We're sorry but this page doesn't work properly without JavaScript enabled. Please enable it to continue.
Feedback

Using Sentinel 2 images to quantify agricultural encroachment in Burkina Faso’s protected livestock reserves

00:00

Formal Metadata

Title
Using Sentinel 2 images to quantify agricultural encroachment in Burkina Faso’s protected livestock reserves
Title of Series
Number of Parts
351
Author
License
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Identifiers
Publisher
Release Date
Language
Production Year2022

Content Metadata

Subject Area
Genre
Abstract
Using Sentinel 2 images to quantify agricultural encroachment in Burkina Faso’s protected livestock reserves In many parts of Burkina Faso, competition over land use has increased tensions and often conflicts between farming and herding communities. Allocating land for farming or grazing is increasingly perceived as a zero-sum calculation among these communities. As a response, the government of Burkina Faso created “Pastoral Zones” across the country as reserves for livestock herders where animals could graze without the risk of entering cropland. Farming in these areas is typically prohibited unless done by herders residing within the reserve. However, farms have appeared in pastoral zones over the years, reducing resources available to herders and exacerbating already fraught tensions between herding and farming communities (Nébie et al 2019). This study uses Sentinel 2 imagery to quantify to what extent agricultural growth is encroaching on two such pastoral zones in Southern Burkina Faso, Niassa and Sondré-Est. This study found a significant growth of agricultural cultivation in both zones between the period of 2016 and 2021. To map agricultural growth, Sentinel 2 imagery was used in Google Earth Engine (GEE). Reproducibility and accessibility were prioritized, hence the use of a free platform and open EO data was prioritised. Google Earth Engine stood out as an accessible cloud platform to easily access the imagery and run the analysis (Gorelick et al, 2017). To visualise agricultural areas, the “3 Period Timescan” (3PTS) Method was employed. This method uses a series of NDVI Images from the Sentinel 2 satellite throughout a growing season to isolate areas of active cultivation. This product consists of a Red-Green-Blue composite of Sentinel-2 Images where the red band represents the maximum NDVI value during the first period of the growing season, the green the maximum NDVI in the middle, and the blue the maximum NDVI at the end. As a result, the method is able to create a seasonal time-series profile of NDVI. A single NDVI product provides an indication of vegetation presence on a given date, but it is not sufficient to distinguish croplands from other types of vegetation. Croplands are thus identified by their temporal evolution of NDVI values throughout the different phases of the agricultural season: photosynthetic activity of crops is low during the planting period (“beginning of the season”, approximated by 15th June to 1st August), increases during the growing phase (“middle”, 2nd August to 1st September) until reaching a maximum value right before the harvest; once harvested, NDVI values decrease drastically (“end of season”, 2nd September to 15th October). Thus, the approach employed for investigating cropland change considers maximum NDVI values for those three separate subperiods of the agricultural season and aggregates this information into a higher-level product, a RGB color composite so-called 3-Period TimeScan, reflecting the vegetation temporal evolution during the agricultural period, at 10m resolution (Boudinaud and Orenstein, 2021). 3PTS images allow for a user-friendly method to visually identify cropland. Cropland pixels from 3PTS images, when visualized in GEE appear in a dark blue due to the sharp changes from the 2nd and 3rd periods of the time series. This contrasts well with natural vegetation, which has a smoother temporal profile with a noticeable peak in the 2nd period and thus appears greener or a much lighter blue. Forests, due to their high NDVI values throughout the entire growing season appear in white, due to the saturation of all 3 bands. Bare soil, with it’s low NDVI values throughout all 3 periods appears as nearly black pixels. Rather than machine learning, visual identification was the preferred method of identification due to the relatively small size of each pastoral zone. The time needed to prepare training data and clean the results of a supervised classification would have exceeded the time to manually identify each area of cropland. As a result, once the images were treated by GEE, they were manually traced within QGIS. The 3PTS script, originally made for GEE was then translated to run in PyQGIS. Once run, the script created a raster image for each year’s growing season in the archive (2016-2021) and polygons were traced over each visualised cluster of cropland. The total surface area of all polygons was then calculated for each year. A github repository contains both the PyQGIS and GEE code and can be run with no prerequisites. The results of the study indicate a significant increase in cultivation in both zones between 2016 and 2021. For Sondré Est, this change amounted to 40% and 160% for Niassa.Curiously, the largest increase in cultivation seems to occur between 2016 and 2017. This is especially so for Niassa. Nonetheless, increases in cultivation increased with each passing year until the present year of 2021. A number of these fields are suspected to be encroachments, given their proximity to the border of the zone and that many are contiguous with the agricultural fields outside of the zone’s borders. However, it is estimated that a number of the fields are the result of the zones’ resident herders planting fodder or other cereals. The latter assumption is made based on the location of the fields in question (far from the borders of the reserves) and their proximity to permanent structures in the reserves (habitations, wells or park buildings).
