As part of the Copernicus programme, Sentinel-3 is a very important mission for ocean colour, sea and land surface, fire, atmosphere and climate purposes. With the two platforms, Sentinel-3 A and B, an optimal global coverage will be now obtained. The next observations and related products are promising!
Check the diverse pictures of this fantastic launch via the Twitter accounts of EUMETSAT and ESA
The current Sentinel-3 services provided by EUMETSAT here
And watch below the movie from EUMETSAT in live in the Sentinel-3 control centre “Sentinel-3: Operating satellites” with Hillary Wilson, EUMETSAT’s Sentinel-3 manager, and Kevin Marston, EUMETSAT’s System operation manager.
Since July, the province of British Columbia (BC), west coast of Canada, has been impacted by massive and violent wildfires. Such events are visually impressive. They can be a constructive force by maintaining the overall health and functioning of the forest ecosystem, but also a destructive force due to devastating impacts on the local population, soil and our atmosphere: e.g. visibility reduction, air quality deterioration caused by smoke fine particles (aerosols) that are harmful for health population and on a longer term, climate.
These effects can easily be seen by different “eyes”. While BC is known for its clean, fresh air, the visibility reduction occurs as the increased concentration of smoke particles leads to more extinction of the sunlight passing through the atmosphere. This results in the formation of haze and the dark grey smoke that we see. The particles can also attract water, acting as condensation nuclei for droplet formation and a subsequent further reduction in visibility as fog and clouds form.
Although the fires themselves are fairly localized in the interior of BC, their overall impact is not, with the smoke traveling far away from the source region – up to thousands of kilometers! Atmospheric and fire emission models are used to predict the amount and trajectory of smoke according to the prevailing circulation and dispersion patterns allowing for air quality forecasts and warnings, such as the Government of Canada’s wildfire smoke prediction system: FireWork.
Satellite observations are vital to monitor such disasters. Not only do they provide an overview image from the top of the atmosphere of the raging fires, which already gives spectacular maps, they also allow detection of the substances released in the air by these episodes and follow their dispersion. This is vital for predicting the impact of air masses far away from the fire sources. Below are some illustrations of these satellite maps over the last 10 days.
Optical images like MODIS sensors (on-board Terra & Aqua platforms) capture the thick smokes directly linked to the fires: such plumes can extent to several hundreds thousands of kilometres. Moreover, measurements in thermal infrared spectrum allow detection of the actively burning areas.
The integrated amounts of aerosol particles that are suspended in the atmosphere can be quantified, and their horizontal distribution monitored. This is an important input for air quality models in charge of predicting the air quality for populations.
Knowing the type of these particles (i.e. whether they are dark or brighter) is of great importance for scientists and researchers. This directly gives insights on how these aerosols affect our atmosphere radiation, and consequently our climate on the long-term. Satellite sensors measuring light in the UV such as the Dutch-Finnish OMI mission, on-board the NASA Aura platform, provides with an important index (here named UV Absorbing Index or UVAI) to identify absorbing (i.e. very dark) aerosol particles on a daily-global coverage.
Finally, wildfires do not only release smoke and particles, which already pose a problem for human respiratory health, but also a mix of toxic gases. One of the most important is CO – Carbon monoxide. Infrared sensors such as the European IASI mission, on-board Metop-A and B, quantify everyday the amount of CO present in the atmosphere. On some specific days in August, concentration values were comparable to those associated with African and South-American fires.
This post was written by Julien Chimot and my colleague Jonathan Izett. Find more information on Jonathan via his Linkedin profile, and his website on his research work focused on the formation, evolution and prediction of fog in the atmospheric boundary layer.
Since mid-July, huge wildfires have raged across south-eastern France, forcing more than 10,000 residents and holidaymakers to flee homes and campsites overnight. Some fires also occurred in the northern part of Corsica. Hundreds of firefighters have been deployed but the battles remain difficult and challenging due to strong winds, dry areas and extreme warm weather.
These impressive fires were spotted by the American Modis-Aqua satellite through the dispersed smoke particles on the last days. The burning areas are not only located within the forests but also next to the coasts and some famous villages (e.g. Saint-Tropez and Bormes-Les-mimosas in the Var department).
Evacuated people are forced to sleep on the beach of Bormes-les-Mimosas in the Var region in southern France. Fires often ravage the French Riviera but this year the area is experiencing an exceptionally hot and dry summer, making it particularly vulnerable. Moreover, current investigations suspect malice causes in this event. The blazes have destroyed brush and vegetation in about 4,000 hectares (15 square miles) of land along France’s Mediterranean coast.
