Didi Abuli (დიდი აბული) vu par Sentinel2-A

Le Didi Abuli domine la chaîne de montagne du petit Caucase en Géorgie. Ce secteur très peu fréquenté situé à proximité de la frontière Arménienne offre un terrain de jeu exceptionnel pour le ski de randonnée. Ainsi sur les traces des copains de la Timuzapata,  avec 2 collègues du Cesbio et 1 de l'ONERA, nous sommes allés arpenter les pentes de ce volcan endormi qui culmine à près de 3300m mais aussi découvrir l'hospitalité des habitants de cette région qui mérite vraiment de s'y attarder.
 
Oui mais quel rapport avec Sentinel2?

Trace de l'ascension du Didi Abuli-Sentinel-2A-2017-06-03

Continue reading

La neige du Haut-Atlas marocain vue par Sentinel-2 et MODIS

Il était une fois, un jeune thésard qui se promenait dans les montagnes enneigées du Haut-Atlas avec un groupe d’amis. Pendant que ces derniers s’amusaient à prendre des selfies avec la neige il pensait à un autre type de selfies ! Il pensait aux photos de la Terre prises par les satellites : ces images refléteront-elles la beauté de cet endroit ? Leur résolution spatiale permettra-t elle de reproduire les subtilités de l’enneigement dans ce massif escarpé ?

 

Photo du Haut Atlas marocain prise à Oukaimeden. L’enneigement varie fortement selon la pente et l’orientation du terrain.

Continue reading

Patagonian skies are not cloudy anymore

"The most usual weather in these latitudes is a fresh wind between north west and south west with a cloudy overcast sky" - Phillip Parker King, Sailing Directions for the Coasts of Eastern and Western Patagonia (1832).

 

Patagonia is a beautiful place to visit but campers know that the weather is extremely variable and the sky is often cloudy. This can be a problem for glaciologists, too, since they rely on optical satellite imagery to study glacier area changes over the last decades (mainly Landsat). Clear-sky optical images can also be used to determine glacier velocity, albedo, front variations, etc.
Continue reading

Sentinel-2A captures a giant ice avalanche in Tibet

The Nature News website reported yesterday on a massive ice avalanche that happened in Rutog, Tibet, on 17 July 2016. This ice avalanche killed 9 people and may be one the largest ever observed. The ice and rock mixture spread over 6 km from the collapse point up to the Aru Co lake shoreline.

Sentinel-2A image of the Rutog ice avalanche acquired on 21-Jul-2016 (4 days after the event). Click on the image to see at full resolution (1 pixel = 10m).

Continue reading

First Sentinel-2 snow map

=>

 

In the framework of the THEIA land data center, we have developed a simple but robust method to map the snow cover from Sentinel-2-like level 2A products. This code was tested with SPOT-4 Take-5 and Landsat-8 series, but it remained to adapt it so that it can run on real Sentinel-2 images! This is now done thanks to Manuel Grizonnet, which allowed us to process the Sentinel-2A image acquired on 06-July-2015 in the Pyrenees as a first example. This image was produced at level 2A by Olivier Hagolle using the MACCS processor. The snow mask from Sentinel-2 images is calculated at 20 m resolution after resampling the green and red bands that are originally at 10 m resolution while the NIR band is at 20 m.

How to make sure everything went well? We can control the snow mask by superposing the mask boundaries on a false color composite:

 

The Sentinel-2A image of 06-July-2015 (level 2A, tile 30TYN) and its snow mask. The snow mask is in magenta and the background image is a color composite RGB NIR/Red/Green. We also show a zoom in the Vignemale area.