Keywords
Google EarthTrailInformation securityPhysical systemSource codeSystem programmingPressureMobile WebTime zoneSurfaceWater vaporIntrusion detection systemFrequencyForestType theoryNatural numberLevel (video gaming)Physical systemMusical ensembleMobile WebTrailRevision controlFrequencyType theoryObservational studyBitPressureCuboidOrder (biology)Physical lawCorrespondence (mathematics)Port scannerVisualization (computer graphics)Line (geometry)Product (business)Natural numberProjective planeMedical imagingMultiplication signForestMereologyRepresentation (politics)QuicksortSign (mathematics)Electronic signatureDifferent (Kate Ryan album)AverageRight angleEstimationDependent and independent variablesElectronic mailing listComputer programmingPixelCovering spaceComputer animationDiagram
Google EarthProduct (business)FrequencyDigital filterMaxima and minimaComa BerenicesComputer-generated imageryVideo game consoleTask (computing)Open sourceResultantPresentation of a groupSoftwareBitVirtual machineFrequencyBelegleserMultiplication signComputer animation
Mobile appGoogolGoogle EarthKeyboard shortcutReading (process)CodeBitQuicksortDifferent (Kate Ryan album)MereologyComputer animation
Mobile appCodeGoogolGoogle EarthKeyboard shortcutSatellite2 (number)Right angleReduction of orderSpacetimeComputer animation
Mobile appGoogolGoogle EarthKeyboard shortcutAssembly languageReading (process)SatelliteCodeRevision controlPower (physics)Visualization (computer graphics)QuicksortTelecommunicationMathematicsComputer animation
Type theorySample (statistics)Time zoneComputer-generated imagerySparse matrixType theoryMultiplication signDifferent (Kate Ryan album)Sampling (statistics)ForestPixelSeries (mathematics)Electronic signatureGraph (mathematics)Slide ruleMedical imagingSparse matrixQuicksortTime seriesRight angleComputer animation
Time zoneComputer-generated imageryBoundary value problemThermal expansionSurfaceSource codeScripting languageSlide ruleRight angleGradientField (computer science)Medical imagingPlug-in (computing)Order (biology)QuicksortTerm (mathematics)Virtual machineTracing (software)BelegleserThermal expansionFrequencyMultilaterationLevel (video gaming)Visualization (computer graphics)EstimationResultantMachine learningComputer animation
Computer-generated imageryTime zoneSource codeKörper <Algebra>Boundary value problemThermal expansionCASE <Informatik>Field (computer science)Multiplication signKörper <Algebra>Medical imagingRight angleNatural numberDifferent (Kate Ryan album)Computer animation
CodeInformation securitySlide ruleHypermediaSurjective functionRevision controlMobile appCASE <Informatik>CodeMathematicsGroup actionComputer animation
Electronic design automationGoogolGraphical user interfaceUniform resource locatorNumerical digitCohen's kappaSatelliteMobile appMobile appComputer animationXML
FrequencyVisualization (computer graphics)CalculationAreaParameter (computer programming)Computer-generated imageryGreen's functionData bufferMobile appPixelLevel (video gaming)Keyboard shortcutSlide ruleFrequencyRight angleComputer animation
Parameter (computer programming)FrequencyGreen's functionComputer-generated imageryData bufferMobile appCalculationVisualization (computer graphics)PixelAreaFile archiverScripting languageSlide ruleRight angleComputer animation
Parameter (computer programming)YouTubeMobile appFrequencyAreaCalculationVisualization (computer graphics)PixelComputer-generated imageryGreen's functionData bufferCohen's kappaGoogolGoogle EarthTrailSlide ruleComputer animation
FrequencyData bufferWordPort scannerCodeMultiplication signSoftware repositoryMobile appTwitterMappingFeedbackUniform resource locatorComputer clusterComputer animation
Transcript: English(auto-generated)
Thanks, everyone. So today I'm going to be talking about a paper that co-authored with Suleiman Jupp on using Google Earth Engine to track agricultural encroachment in Burkina Faso, specifically in what we call pastoral zones, or what we might know as livestock reserves.
So to give a little bit of background about what we're talking about here, the first thing is to understand pastoralism in Burkina Faso. When we talk about pastoralism, we're talking about a mobile system of livestock herding. This means that livestock herders don't stay in one place for the whole year.
They move their animals by the season in search of pastures as the seasons change, moving from the rains into the dry season. And in particular, we're looking at Burkina Faso, so right here in this red box. This is a map of what these transhumous movements might look like.
As you can see, it's a diverse map. You have long movements that seem to stretch across the entire band of the Sahel Sahara, and then you've also got small movements, which might be taken over the course of less than a season. So all that's to say is we're looking at a very heterogeneous type of movement system.
