Archives pour l'étiquette spring 2016


Sentinel 2 to monitor forest fires in Siberia?

On the 20th of July, a colleague in Igarka (Maxime Deschuyteneer) informed me of a forest fire near Igarka that could be responsible for “a small greenhouse effect that make you cough”…Indeed, forest fires are recurring problems in Central Siberia, mainly during June and July because of a sharp increase of temperatures. These fires have widely increased last years and the year 2016 would be the most “blazed” of history since, according to GreenPeace Russia, 3.5 million ha of forests have been burnt1 (seven French departments!). On 24th September, NASA has also published an Aqua MODIS scene from 18th September 2016 showing huge plumes moving towards North East of Russia as well as zones (in red) where the satellite has detected unusual warm temperatures associated with fire2. The extent of MODIS image on the map of concentrations of aerosols allows presenting the scale of the phenomenon.


(at the top)- NASA’s Aqua satellite scene (MODIS) showing huge plumes2; (at the bottom) – Map of concentration of aerosols2

I used the amazing tool Sentinel Playground to verify if the fire described by Maxime was captured by Sentinel 2 (S2) and especially if it was possible to monitor its spread and its damages. Done! The figure below shows 8 S2 acquisitions between on 16/07 and 04/09 from only several kilometers up to Igarka. Thanks to a false color composite where the band 12 (the thermal channel, SWIR) is placed in the red channel, forest areas affected by the fire can be clearly distinguished. By carefully observing, the flames on the front line of the fire are even visible on 19/07, 23/07 and 26/07. Despite some clouds, this time series allows learning a little more on burned areas and the spread of fire.


Sentinel 2 time series of the forest fire in Igarka (67°28’19”N, 86°33’31”E)

After roughly delineating the affected zone, we have performed a supervised classification based on a tree decision algorithm (CART3) for each S2 scene. The number of training samples varies at each date since the goal was to get as the more accurate classification as possible and not necessarily providing an operational processing chain for this type of work. Only the S2 scene from 22/08 was not selected for analysis because of too much hazing effects. Thus, the figure below illustrates classification results for each date where burnt/not burnt forest areas have been computed and reported under each of them. We can see that the fire has widely spread in a 10 days period after its starting (about 19/07) where burnt areas have doubled between on 19/07 and 23/07 and then between on 23/07 and 26/07. Furthermore, by approximately drawing a line on the front of the fire, we can say that the fire has moved from North to South at an average speed of 10m/h during the first period (19/07-23/07) and 17m/h in the second one (23/07-26/07). In a second stage (from 05/08), the fire has stopped advancing but encircled zones have continued to burn. It is visible on 08/08 and 04/09. Nonetheless, classification results on 23/07 are imperfects because of cloud shadows effects and some areas in the south were not classified as “burnt” whereas they were on 19/07. Therefore, burnt areas are probably under-estimated at this date.


Classification results of burnt and not burnt forest areas based on a supervised tree decision algorithm (CART)

This is a good example of the various possibilities offered by Sentinel 2, namely for monitoring fast natural phenomena in sparsely populated regions (for example, the peak flow of the Yenisei river). If some readers are better qualified than we are on this topic and want to pursue this work, the script GEE is available here and we remain available for any questions. See you for new S2 observations in Igarka! (And thank you to Simon Gascoin for his help in the implementation of the script GEE).

3Breiman, L., Friedman, J.H., Olshen, R.A., and Stone, C.I. (1984). Classification and regression trees. Belmont, Calif.: Wadsworth.


June 10: Calm on Yenisei is coming back…New field works are coming!

30 minutes ago, Théo (the historical sampler of the Yenisei) tooks the hovercraft for the Igarka airport to go back in France.
10th of June 2016, 10h30, Spring continues to settle in Igarka and gives to the Yenisei banks greenish colors that would almost make us think to “La prairie des Filtres” (special mention for my friends in Toulouse).
The flood peak has now passed (05/06) and we sampled each day the last week for ensuring to have a maximum amount of data on this so particular period.
Yenisei waters were black, full of organic matter and it was an amazing spectacle to observe remaining ice breaks from the branch of Igarka leaving to Arctic ocean.
From now, we reduce the field sampling frequency to one or two times per week but others field works are coming. Igarka will also serve as a calibration site for the SMOS (Soil Moisture and Ocean Salinity) satellite whose the mission is driven for part by a team of CESBIO (Centre d’Etude Spatiale de la BIOsphère) in Toulouse.
SMOS aims to characterize soil moisture across the Earth all 3 days with a spatial resolution of ~ 40-50 km and already provides different products for users working on this topic.
You can visit the SMOS blog at this address: HERE.
Arctic region soils are critical for SMOS since their strong organic matter concentrations need to adapt the calibration of the delivered products by this satellite.
Next week, the TOMCAR team will take care of installing two humidity sensors that, we hope, will allow to improve the SMOS products acquired in Arctic regions.
Additional news on satellite activities in the region of Igarka are coming, namely about the DOC monitoring in the Yenisei.
Allez les bleus 😉


June 2: Back on the water

Today was the first sample collected by boat of the 2016 mission.
We had some trouble to access the river bank, because of deep remains of snow we got the car caught, but we finally did it with a second car!
To get in the sampling zone we had to cross slowly deriving icebergs.
The water is now full of color, fDOM values have almost doubled since the last sample 4 days ago.
Navigation is already quite dynamic on the Yenisey, with commercial boats unloading goods for Igarka shops and huge barges carrying containers and materials crossing even northern.


June 1: First boat !

First boat arrived yesterday!
People are unloading goods going back and forth with hovercraft.
The channel of the Yenisey is clearing progressively, there are still icebergs deriving and « ice jam » may occur again when the remaining ice from upstream islands and branches will break. Safe sampling need to wait a little as moving icebergs might be several times the size of our boat.

CIMG9942 2

May 28 : Wider, thiner, faster

Second sampling with the hovercraft today.
Cracks are multiplying and getting wider, ice surface has patchy brown colors and is getting thiner while the water level is rising faster and faster.
Ice blocks have started to climb on the river bank in some places and first compression crests are growing little by little here and there.

CIMG9862 2

May 25 : Still on the ice but…

We are still sampling on the ice but yesterday the snowmobile and the trail were surfing!
In the sampling zone there is a thin layer of ice covering about 40 cm of melted snow on top of a thick ice layer of about 80 cm.
Seagull are arrived which is, according to locals, a sign that the ice break is coming.
Fishermen have brought back their cabines on the earth and are now fishing close to the river bank.
fDOM values are around the levels of the days just before the ice break of the previous years.
The water is now rising quite steadily (our landmark, a old stranded boat is disappearing).