Wildfires in California spotted by Sentinel-2A

Wildfires were raging in California this summer. Let's see if we can spot some of them in Sentinel-2A images with the new Sentinel-Playground by Sinergise.

 

The combination of the SWIR and NIR bands of Sentinel-2 or Landsat enables to produce accurate maps of burnt areas. The SWIR band is sensitive to the water content in the soil and vegetation, while the NIR band is sensitive to the vegetation health (photosynthetic activity).

 

In addition, the radiance measured by a spaceborne sensor in the SWIR wavelengths increases if the surface is very hot (as taught us Prof. Planck in Hawaïï).

 

As a result, a simple color composite of bands SWIR/NIR/Red gives a stunning view of burnt areas and can highlight ongoing fire areas if the smoke is not too opaque.

 

In Coalinga, the Mineral Fire burnt nearly 3000 ha close to the city of Coalinga.

Mineral wildfire near Coalinga, Californa USA. Time series of Sentinel-2A images (RGB color composite of bands SWIR/NIR/Red = 12/8A/4)

 

Continue reading

Sentinel-2A and Landsat-8 team up to track Kilauea Volcano's lava flow

=>

 

Kilauea Volcano in the largest active volcano in Hawaii and one of the most active on Earth. It has been erupting for over 30 years now but hit the headlines recently because a large lava flow traveled up to the ocean, adding 2 brand new hectares to the Hawaiian Islands.

Aerial view of the 61G lava flow ocean entry on August 19, 2016. Credit: U.S. Geological Survey Department of the Interior/USGS U.S. Geological Survey.

Continue reading

Kittens time series

=>

Optical remote sensing is great to map the snow cover extent in mountain regions as long as there is no cloud above the land surface. Radar remote sensing of the snow cover is not operational yet mainly because the backscatter from the snow surface is strongly dependent on the snowpack liquid water content. On the ground, however, thousands of people are observing the snow cover in the mountains, everyday. Some of them take photographs and kindly upload them to photo-sharing websites with a public license. Many of these photos are geotagged, either because the cameras have built-in GPS, or because the users added geographical coordinates when publishing their album.
Continue reading

Take5 goes to the movies

=>

How to go from 1 image every 5 days to 24 images per second ?

It was possible, thanks to CNES funding, thanks to an imaginative producer, Gérard Dedieu (who does not smoke cigars yet), thanks to a talented film director and scenarist, Thierry Gentet (the only film director who understands space mechanics), and thanks to his team, Mira Production, who are even able to shoot beautiful images in our  ... splendid CESBIO offices, and thanks to a series of promising actors and actresses Anne Jacquin, Valérie Demarez, Virginie Lafon, Valery Gond, Jean-Pierre Dedieu, and another one, the last one, who cannot say a full sentence before the 5th take.

 

We hope this little film will help you understand or explain the possibilites and opportunities offered by multi-temporal images at a high resolution, and that it will give you ideas to use the new SPOT5 (Take5) data.

 

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

Land cover maps quickly obtained using SPOT4 (Take5) data for the Sudmipy site

=>

At CESBIO, we are developing land cover map production techniques, for high resolution image time series, similar to those which will soon be provided by Venµs and Sentinel-2. As soon as the SPOT4 (Take5) data were available over our study area (Sudmipy site in South West France), we decided to assess our processing chains on those data sets. The first results were quickly presented during Take5 user's meeting which was held last October.

1. Experiments

In this post we describe the work carried out in order to produce these first land cover classifications with the SPOT4 (Take5) Sudmipy images (East and West areas) and we compare the results obtained over the common region to these two areas.

 

Prior to the work presented here, we organized a field data collection campaign which was synchronous to the satellite acquisitions. These data are needed to train the classifier training and validate the classification. The field work was conducted in 3 study areas (figure 1) which were visited 6 times between February and September 2013, and corresponded to a total of 2000 agricultural plots. This allowed to monitor the cultural cycle of Winter crops, Summer crops and their irrigation attribute, grasslands, forests and bulit-up areas. The final nomenclature consists in 16 land cover classes.

 

The goal was to assess the results of a classification using limited field data in terms of quantity but also in terms of spatial spread. We wanted also to check whether the East and West SPOT4 (Take5) tracks could be merged. To this end, we used the field data collected on the common area of the two tracks (in pink on the figure) and 5 level 2A images for each track acquired with a one day shift.

 

OUEST EST
2013-02-16
2013-02-21
2013-03-03
2013-04-17
2013-06-06
2013-02-17
2013-02-22
2013-03-04
2013-04-13
2013-06-07
2. Results

The first results of supervised SVM classification (using the ORFEO Toolbox) can be considered as very ipromising, since they allow to obtain more than 90% of correctly classified pixels for both the East and the West tracks and since the continuity between the two swaths is excellent. Some confusions can be observed between bare soils or mineral surfaces and Summer crops, but these errors should be reduced by using LANDSAT 8 images acquired during the Summer, when Summer crops will develop.