Continue reading

Premier masque de neige Sentinel-2

=>

Dans le cadre du Centre d'expertise scientifique THEIA "surface enneigée" nous avons développé une méthode simple et robuste pour détecter la neige à haute-resolution à partir des produits de niveau 2A de type Sentinel-2. Ce code a été testé sur des séries SPOT-4 Take-5 et Landsat-8, mais il restait à l'adapter pour qu'il puisse tourner sur de vraies images Sentinel-2 ! C'est chose faite grâce à Manuel Grizonnet, ce qui nous a permis de traiter l'image Sentinel-2A du 06-juillet-2015 sur les Pyrénées. Cette image avait été produite au niveau 2A par Olivier Hagolle avec la chaine MACCS. Le masque de neige est calculé à 20 m de résolution après ré-échantillonnage des bandes vertes et rouges qui sont d'origine à 10 m de résolution alors que la bande MIR est à 20 m. Continuer à lire

SAT-IRR: Satellite for Irrigation Scheduling

=>

Following the irrigation scheduling experiment in Morocco during the SPOT4-Take5 experiment (Le Page et al, 2014), a Web tool owing to help the irrigation decision making is under development (http://osr-cesbio.ups-tlse.fr/Satirr). The tool is functional on three Landsat8 tiles: Marrakech in Morocco, Kairouan in Tunisia, Toulouse, France.

 

As the tool is addressing irrigators, the idea is to set an irrigated plot of the simplest and fastest possible way. After drawing his plot on a base map, the user answers to 4 questions. He chose his culture among 7 options currently parameterized (corn, wheat, olive ...), its soil among the 12 USDA typical soils, the sowing date and irrigation method (flooding, sprinkler or drip). This rough initialization is adequate launch the service although, at any time, the user can change the plot contours or refine parameterization if he knows the soils, the peculiarities of its crop, etc.

satirr_screenshot

Screenshot from SAT-IRR web interface. The four icons allow modifying the plot parameters and contours, input irrigation, and consulting the results as graphs or tables. The graph results show a small Openlayers window with the last NDVI image, the sequence of NDVI thumbnail images, and 4 graphics: The “atmospheric part (rainfall, Reference Evapotranspiration and actual evapotranspiration), the second graph shows the status of the soil water content separated in three layers, the third graph shows the evolution of Basal Crop Coefficient and Fraction Cover, and the last graph is NDVI. The blue square at the right of the graphs are projections for the next month, including the green bars which are irrigation recommendations

Initially the server makes an approximation of an average behavior of the plant. For this, a monthly climatology is compiled (multi-annual average of weather parameters) and then interpolated to daily values, while the average behavior of the plant is extracted from FAO-56 Tables "FAO Irrigation and Drainage No. 56 : Guidelines for Computing Crop Water Requirements" (Allen et al, 1998). In a second step, the satellite images already on the server are examined and the relationship between NDVI and Basal Crop coefficient (Kcb) and percent of ground covered by vegetation (Fraction cover, Fc) are determined at each date by averaging on the plot.
Past weather is populated by measurements on the nearest synoptic station of the World Meteorological Organization and synthesized in the form of daily reference evapotranspiration (ET0) and rainfall. Forecasts are obtained by the API of the Norwegian Meteorological Institute.
Finally, a water balance very close to the one described in the FAO-56 is calculated by combining typical crop behavior and climate, satellite imagery, weather data and forecasts and projection into the future of crop development. The goal is to propose a date and dose of irrigation.

 

In addition to updating the weather (measures and forecasts), the server will check every day for the availability of new images (only Landsat8 for the time being). If there is availability, the tile is downloaded, it is then corrected for atmospheric effects using the information provided by the nearest photometer from the Aeronet network using the SMAC code (Rahman & Dedieu, 1994), then a cloud mask is created and NDVI is calculated. This image is stored as the original file is discarded to not overload the server.
All parameterization / measures / prediction are stored in a postgres / postgis database that links with a web client interface. The user can view the results in tables or graphs, and add its own irrigation in another dedicated interface.
While the interface is still a little rough, we are essentially considering developments on the server side:

  • Adaptation to Sentinel-2: the transition to S2 should not be a hassle. However, it will be necessary to adjust the calculation of the tiles to download, the download code and format reading.
  • Use of Sentinel-1: In the current state, the well-performance of water balance is based on the actual information of irrigation provided by the user. We plan to test the use of S1 images to determine the irrigation dates.
  • Access to local agro-weather stations: As part of the development of the Environmental Information System in Cesbio, telemetry of several weather stations has been settled up (eg, see http://trema.ucam.ac .ma) (Jarlan et al, 2015), we have to make these stations accessible through a standardized web service like Sensor Web.
  • Introduction of collective irrigation network. The PhD work of Kharrou (2013) and Belaqziz (2013, 2014) have shown that remote sensing can be used to optimize water rotations of an irrigated command. We plan to offer the possibility of introducing a set of plots to associate it with a distribution network and ultimately offer an optimized arrangement of the water rotation. However, at present, this goal is more into the order of a challenge!
  • We are currently working on a procedure to introduce wheat yield using remote sensing data (J. Toumi PhD Thesis) and further expect to input an early wheat yield prediction into the tool.

If you want to try it out, be my guest, it's free. If you want to try it out on other regions, please contact me!

 

References:

  1. Le Page M., J. Toumi, S. Khabba, O. Hagolle, A. Tavernier, M. Kharrou, S. Er-Raki, M. Huc, M. Kasbani, A. Moutamanni, M. Yousfi, and L. Jarlan, “A Life-Size and Near Real-Time Test of Irrigation Scheduling with a Sentinel-2 Like Time Series (SPOT4-Take5) in Morocco,” Remote Sens., vol. 6, no. 11, pp. 11182–11203, Nov. 2014.
  2. Allen R., L. Pereira, D. Raes, and M. Smith, FAO Irrigation and Drainage n°56: Guidelines for Computing Crop Water Requirements, no. 56. FAO, 1998, pp. 273–282.
  3. Rahman H. and G. Dedieu, “SMAC: a simplified method for the atmospheric correction of satellite measurements in the solar spectrum,” Int. J. Remote Sens., vol. 15, no. 1, pp. 123–143, 1994.
  4. Kharrou M.H., M. Le Page, A. Chehbouni, V. Simonneaux, S. Er-Raki, L. Jarlan, L. Ouzine, S. Khabba, and A. Chehbouni, “Assessment of Equity and Adequacy of Water Delivery in Irrigation Systems Using Remote Sensing-Based Indicators in Semi-Arid Region, Morocco,” Water Resour. Manag., vol. 27, no. 13, pp. 4697–4714, Sep. 2013.
  5. Belaqziz S., S. Mangiarotti, M. Le Page, S. Khabba, S. Er-Raki, T. Agouti, L. Drapeau, M. H. Kharrou, M. El Adnani, and L. Jarlan, “Irrigation scheduling of a classical gravity network based on the Covariance Matrix Adaptation – Evolutionary Strategy algorithm,” Comput. Electron. Agric., vol. 102, pp. 64–72, Mar. 2014.
  6. Belaqziz S., S. Khabba, S. Er-Raki, L. Jarlan, M. Le Page, M. H. Kharrou, M. El Adnani, and A. Chehbouni, “A new irrigation priority index based on remote sensing data for assessing the networks irrigation scheduling,” Agric. Water Manag., vol. 119, pp. 1–9, Mar. 2013.
  7. Jarlan L., S. Khabba, S. Er-raki, M. Le Page et al, “Remote sensing of water resources in semi-arid Mediterranean basins: The Joint International Laboratory TREMA,” Int. J. Remote Sens., vol. (under review), 2015.

2015

=>

Dear reader,  may 2015 bring you health and happyness !


2015 should be a great year for the time series of optical images at a high resolution, with the launch of Sentinel 2A in Mat, in less than four months ! We have been waiting for it for so long that we can hardly believe it will be launched so soon. And to help you wait until then. SPOT5 orbit will be changed, to observe a hundred of sites with a 5 day repetitivity similar to that of Sentinel-2, until the end of August.

