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

Effect of water level on bird habitat at lake Maggiore

00:00

Formal Metadata

Title
Effect of water level on bird habitat at lake Maggiore
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
Lake Maggiore and the Ticino River are water bodies shared by Italy and Switzerland: they are important resources for drinking water, irrigation and hydroelectricity generation as well as for tourism and biodiversity. The cross-border character and the often conflicting needs of the different users make the shared management of this resource very complex, but of great importance. The `‘Parchi Verbano Ticino´’ project, funded by Regione Lombardia / EU – INTERREG Italia Svizzera 2014/2020, aims to study the effects of water levels of the lake on various environmental components with a particular focus on protected natural areas. The level of the lake is regulated by a dam located at the southern shore of the lake. In this framework, this study aims to analyse the effect of water level on bird migration by: 1) Calculate the inundated bird habitat using a simulation based on measured water level; 2) calculate the inundated habitat from Sentinel-1 remote sensing imagery 3) Use the flooded area derived from S1 as ground truth to validate the previous simulation 1). The study area is centerend around Bolle di Magadino (Switzerland, 8°51’56.90”E, 46°9’42.17”N, a protected wetland located on the north shore of lake Maggiore at the confluence with the Ticino river. The area is a recognized nesting and stopover site for birds, listed as a Ramsar Wetland of International Importance and as Important Bird and Biodiversity Area (IBA). We defined the habitats of interest using a vegetation map provided by Fondazione Bolle di Magadino. The vegetation types collected from a phytosociological field study were aggregated into ten land cover classes that described the habitat types and land use. The final habitat map covers an extent of 6.7 km², including the 1500 ha of wetland called Bolle di Magadino. Daily passage of migrant birds have been recorded at Magadino ringing station and since 2019, traditional net captures were coupled with an Avian Vertical-looking Radar In this study we focus on the following periods, during which bird monitoring systems were both deployed: P1: 2019-05-01– 2019-06-20; P2 2019-10-01–2020-02-20 and P3 2021-02-01– 2021-07-20. The lake level measured at the hydrological station of Locarno (CH) was used to determine the inundated area in GRASS GIS using ther.fill module (GRASS Development Team, 2022) and a DTM that included the lake bathymetry (cell size 0.5 m). The lake level fluctuated between 192.3 and 194.9 m.a.s.l. over the study period, with a minimum in April and May, when the waters are used to irrigate the rice fields downstream, and a maximum in late autumn. We used the Google Earth Engine Platform (GEE) (Gorelick et al., 2017) to extract Sentinel-1 Synthetic Aperture Radar (SAR) images (ESA, 2021) as they are suited for surface water mapping and not affected by cloud coverage (Ovakoglou et al., 2021). We used Edge Otsu Algorithm with terrain correction (Markert et al., 2020) to estimate the inundated areas of the collection of a total of 236 images for the three time periods when bird migration was also monitored. The calculation was implemented GEE with the approach described by Gorelick et al. (2017) calibrating the threshold for our study area, adapting the code provided by Open Geo Blog (2021). The inundated areas were then overlapped with the land use map in order to estimate the extent of the submerged vegetation over the three time period defined. The resolution of all maps was 10 m, except the DTM that has 0.5 m, the CRS used in this work were WGS84 in GEE and the local CRS GCS CH1903 for all the other analysis. The area covered with water, according to S1, varied between 103.9 and 471.7 ha (210+- 83.6), some of the surface is a permanent wetland so it is never completely dry. Each habitat was affected differently by the flooding: when the water was at its highest, croplands were completely inundated, grasslands and reeds were submerged for 80% of their extent whereas urban areas and infrastructures were not affected (less than 1% underwater). The flooded area calculated by filling the terrain model at the level of the ranged from 130.4 to 248.90 ha (208+- 48.6). The correlation between inundated areas obtained using r.fill in GRASS and S1 on the same dates was fair for P1 and P2, but not for P3, when the water of the lake were taken for irrigation, but the habitats were flooded by rainfall. An interpolation of the flooded from S1 is a more efficient way to obtain an estimate of the flooded habitat on a daily basis, that is necessary to study its effect on migratory birds. The results presented here will contribute to the definition sustainable management tools of water management of lake Maggiore taking into account the effect of lake level on biodiversity in general and on bird habitat in particular.