Now, in the news over the past 10, 15 years has been a lot of discussion about clashes between farmers and herders, between more or less sedentary agriculturalists and mobile livestock herders. This has been growing as we've had more land use conflicts over the years.
Now in response, the Burkina Bay government created pastoral zones partly to maybe settle some of these mobile pastoralists and in others to guarantee access to pasture, right, is that a lot of the conflict happens over deciding land use between pasture and agriculture. So these pastoral reserves were created.
Now what this project is going to be looking at is are these pastoral reserves actually looking at grazing, are they actually serving grazing, or are they actually being used as well for agriculture? One of the common narratives, actually the most common narrative about these conflicts
between farmers and herders is that it's these mobile herders encroaching on sedentary agriculturalists. It's a very simplified narrative and like most simplified narratives, if you take it alone, it's bullshit. And I think that what we're going to try and do is hopefully this study will use some empirical evidence to show that it's actually more complicated like that and that this
encroachment, as we like to call it, can go both ways. This was a research project that was funded by SNV Burkina Faso as part of a wider project to understand land use pressures on pastoral systems. So okay, now that we've gone over the background, where are the places we're
looking at? We're looking at two pastoral zones in particular, Niassa and Sondraest. Both of these zones are located in the south of the country and are home to several thousand pastoralists. So on the left, you can actually see a land use map, which I think gives a pretty interesting illustration of what we're looking at.
So surrounding the two pastoral zones that you can see shaded with a line, with a line fill, is cropland. So what's in the legend known as Agricol? Within the fill, you can see that it's Tapirbasse, which is grassland.
So you can see that these are small islands of grassland within a sea of cropland that goes right up to the border of the zone. As you can imagine, this puts significant pressure on that land. So again, the big question, are these zones being fully used for pasture, as is
indicated by the laws in their creation, or are they being used for other purposes? And are they being used by pastoralists, or are they being used by other communities? Which is the big money question. So in order to figure this out, we used a methodology originally developed by
L'Orboudinot at WFP, a world food program, called the three-period time scan. Now what this does is it creates a pretty simple RGB image corresponding to three periods of the agricultural season. The red, which is the first band, is the beginning of the growing season. So about the first month and a half, starting from the sowing period to the
corresponding to the sowing period. The middle is the green band, which corresponds to the second month of the season. And then the end, so the harvest, is blue. And what's actually pretty cool about this is in the Sahel, this corresponds
really well to the growth signature of rain-fed agriculture. So if you look here on this map, you can see an example of what that looks like. All of that dark blue is active cropland. The lighter shaded around it is natural vegetation.
Because what you can see, this chart here that shows, for that exact pixel, that's blue, you can see where the RGB is generated, right? Each of those NDVI values corresponds to a band. And you can see how it contrasts to the NDVI signature of the zone, right? That's the generalized average.
So to give you a better idea, this allows us to do is it allows us to signify different land use types very easily, very visually, right? So the RGB signature of each land use type is going to be different across the season. Agriculture starts out low.
Understandably, the sowing period, cropland is bare, and then it peaks in the third part. Woody cover, or forest in general, is going to be high throughout the growing season, which makes sense. Whereas natural vegetation in the Sahel will have a peak in the middle, corresponding to about a week or two weeks right after the peak of rainfall,
and then it goes back down as biomass begins to dry out. And you can see sort of what the RGB will look like for each of these different values, for agriculture, for forest, for natural. So you can start to see how this quite simple product that we build in Google Earth Engine is able to very quickly make a visual representation of active cropland.
We use Google Earth Engine because of simplicity of use, because while it is not open source in the, while it's not an open source software, it is free and easy to access. And rather than make a virtual machine somewhere, we found that this was the easiest way to be able to not just show the results,
but actually help other researchers to reproduce the results. At the end of the presentation, I'll talk a little bit about how it's being used elsewhere and some of the other uses that the three-period time scanners had. So here is a sort of a, this is a little bit of a gift to show you,
once we've loaded into Google Earth Engine, the differences you can see. So this is in the southern part of Mali, sorry, southern center of Mali. And you can see active cropland being changed from 2016 to 2019 in just a second.
You can see a reduction, right? You can see that these very blue spaces are being taken over by these lighter natural vegetation. This corresponds to cropland abandonment due to violent conflict. And again, I think it shows the sort of power of this visualization and how easy it is to communicate changes in cropland.
So how do we apply it to pastoral reserves? So this is, remember there were two zones we were looking at, Sondreest and Niassa. Right here, we have the zone of Niassa. And you can see all of that cropland around it,
which is signified by the second, time series, not time series, the second series on the graph here. Now, we've done the same thing as I showed you in an earlier slide where we just took samples of different land use types and mapped the NDVI signature across the growing season.
