). Open water appears in dark blue while other features as soil or vegetation within the river bed or lake table look red in the image. Nisser, Fyresvatn and Nesvatn Lakes, Norway - April 1st, 2019 (credit: © European Union, Copernicus Land Monitoring Service 2021, European Environment Agency (EEA))
In analogy with the snow monitoring products FSC and PSA, ice products also come in a combination of low, and high-level datasets. As such, the Aggregated River and Lake Ice Extent (ARLIE) is derived from the RLIE by grouping its pixels into 10 km segments of larger rivers and of entire lakes. The ARLIE product is still evolving, and once finalised later in 2021, it will provide aggregated information as a service and in a tabular format. This will considerably simplify the use of ARLIE data, since users do not require handling of spatial data.
Access to the Snow and Ice products
Time-critical applications such as avalanche bulletins and weather forecasting require snow and ice data within a minimum time lag after sensing by Sentinel-2. The maximum acceptable latency time for these applications is 12 hours after sensing while 6 hours is the ideal user requirement. As a result, every effort is made to produce and publish NRT HR-S&I products within 12 hours after Sentinel-2 data are available. Since its operational start-up, the HR-S&I monitoring service has demonstrated its ability to generate these products within the time limits required by users.
The maximum acceptable latency time for applications such as avalanche bulletins, weather forecasting, etc. is 12 hours. In February 2021, 80% of NRT HR-S&I products (i.e. FSC and RLIE products) were published within 12 hours after measurement (credit ©: Magellium)
The HR-S&I products are available for browsing and downloading on the WEkEO DIAS (Data and Information Access Services). HR S&I is the first Copernicus Land product fully implemented and accessible on WEkEO. This allows centralised access to Copernicus Earth observation data, and scale processing resources in a way to meet the high demand on the timeliness.
Machine to machine data access mechanisms allow efficient data flows, essential for the integration of HR-S&I products into external operational downstream services.
Will the HR-S&I monitoring service be extended?
After nearly one year in operation, the HR-S&I products will continue to be improved and extended based on user needs. In the medium term, improvements to the algorithms for detecting clouds, snow and ice will mostly be made in upstream R&D projects and will then be turned into HR-S&I operational products when required by users.
In the more immediate future, the HR S&I part two will extend the product portfolio with active radar data from Sentinel-1. This will enable all-weather, day and night time observations to be brought to the service as well as the current optical based data to be completed. Thanks to radar technology the HR S&I product portfolio will also be extended by snow conditions (wet or dry snow) which will give users the ability to track snow melting information across Europe with very high spatial details. The CLMS, under the European Environment Agency (EEA) delegation, expects multisensor snow and ice observations at a much higher frequency will help it to get as close as possible to the users’ needs. This extension is expected to enter into operation in the third quarter of 2021. The result of which it is hoped will continue to bring timely, reliable and high quality information to a range of users and stakeholders for their ongoing requirements and in particular in the monitoring of the effect of climate change and the assessment of natural hazards and events like Storm Filomena.
Further information and developments regarding the High Resolution Snow and Ice Monitoring products by the Copernicus Land Monitoring Service (CLMS) can be found here.
Thu, 18/03/2021 - 12:00