 

This post is also an opportunity to recall what was achieved in 2014 :

  • In February, we started the Sen2Agri project for l'ESA, with our partners, UCL, CS-SI France and CS-SI Romania. This project which is keeping us busy busy aims at producing a system for automatic classification of crops at the scale of whole countries.
  • In March, THEIA released a new version of SPOT4 (Take5) data, in particular with an enhanced geometrical superposition. This data set has now been downloaded by more than 600 users.
  • In May, the production in near real time (with a delay of 15 to 20 days) of LANDSAT 8 L2A products over France started (this type of product is still not available at USGS, but not for long, I guess...), followed by the processing of de 3 years of LANDSAT 5 & 7 data. 5 years of data, from 2009 to 2014, with an interruption in 2012, after LANDSAT 5 end of life and before LANDSAT 8 launch.
  • In September, CNES finally decided to launch SPOT5 (Take5) experiment,  thanks to a large contribution from ESA. The call for sites proposal was launched by ESA in November, and was largely successful, with 62 proposals for a hundred of sites. The analysis of this proposal is now on-going, and the choice will be difficult as the proposed applications are rich and very diverse.
  • In September too, CESBIO hosted the meeting of THEIA Scientific Expertise Centres, which allowed to identify about 20 products that the centre could prepare in the coming years. These products are described in the recent THEIA bulletin (in French).
  • In November, CNES hosted the SPOT4 (Take5) users meeting, to summarize data use, with a hundred of participants and  23 excellent presentations of the applications allowed by the experiment.
  • In December, we finally launched the call for sites proposal for Venµs mission, which will be launched in 2016. The deadline is January the 28th, 2015.

 

We will have several topics to comment for this blog that starts its third year, with a large audience : it received 23000 visits, and 47000 pages were read, a 50 % increase compared to last year. These weeks, it is getting harder for me to maintain a rhythm of a post per week that I tried to keep in the first two years. Contributions of data users are welcome !

In white, the countries from which no visit was observed in 2014 (Ouganda, Afghanistan)

Liste of the 10 countries from which the most frequent visits are observed.

2015

=>

Chers lecteurs de ce blog, que cette année 2015 vous apporte joie et santé  !


L'année 2015 sera une grande année pour les séries temporelles optiques à haute résolution, avec le lancement de Sentinel-2A début mai, dans moins de quatre mois ! Nous attendons ce satellite depuis si longtemps qu'il est difficile de croire que cette date soit si proche. Et pour nous faire patienter, dès début Avril, l'orbite de SPOT5 sera modifiée, et le satellite observera une centaine de sites, avec une répétitivité de 5 jours identique à celle de Sentinel-2, jusqu'à la fin du mois d'Août.

 

Cet article est aussi l'occasion de se pencher sur ce qui a été réalisé en 2014 :

  • En début d'année, nous avons démarré le projet Sen2AGri pour l'ESA, projet qui depuis nous occupe fortement et dont l'objectif ambitieux est de mettre en place une chaîne de classification automatique des zones agricoles et des types de cultures fonctionnant à l'échelle des pays entiers.
  • Au mois de Mars, THEIA a diffusé une nouvelle version des données SPOT4 (Take5), améliorant notamment la superposition géométrique. Ces données ont maintenant été utilisées par près de 600 utilisateurs.
  • En mai, c'est la production en temps quasi réel (15 jours à un mois) des données LANDSAT 8 de Niveau 2A sur la France qui a démarré (ce type de données n'est pas encore disponible aux USA, mais pas pour longtemps...), suivi du retraitement de 3 ans de données LANDSAT 5 et LANDSAT 7.  5 ans de données sont maintenant disponibles, de 2009 à 2014, avec une interruption en 2012, après la fin de LANDSAT 5 et avant le lancement de LANDSAT 8.
  • En septembre, nous avons obtenu la décision de lancer l'expérience SPOT5 (Take5), grâce à une importante participation financière de l'ESA. L'appel a proposition de sites a été lancé par l'ESA en Novembre, et a connu un beau succès, avec 62 propositions, pour près d'une centaine de sites. L'analyse de ces propositions est en cours et le choix sera difficile, car les applications proposées sont très riches et très diverses.
  • En Septembre aussi, s'est tenue au CESBIO la réunion des Centres d'Expertise Scientifiques de Theia, qui a permis d'identifier une vingtaine de produits différents que le pôle pourrait lancer dans les prochaines années. Ces produits sont résumés dans le dernier bulletin de THEIA.
  • En Novembre, le CNES a accueilli les journées des utilisateurs de SPOT4 (Take5), pour faire le point de l'utilisation des données, avec une centaine de participants et de belles  présentations des applications permises par l'expérience.
  • En décembre, nous avons enfin lancé l'appel à propositions de sites pour Venµs, dont le calendrier de lancement est enfin stabilisé. Vous avez jusqu'au 29 janvier 2015 pour y répondre.

