European Facility For Airborne Research

DeInVader

DeInVader: Tracing the invasion of an exotic tree species in protected West-Mediterranean dune ecosystems

Scientific project

Detecting and analysing the spread of exotic Acacias in West-Mediterranean biodiverse costal dunes with hyperspectral images and LiDAR

None

GROSSE-STOLTENBERG Andre

PhD Student and Research Assistant at the Institute of Landscape Ecology, Münster University. Expertise in field methods of vegetation ecology including hemispherical photography. Research focus on applying remote sensing (LiDAR, hyperspectral) in assessing biological plant invasions.
English: Fluent
Portuguese: Fluent
French and Spanish: Basic

DO228 - ARF

Invasive species can have a high impact on ecosystem services, biodiversity and ecosystem functioning (Rejmanek et al. 2005, Ehrenfeld 2010). Most recently, the fusion of high resolution hyperspectral and LiDAR data with ecological data has shown to be a powerful tool to detect invasive species and assess its impact in Tropical rainforests. Especially the combination of hyperspectral and structural (LiDAR) data is promising to be the decisive tool to detect an invader that also occurs in the understorey of forest (Asner 2008, Huang and Asner 2009) and thus cannot be easily detected using spectral data alone (e.g. Joshi et al 2006). To my knowledge this proposed project is the first study that combines hyperspectral and LiDAR remote sensing data with such fine spatial and spectral resolution with a variety of ecological ground data to detect an invasive species in Mediterranean open dunes and dune forests.
The principle objective of this project is to map the current spatial distribution of the ecosystem engineer Acacia longifolia in Mediterranean dune ecosystems, and to identify different stages of invasion in different invaded habitat types. This will be the baseline data to scale local impacts (e.g. Marchante et al. 2009, Rascher et al. in press, Werner et al. 2010, and references therein) of the invader up to landscape level.
Desirable outcomes include observing impacts of nutrient enrichment due to Acacia's nitrogen-fixation on surrounding vegetation and soil, and the remote sensing of stable isotope composition (see Wang et al. 2010) of A. longifolia and the surrounding vegetation to reveal impacts on ecosystem processes (see Werner et al. 2010 and references therein). Further important information will be gained on the species and functional type diversity as well as the biomass in the studied area to improve the land management schemes.
Airborne hyperspectral and LiDAR remote sensing data with a fine spectral and spatial (~1m and accordingly >5points/m²) resolution will be collected in the Nature Reserve of Santo André and Sancha, a mediterranean dune ecosystem in South-West Portugal. Field spectra will be collected simultaneously for calibration purposes using an ASD FieldSpec pro. Samples of canopy leaves of A. longifolia and representative native shrubs will be taken and analysed for e.g. water content, Chlorophyll A, nitrogen, and tannin. Acacia populations at different stages of invasion, representative native shrub and tree and of other exotic tree species, and vegetation types will be mapped with a dGPS. Detailed digital distribution maps of protected plant species and vegetation types for further analysis are available.
This project could set the basis for studies on A. longifolia covering larger areas along the Portuguese coast. This project will also facilitate the monitoring of endangered, protected and diverse ecosystems worldwide (Richardson and Kluge 2008, Horus Institute 2005) by identifying areas that are vulnerable to invasions, lead to an early-detection scheme and evaluate Acacia control measures (see Huang and Asner 2009). Future projects could then up-scale even further using satellite data provided by German hyperspectral satellite mission EnMAP starting 2014.
The processed data will be analysed using the software ENVI, eCognition and Polyworks (LiDAR only). Geospatial analysis will be conducted in ArcGIS. For vegetation analysis, we will use ordination software like Canoco and PcORD. Further statistical analyses, e.g. regression modelling, will be conducted with “R”. The data will be archived at the Institute of Landscape Ecology of the University of Münster. Any participant is free to archive the data at her/his institute, too.
A large part of the results of this study will be published within my cumulative PhD thesis. As the applicant group is quite interdisciplinary and the available data will be quite large, we expect high publication output in different areas of research which will assure a broad dissemination of the results. The results will also be presented at conferences on (landscape) ecology, biological invasions and remote sensing. As the area of study is already subject to Master theses at the the Universities of Bielefeld, Lisbon and Münster, Bachelor and Master students will be involved and encouraged to use the data for their theses. The data and the results of this project can be included, in the future, for undergraduate and graduate classes in biology, ecology, modelling and remote sensing.