Merging of the land cover maps obtained on the East and West Sudmipy tracks (the cloudy areas were cropped out). The comparison against the ground truth (the black dots on the map to the South-West of Toulouse) results in a kappa coefficient of 0.89 for the West and 0.92 on the East.

 

West EAST

This zoom compares the results obtained on the common area of the two tracks (West to the left and East to the right). The two classifications were obtained independently, using the same method and the same training data, but with images acquired at different dates and with different viewing angles. The main errors are maize plots labeled as bare soil, which is not surprising, since this crop was just emerging when the last image was acquired. There are also confusions between wheat and barley, but even on the field, one has to be a specialist to tell them apart.


3. Feedback and retrospective

After performing these experiments, we were very satisfied with the operationnality of our tools. Given the data volume to be processed (about 10 GB of images) we could have expected very long computation times or a limitation in terms of memory limits of the software used (after all, we are just scientists in a lab!). You will not be surprised to know that our processing chains are based on Orfeo Toolbox. More precisely, the core of the chain uses the applications provided with OTB for supervised training and image classification. One just have to build a multi-channel image were each channel is a classification feature (reflectances, NDVI, etc.) and provide a vector data (a shapefile, for instance) containing the training (and validation) data. Then, a command line for the training (see the end of this post) and another one for the classification (idem) are enough.

Computation times are very interesting: several minutes for the training and several tens of minutes for the classification. One big advantage of OTB applications is that they automatically use all the available processors automatically (our server has 24 cores, but any off the shelf PC has between 4 and 12 cores nowadays!).

We are going to continue using these data, since we have other field data which are better spread over the area. This should allow us to obtain even better results. We will also use the Summer LANDSAT 8 images in order to avoid the above-mentioned errors on Summer crops.

4. Command line examples

We start by building a multi-channel image with the SPOT4 (Take5) data, not accounting for the cloud masks in this example :

otbcli_ConcatenateImages -il SPOT4_HRVIR_XS_20130217_N1_TUILE_CSudmipyE.TIF
SPOT4_HRVIR_XS_20130222_N1_TUILE_CSudmipyE.TIF
SPOT4_HRVIR_XS_20130304_N1_TUILE_CSudmipyE.TIF
SPOT4_HRVIR_XS_20130413_N1_TUILE_CSudmipyE.TIF
SPOT4_HRVIR_XS_20130607_N1_TUILE_CSudmipyE.TIF -out
otbConcatImg_Spot4_Take5_5dat2013.tif

We compute the statistics of the images in order to normalize the features :

otbcli_ComputeImagesStatistics -il otbConcatImg_Spot4_Take5_5dat2013.tif -out
EstimateImageStatistics_Take5_5dat2013.xml

We train a SVM with an RBF (Gaussian) kernel :

otbcli_TrainSVMImagesClassifier -io.il otbConcatImg_Spot4_Take5_5dat2013.tif
-io.vd DT2013_Take5_CNES_1002_Erod_Perm_Dissolve16cl.shp -sample.vfn "Class"
-io.imstat EstimateImageStatistics_Take5_5dat2013.xml -svm.opt 1 -svm.k rbf
-io.out svmModel_Take5Est_5dat2013_train6.svm

And Voilà !, we perform the classification:

otbcli_ImageSVMClassifier -in otbConcatImg_Spot4_Take5_5dat2013.tif -mask
EmpriseTake5_CnesAll.tif -imstat EstimateImageStatistics_Take5_5dat2013.xml
-svm svmModel_Take5Est_5dat2013_train_6.svm -out ClasSVMTake5_5dat_16cl_6.tif

What about playing Take Five again with SPOT-5 ?

=>

This post is regularly updated with news (the official date of Sentinel-2 launch) or to add new arguments.


SPOT5 will soon end its career. After 12 years of image acquisitions, the satellite will retire in 2015. CNES might launch a call for scientific experiments with SPOT5 before the satellite de-orbitation. I am quite sure it would be useful to repeat the SPOT4 (Take5) experiment, for the following reasons :

 

  • The official Launch date of Sentinel-2A in now March the 30th, 2015. Even if the satellite is launched in time, the routine acquisition on the 10 day cycle orbit will start one or two months later, and it will take several months to distribute data operationally. Data from the second Sentinel-2 satellite will probably only be available the year after, we will thus have to wait until 2016 to get data with a 5 days repetitivity.
  • A new Take5 experiment based on SPOT5 would allow to go on preparing uses, methods and applications based on time series, to get ready to make an operational use of Sentinel-2 just after the data are released.
  • SPOT5 provides multispectral images with a 10 m resolution, just like Sentinel-2. A new Take5 experiment would allow better simulations of Sentinel-2 data.
  • SPOT4 (Take5) experiment was held in Spring and ended in June, we might this time try to extend the time period towards the Summer to monitor the summer crops.

For the first presentation of SPOT4(Take5) proposal, I had been told that it had no chance to succeed, and I had even used this drawing as my last slide. SPOT5 (Take5) chances of success are the same, but isn't it worth trying ?