Now, it's a little bit different than the earlier one because when you look at the Sahel, you're looking at a very agroecologically heterogeneous zone. Grassland, rather than peaking in the middle, actually sort of peaks at the beginning, but gradually drops. Cropland, again, similarly, very low at the beginning, peaks at the end.
Forest remains high throughout. Sparse vegetation remains unsurprisingly sparse. And this allowed us to sort of get a general idea of, okay, well, what is cropland looking like here? And so this is an image from 2016, and you can see all of these cropland pixels completely surrounding the zone.
Now, the results showed a significant level of intrusion of agriculture from neighboring zones and also a growth of what we believe to be locally produced agriculture within the zones.
So these are these two three-period time scan images. Probably come back to these, but I would like to just show you a visualization that's a little easier to read. So on the left, we have Niassa, and on the right, we have Sondreest. So Sondreest, you can see the lighter shade is 2016 cropland, and the darker is the expansion in 2021.
Now, it's not insignificant. There's a large expansion of cropland. However, we found that these were in the center of the zone, which seemed to indicate, as we'll show in later slides, that this was actually the pastoralists themselves growing these fields. It wouldn't really make sense for an encroaching agricultural community
to come and build, come and make fields right outside the habitation of the zone's inhabitants. Niassa, though, shows a much different story. You can see there's a sort of gradient moving north to south. And as we'll show in a later slide, a lot of these fields are contiguous
with the cropland in the neighboring zone, which seems to indicate that this isn't actually being used for pasture, but this is growth of the existing cropland into the agriculture, into the pastoral zone. So we used a QGIS to calculate the growth.
In order to do that, what we did was we wrote a script using the Google Earth Engine plugin for QGIS, and we basically took the Google Earth Engine script we wrote, and we translated it into PyQGIS, which allowed us to visualize the images in QGIS and honestly manually trace them.
This wasn't really possible with machine learning because so much cleaning had to be done that it was actually just faster to trace it ourselves given how small the zone was. Niassa, as we showed you, really had an astronomical growth of cropland within the zone.
I mean, just absolutely astounding, going from pretty negligible in 2016 to 2021, something like 140% growth. Sondreest, again, the growth was quite large, but again, in relative terms, not quite as shocking.
So to provide a little more evidence on what we mean by the difference between fields planted locally by the zone's residents and the encroachment, on the left is an example in Sondreest. So we've got a big aerial image of some fields in the center. And is this a... here we go.
So you can see here, these are actual habitations. The fields meet pretty closely. Whereas on the left, sorry, on the right, again, you can see the contiguous nature, right? You can see that the cropland that's in the zone oftentimes connects to fields nearby.
And we found this was the case throughout the border. So what does this mean? First of all, this methodology is reproducible. We've used it for cases in humanitarian action to look at food insecurity. So measuring, giving a very basic idea
of cropland changes from one year to the next. And it's been used already in multiple countries in the South for a couple of years now. Thematically, the thing that I find interesting is that this challenges the dominant narrative on these farmer-herder conflicts, which is this idea of an invasion of herders onto farmland.
It's a narrative that you can find being parroted in a lot of media in the US and Europe when we talk about these conflicts. So we have evidence to show that it's much more complicated than that, that it's not a simply one-sided case of one side encroaching on another, but there's actually encroachments happening on both ends.
And I guess importantly to all of the geeks in the room is we do have the code and the app available. Let me see if I can actually click here. On the next slide, actually, we have... Whoa! Can I have that back, actually?
Oh, that's cool. Oh, wonderful. Thank you. So we do have an app that we've loaded into Google Earth Engine that allows you to basically select anywhere you want. I don't have a keyboard right now, so I'm just going to go here, right? And basically you can go anywhere in the world.
You can adjust the dates if you want, so you can use your own period. And basically recreate this map. The URL was in the previous slide. And yeah, basically you can add it. There's also a GitHub if you want to use the PyQGIS script.
That was again on the previous slide. You can do the entire Sentinel-2 archive, right? So you can go 2016 to 2021, which is the most recent growing season, and we're going to be keeping updating it every year with each growing season. So the growing season for West Africa will end in October,
and so we will have new images ready for 2022, ready for easy use in 2022. In October of this year. Would it be possible to go back to the slide, actually? Thank you. So yeah, you can use this.
You can really do whatever you want with it. We certainly don't mind. Oh, excellent. Thank you. There we go. And so yeah, you have... I created a tiny URL, time scan app, all one word if you want to use the app. Time scan repo if you want the GitHub.
And or you can email me, alex.dacart.com, or find me on Twitter. Always happy to talk about cows and maps, or just, you know, again, please, if you want to look at the app or the code and you have any feedback, we would absolutely love to hear it. We're always happy to learn what kind of improvements we can do.
And yeah, I think that's all from me. Thank you very much.