Keywords
202
Thumbnail
1:16:05
226
242
Sound effectVideo gameGoodness of fitPresentation of a groupGroup actionVideo gameHuman migrationComputer animation
Data managementVolumeSurfaceCone penetration testField (computer science)Data managementProduct (business)Computer animation
Digital photographyData managementFinitary relationSoftware-defined radioAreaWebsiteConservation lawSimulationSineWorkstation <Musikinstrument>Musical ensemblePhysical systemGoogle EarthAlgorithmSeries (mathematics)Grass (card game)Thresholding (image processing)PixelOpen setBlogCodeSatelliteComputer-generated imageryProcess (computing)UsabilityEvent horizonDependent and independent variablesWater vaporData conversionAreaDirection (geometry)Data managementRow (database)Group actionProjective planeLevel (video gaming)Thresholding (image processing)AlgorithmPattern languageState observerWebsiteWeb 2.0Multiplication signMetreMereologySimulationWeb pageTime seriesCodeGrass (card game)Observational studySpring (hydrology)Order (biology)Human migrationDifferent (Kate Ryan album)Physical systemRoutingReverse engineeringHypermediaInformationMedical imagingBasis <Mathematik>Integrated development environmentReal-time operating systemCondition numberUniverse (mathematics)Source codeCalculationMatching (graph theory)Field (computer science)FrequencyResultantImage resolutionComputer animation
Grass (card game)Executive information systemAreaObservational studyGrass (card game)Computer animation
SatellitePairwise comparisonSimulationWater vaporArea
EstimationAreaSpring (hydrology)Type theorySatelliteBounded variationCorrelation and dependenceSign (mathematics)Cross-correlationGrass (card game)Network topologyCodeGoogolScripting languageDigital photographyPhysical systemSpring (hydrology)Field (computer science)SatelliteObservational studyResultantHuman migrationWater vaporMaxima and minimaMatching (graph theory)SimulationTemporal logicCross-correlationImage resolutionNumberAreaGraph (mathematics)FrequencyDifferent (Kate Ryan album)Real-time operating systemCorrelation and dependenceStandard deviationSoftware developerNegative numberExpert systemoutputMobile appMultiplication signSource codeTwitterType theorySeries (mathematics)State observerData managementMathematical analysisDecision theoryPattern languageCodeCross-correlationMotion captureReal numberTime seriesComputer animation
Transcript: English(auto-generated)
Okay, good morning again. Now I am the speaker of this next presentation. I'm Clara Tatone, I am a researcher at the University of Insubria. And today I will present a work carried out together with BirdLife Switzerland and
Fundazione Bolle di Magadino about the effect of water level on migratory birds' habitat at Lake Maggiore. Lake Maggiore is the second largest Italian lake, even though it's not completely Italian. Here we are at Florence, about 500 kilometers north, lays Lake Maggiore,
that is divided between Italy and Switzerland. Lake Maggiore, despite being a natural lake of glacial origin, is regulated by two dams.
One upstreams the Verzasca Dam, who is in Switzerland, and it is a dam mainly devoted to hydropower production and electricity. Downstream on Ticino River, in the Italian territory, we have the Miorina Dam,
who is a dam that regulates the level of the lake, with in mind the rice fields that we have in Italy, that are in the plain downstream of the lake. In 1940 it was supposed to establish an international commission that regulates the water level
and reached an agreement between the two countries. We are still waiting for that. The current management is mainly driven by the Miorina Dam. Not surprisingly, this situation led to water wars,
because of the different interests at stake between the two nations that share the lake. So, in summer, like this one, where we are experiencing a drought, the water level drops because the water is needed for agriculture, so Miorina Dam is open,
and water is needed for electricity, so Verzasca Dam is closed, so the water level lowers, putting at risk even the boat navigation, because the boat cannot moor any longer. On the other hand, in winter and autumn, the problem is the reverse.
Italy is trying to capitalize water for the spring irrigation of the rice field, so the water level rises, and some people have to go to work canoeing.
So, those are examples of the title of newspaper and media. In order to not solve this problem, but at least provide information about this problem, the Parque Verbano Ticino Interreg project was developed,
a project between Switzerland and Italy with many partners involved. The aim of this project is to research about the different needs of people and interest at stake about the water of this lake, and try to find useful guidelines, because I mentioned agriculture,
but there is also the flooding risk, and tourism that are at stake in this project. Today, I'm going to focus on biodiversity, because, of course, water is very important for biodiversity. In fact, Lake Maggiore is on one of the main migratory routes for birds coming from Africa in spring,
and going back to Africa in autumn. And the area called Bolle di Maggadino is a protected area in the Swiss part of the lake at the north, that is a recognized stopover site for birds by the Ramsar conversion and by the bird habitat directive.
This is our study area in large, so for this project, we asked, what is the effect of water level management on the habitat for birds at Bolle di Maggadino?
In order to try to answer this question, we wanted to calculate the flooded habitat when the water level rises or lowers, because the slope is very mild in this area, so even a tiny increment or decrement in the water level can have sensible effect.
To do that, we tried to calculate the inundated area from two sources, the water level measured hourly by the hydrometer near our study area, and by remote sensing using Sentinel-1 imagery.