Vesc 2017 fieldwork has come to an end! A lot of works was achieved by the 2nd year Bsc TU Delft students with their supervisors.
Some numbers to summarize: 48 students working during 3 weeks in France (Vercors & Drome region) + some more back in Delft on analyzing the geological formations and Sentinel-2 satellite data, 16 land cover final classification maps produced, many more maps analysing vegetation seasonal cycles and the impacts of cloud contamination, more than 1600 GPS waypoints collected for identifying reference land cover classes, 16 GPS elevation maps compared with satellite remote sensing, 32 final reports delivered (Acquisition and Integration reports)!
Back in Delft after the Vesc 2017 FieldWork performed in France, Bsc TU Deft students have been very busy with final processing of Sentinel-2 datasets, interpretation and evaluating their land cover performances. Many analyses have been done to find out how much information could be extracted and how much clouds, observation spatial resolutions, instrument defaults et many other elements could affect their results.
Among other things, first seasonal analyses have been achieved on Sentinel-2 surface reflectance between 2016 and 2017. Although this work is preliminary, and we likely need to improve our methodologies and tools in the future, nice results were shown.
All the groups, with their different approaches and land cover tools, managed to detect and observe, from remotely, vegetation cycles in France, through the seasons, with Sentinel-2 sensor. In particular, it is very impressive to see how healthy and very green biomass (e.g. deciduous trees and grasses) grow up from the end of Winter to Summer and start disappearing again starting mid-Autumn. From mid-Autumn to end of Winter, a lot of classes more related to bare soils start appearing. Although the given names (or identifications) are likely not true, they well represent land covers with very little to null vegetation present at their surfaces.
A great work achieved in a short time by TU Delft Bsc students! Looking forward to learn and discover more in the next years!
Previous (2014-2015-2016) field-campaigns in Vesc, France, land covers classification with LandSat-8 here
Performing some false colour composites, by combining different spectral bands of Sentinel-2, will give us some first hints about the different land covers, their spatial distribution and to define some reference classes.
The picture below depicts 3 types of false colour composites in this region:
the first one is the “natural colour” band combination (RGB 4-3-2)
the second is “standard false colour” composite (RGB 8-4-3), a very popular band combination useful for vegetation studies
and the third provides a “natural-like” rendition (RGB 12-8-3), popular for geology, agriculture and wetland studies.
What do all these colours tell us about our region? Which land covers seem to be easy to distinguish from the Sentinel-2 observations?
Stay tuned! We should find the answers within the next weeks with our students!
PS: a small cloud is hidden somewhere in all these images. Are you able to find it?
Previous (2014-2015-2016) field-campaigns in Vesc, France, land covers classification with LandSat-8 here
This year, like the last 3 years, we go back again to France, next to the Vercors region, for an exiting field-campaign with the TU Delft Bachelor 2nd year Applied Earth Sciences students!
Coordinated by Dr. J.C. (Jan Kees) Blom & Dr. Roderik Lindenbergh, the students will have 2 key goals during this field-campaign: a geology goal (geology excursion, analyses of the geological formation, etc…), and a remote sensing goal. I will be there assisting the students with the data, the technology tools, and preparing their reports for this last goal.
The remote sensing goal is about performing a land cover classification based on optical spectral measurements acquired by satellites. GPS way points will be used to locate and identify reference land cover spectral signatures. The last years, we exploited the LandSat 8 data. But this year, for the first time, we will work with the very new Sentinel-2 reflectance measurements.
This is a good opportunity for the students to discover what satellite instruments measure, how we can interpret these measurements, and which parameters we have to pay carefully attention. A good introduction for further studies on remote sensing if they are interested in the future!
Among all the questions that we will have to investigate with the students?
How does each type of surface reflect the Sunlight in each spectral band (visible, infrared…)?
Which land covers may be easily distinguished in a Sentinel-2 data?
What does Sentinel-2 indicate us about the vegetation type, and its evolution?
How to compute the spectral signatures of a river? Deciduous / coniferous forest? Farmlands? Cities? Agriculture? Bare soils?
How to define a reference land cover? How to locate it w.r.t. Sentinel-2 pixel?
How to create a supervised land cover classification tool? And a prior training site?
Everything is ready now! In 1 week, we start this campaign! And in about 4 weeks, we should have altogether performed a Sentinel-2 land cover classification of the Vesc & Vercors area!