 

Nous aurons donc de nombreux sujets à commenter pour ce blog qui commence sa troisième année, avec un grand nombre de fidèles lecteurs. Le blog a reçu 23000 visites, et 47000 pages ont été consultées, en augmentation de 50 % par rapport à l'an dernier. Depuis quelque temps, j'ai un peu de mal à maintenir le rythme d'un article par semaine que j'essayais de tenir depuis deux ans, mais l'année 2015 devrait être riche en événements. Les contributions des utilisateurs de données sont aussi bienvenues !

En blanc, les pays pour lesquels aucune visite du blog n'a été recensée en 2014 (Ouganda, Afghanistan)

Liste des 10 pays dont proviennent le plus fréquemment les consultations du blog. Ca fait plaisir de voir la France en aussi bonne position dans un classement international

 

Comparison of Level 3A compositing methods

=>

As said in a previous post, we are testing various methods of level 3A production, using SPOT4 (Take 5). The Theia Land Data Center will the use these methods to process Sentinel 2 data. In case you did not click on the link above, let's recall that the level 3A products are monthly composite products of cloud free reflectances. For each pixel, our method computes the weighted average of the reflectances of the dates when the pixel is cloud free. For more details, you will need to follow this link.

 

The work of Mohamed Kadiri at CESBIO, which is funded by the CNES budget for Theia, adressed first the definition of quality indexes for composite products (for more details, may I suggest that you follow this link ?). This work showed that our product has nice performance, but we knew some one would ask us to compare them to the classical methods for level 3A products.

 

Therefore, we compared our product with the famous NDVI Maximum Value Composite (NDVI MVC), developped by our remote sensing ancestors, and used since the most remote antiquity to process AVHRR time series. This method consists in using for each pixel of the level 3A, the reflectances of the date which has the greatest NDVI.  Why ? Mostly because the NDVI of a cloud is very low, often negative, and therefore this method will rather select cloud free pixels. The NDVI MVC comes from a time when the cloud masks were not very accurate.

 

Example of a monthly synthesis obtained with the NDVI MVC methods Example of a monthly synthesis obtained with the weighted average method

This post uses the SPOT4-Take5 data to show a comparison of the performances obtained on the Versailles site, with the NDVI MVC method on the left, and the weighted average on the right. One can clearly see, on the left, the presence artefacts made of whiter and darker dots which are not seen on the image on the right. These artefacts appear when the selected date changes from one pixel to the other. These artefacts are much less visible on the vegetation covered plots, as, for this composite obtained in spring, the vegetation increases quickly, and all the pixels come from the last cloud free date of the synthesis.

 

If we have a look at our quality indicators, which were described in our previous post about composite products , it is obvious that the performances obtained by the weighted average method are much better than those of the NDVI MVC method, either as regards the similarity to the central date image of the Level 3A (in yellow, for the 70 % best pixels and in green for the 95% best pixels), and moreover as regards as the amplitude of artefacts (in blue). The abscissa of the plot is the half of the number of days used in the synthesis, and our recommended value is 21.

 

 

NDVI Maximum Value Composite Weighted Average Composite