  • The SPOT4(Take5) experiment was set up very quickly and we lacked time to convince international partners to take part to the experiment. This time, thanks to SPOT4(take5). Now, thanks to SPOT4(Take5) little celebrity, it should be easier to involve new international partners, and to reach new users. The cost for international partners to get access to data for one site every fifth day during five months was about 4000€. Please contact me if you think your organism might participate.
  • SPOT4 (Take5) sites were chosen very quickly, and many users complained they had no time to set-up ambitious measurement campaigns, to hire people for ground truth measurements and define protocols. This time, we might obtain a longer prior notice period to set things up properly.
  • The dreadful weather we had in Europe during SPOT4(take5), nearly spoiled a few experiments, but we may hope that it would not happen again if we tried it one more time.
  • ESA and JRC excepted, we did not have time to involve European partners in SPOT4 (Take5), and 95% of the sites chosen by ESA and JRC were outside Europe. This time, we could focus part of the experiment on Europe.
  • We might try to have the experiment running for a longer period that in the first time, but it will mean a  larger cost for CNES, and we willl need to have a convincing set of experiments to convince CNES.

Even if SPOT4(Take5) was a success, we will need to build an excellent proposal in order to convince CNES, in a constrained funding context. After the experiment was already funded once, it is not a premiere anymore, and its impact will be less straightforward.

 

We thus need to compensate with original ideas and a large support.  If you are interested to participate to a possible SPOT5 (Take5), please leave messages on this blog or on my email, Please do not forget to provide us the results you obtained with SPOT4(Take5).

 

Feedback on the irrigation scheduling experiment using remote sensing images

=>

CESBIO contributes to an international joint laboratory in Morocco, called TREMA, "Télédétection et Ressources en Eau en Méditerranée semi-Aride", which means "Remote Sensing and Water Resources in Semi-Arid Mediterranean". This year, this laboratory has embarked on an ambitious experiment of irrigation scheduling by satellite imagery, on a wheat plot near Marrakech. This experiment was already described in March, and it gave very promising results.


The main objective of the experiment was to see if  the logistics of irrigation scheduling by water balance model were feasible in real conditions. For this, a farmer accepted to play our game on two four hectares plots of wheat: Irrigation of the reference plot was driven by the farmer in the usual way. The test plot irrigations was driven by our tool SAMIR (FAO-56 model forced by satellite imagery).

 

Since the sowing late December to the harvest in early June, a weather station installed on a reference culture has given us the daily reference evapo-transpiration measurements. On the other hand, to control a posteriori the quality of our estimates of water requirements for irrigation, two flux measuring stations were set up. We also acquired a series of images SPOT5 early in the season to compensate for the slightly late start of SPOT4 experience (TAKE5) which began in February.

 

In addition to a clear weather throughout the season, we were able to benefit from the excellent work of the SPOT4 (TAKE5) team which provided us with the georeferenced images very quickly. The NDVI evolutions were thus available in a relatively short time. As an end user, the Office of Agricultural Haouz allowed us to perform the irrigation of the test plot in the best conditions while being subjected to the constraints of the canal system.

On the ground, everything did not work as well as we planned. Following a misunderstanding with the farmer, we completely missed the second irrigation and the fertilizer application was not timely. Indeed, the study plot is installed on a heavy clay soil that forms a crust. We were not aware that, a few days after sowing, a specific irrigation is needed to ease the emergence of plants. On the other hand, the farmer applied nitrogen fertilizer on two plots just after irrigation of the reference parcel and relatively far from the irrigation of the test plot. Under these conditions the nitrogen is relatively less soluble, and our test plot lacked fertilizers.

Our experiment has been seriously hampered by the misunderstandings with the farmer. But despite the bad start, the experiment was pursued to its end.

 

This plot shows the changes throughout the course of the experiment of the water supply from rainfall and irrigation, the evapo-transpiration ETobs measured in the field and the Evapo-Transpiration ET estimated by SAMIR model, using the vegetation status from SPOT4 (Take5) images. On this plot, the dates of irrigation were suggested by the model.

 

To our surprise, the results are extremely promising. Indeed, despite a 20% lower biomass compared to the plot driven by the farmer, we got a equivalent performance in grain yield. This can be explained by the fact that, although the average number of wheat blades was much lower on the test plot, it is very likely that the reference plot, irrigated by the traditional method, has suffered water stress in late March limiting the filling of grain.

 

 

This full-scale experiment finally turned out to be very instructive. First,  imaging/weather/irrigation logistics worked great : the weather data transmission, the reception and the geometric and radiometric correction of images, the model runs and  irrigation decision were largely automated. The SPOT4 (Take5) data, that prefigure those of Sentinel-2, proved perfectly suited to this application. Unfortunately, the clay crust has severely limited the emergence of culture. Yet this phenomenon, well-known to our farmer, taught us to cultivate humility ;-) , and we will consider the introduction of the risk in a decision support system. Finally, the functional constraints of the gravity irrigation system have taught us that our tool should be more flexible to recommend an irrigation period instead of a single date, and that we should link the service to weather forecasts.

 

Following this experiment, we started developing a Web service (SAT-IRR) that should shortly provide the essential functions of an irrigation decision support with a simplified interface.