Then we wanted to compare the area derived from Sentinel-1 and from simulation based on water level, and finally evaluate what effect it has on bird migration. So we have to collect a lot of data. Concerning birds, we have a long time series since 2001 collected with traditional ways,
that is, capturing the birds with nets, marking them with rings, and let them free to fly where they want. Within this project, the University in Subria bought this nice gadget, that is land radar, especially designed for bird observation.
I cannot show web pages in this convention, but if you want to see real-time migration, scan the code and you will connect to our web page, where the radar is outputting continuously the passage of the birds.
Our monitoring times where both systems were deployed were three, two in spring and one in autumn. This is the workflow that we followed for this work.
First of all, I will speak about the simulation done with GRASS GIS with Air Lake package. We took the daily water level from Agrometer, Switzerland, and a DTM with lake bathymetry with 0.5 meter resolution,
and we calculated the inundated area. Then we match it with a habitat map developed by Fondazione Bolle di Magadino from fieldwork, and we were able to have a time series of flooded habitat. On the other hand, we took Sentinel-1 SAR imagery that is not available on a daily basis,
and in Google Earth Engine, we used the Edge OTSU algorithm to calculate the inundated area for the available time of recording. We did the same with the habitat map, and finally, we analyzed how well the time series matched with R.
Finally, we crossed everything with our data about birds. Sentinel-1 is a project from Copernicus. It provides imagery of synthetic aperture radar,
and it works even in every meteorological condition, so we processed over 2,000 images, and we reclassified the inundated habitat. The thresholds were calculated for each of the periods.
Going to the results, we see that the simulation and Sentinel data don't really match. In gray, you have the water level, and in red, the simulated inundated area,
and they match perfectly, of course. SAR has a different pattern that is more linked to rainfall. This is an example. This is our study area, Bolle di Magadino, and this is what happens, according to GRASS, when the water levels rise. You see that the protected area is completely flooded.
For comparison, I'm showing you what happens with the simulation based on satellite imagery. As you see, satellite imagery is able to capture water also inland, like multi-paddles that can develop around the study area.
So, we compared the inundated habitat, and we find that there are significant differences between the methods we used. So, which is the right one?
And which of the methods provides the results that are useful for understanding bird migration? We have to match the data with birds. So, first, we do also some cross-correlations to see what happens between simulated flooded area and satellite imagery
that we suspect is the ground truth. And we found out that the two systems don't match at all in spring when the dam is open for irrigating the rice field.
Whereas, in autumn and winter, there is a better match. So, when the dam is not operational, satellite and water simulation goes well. In spring, when there is rainfall and dam operational, the two systems don't match.
Then we match our results with bird data. And we find some significant temporal correlation with rainfall, which is known in literature and also by common sense that birds don't like to fly when it rains. Who would like to do that? Come on.
And so, we have a negative peak of birds' capture when we have a peak of raining. I'm showing you the results only for one of the periods because they are more alike, and this is spring 2001. Then we found that the number of birds
are correlated with the flooded area calculated by simulation from water level. We also found that there is a correlation that is not shown in this graph
between the inundated water vegetation and the data derived from Sentinel. And this also makes sense because we have also water bird migrating, and when the water level is too low and the water vegetation is dry, birds don't like it.
So, to come to the conclusion, we have found out that the two methods yielded significantly different results for all habitat types. And that is because of the fact of the management of the dam
and the fact that our water level simulation does not include rainfall that is captured by SAR. But SAR is available every two or maximum six days for our study area. That is a two-course resolution because we have daily and even hourly data of bird migration.
And we know that the decision of the birds to stay or live is made not in six days, but in the same day. We observed seasonal patterns and we see that there is a significant agreement between the two methods,
so we have to find a way to include the rainfall in our simulation or to find some other source for the satellite imagery. During spring, as I mentioned, the two methods are completely uncorrelated
and provide opposite trends. And, I didn't mention before, the data derived from satellite I've mentioned has a higher standard deviation.
So, concerning the time series analysis, we found cross-correlation that are significant and positive and are supported by biological data and observation from literature, especially the negative correlation with rainfall
and the negative correlation when the water vegetation is dry. As future development, we wanted to try to find a way
to select one of the two methods or to improve the classification of SAR imagery. And this is also why I'm here because there are a lot of experts in hydrology and I hope to have some good inputs from you. I didn't show here the data from the radar because we are beginning to analyze them
and it's also one of our future developments. So, if you want to reuse our code, it is available. You can find it here. If you want to see immigration in real time, you can scan the MTR code,
thanks to Damiano Prettoni, one of the co-authors who developed these Shiny apps. And I want to thank Francesca Gianetti, who introduced me to Google Earth Engine and Marco Cioli for their support. So, I'm looking forward for your suggestion. Thank you.