Bibliography
Asner GP, Jones MO, Martin RE, Knapp DE & Hughes RF. 2008. Remote sensing of native and invasive species in Hawaiian forests. Remote Sensing of Environment, 112: 1912–1926.
Ehrenfeld JG. 2010. Ecosystem consequences of biological invasions. Annual Review of Ecology, Evolution, and Systematics, 41: 59-80.
Horus Institute. 2005. Acacia longifolia. Available from: http://www.institutohorus.org.br/download/fichas/Acacia_longifolia.htm [Accessed November 16, 2010].
Huang C & Asner GP. 2009. Applications of remote sensing to alien invasive plant studies. Sensors, 9: 4869-4889.
Joshi C, De Leeuw J, van Andel J, Skidmore AK, Lekhak HD, van Duren IC, Norbu N. 2006. Indirect remote sensing of a cryptic forest understorey invasive species. Forest Ecology and Management, 225 (1-3): 245-256.
Marchante E, Kjøller A, Struwe S & Freitas H. 2009. Soil recovery after removal of the N2-fixing invasive Acacia longifolia: consequences for ecosystem restoration. Biological Invasions, 11 (4): 813-823.
Rascher KG, Große-Stoltenberg A, Máguas C, Meira-Neto JAA & Werner C. In print. Acacia longifolia invasion impacts vegetation structure and regeneration dynamics in open dunes and pine forests. Biological Invasions.
Rejmanek M, Richardson DM & Pysek P. 2005. Plant invasions and invasibility of plant communities. In: van der Maarel E (ed.). Vegetation Ecology, 332-355. Blackwell, Oxford.
Richardson DM & Kluge RL. 2008. Seed banks of invasive Australian Acacia species in South Africa: Role in invasiveness and options for management. Perspectives in Plant Ecology, Evolution and Systematics, 10 (3): 161-177.
Wang L, Okin GS & Macko, SA. 2010. Remote sensing of nitrogen and carbon isotope compositions in terrestrial ecosystems. In: West JB et al. (eds.). Isoscapes: Understanding movement, pattern, and process on Earth through isotope mapping, 51-70. Springer, New York.
Werner C, Zumkier U, Beyschlag W & Máguas C. 2010. High competitiveness of a resource demanding invasive acacia under low resource supply. Plant Ecology, 206: 83-96.

For imaging spectroscopy, "clear sky" conditions are preferred. Acceptable is a sky which is covered up to 1/8 by cumulus clouds. Not acceptable are cirrus clouds because the illumination conditions would vary too much for imaging spectroscopy.
The LIDAR measurements do not require these strict conditions, but as LIDAR and hyperspectral data will most likely be collected at the same time, there should be optimal weather conditions.
In the case that the acquisition of the LIDAR data fails for some reason while the collection hyperspectral data was successful, the LIDAR data can be collected another day when the illumination conditions are not perfect. Good, clear atmospheric conditions would deliver the best LiDAR results.

The flight should take place in March/April. This is the time when the invasive species in focus, Acacia longifolia" is flowering which might enhance discrimination from the native species in the visible part of the spectra.
In addition, the flights shall take place between 2 hours before and 2 hours after local solar noon to guarantee sufficient illumination for imaging spectroscopy. 2.5 hours before/after solar noon might work as well if time is short (other projects, only few sunny days during a period of generally bad weather for the proposed time period, etc.).
The light conditions are more important for the acquisition of the hyperspectral than of the LiDAR data. For both data, clear atmospheric conditions are essential to get the best results.

The area of study consists mainly of the coastal dunes of the Nature Reserve "Lagoas de St. André e Sancha between Sines and Melides in Southwest Portugal. As the site is known from field trips, logistics such as accomodation and authorization for field work is assured. This area has been chosen for the following reasons:
The area has a high biodiversity value. It is very rich in species and landscape structures. Thus, it is protected by national (Reserva Natural das Lagoas de Stº André e da Sancha) and by European law (NATURA2000 habitat (PTCON0034) and partly by the bird directive (PTZPE0013, PTZPE0014)). Several rare and endemic plant species that are protected by national and European law (Annex II of the NATURA2000 directive) occur in this area. They are potentially threatened by the ecosystem changes caused by Acacia longifolia.
There are detailed digital vegetation maps available for this area, and there is detailed knowledge about all occuring plant species. In additon, selected A. longifolia occurences in different habitats (open dunes, dune forests with Pine or with Eucalyptus, Acacia shrublands) at different invasion stages (from establishing to dominating) have been mapped with Trimble dGPS for ground truthing. This map will be extended for the flight campaign. Historical photographs for this site are available which allow a sound comparision of the historical and mapped distribution of A. longifolia.
The Geographic Institute of the Portuguese military (IGeoE) has been informed about a planned flight campaign in this area and there were no concerns mentioned.
Within the area of study there are sufficient reference targets for atmospheric correction such as roads, parking areas, sandy beaches, lagoons, buildings, that will be also suitable as dGPS-based ground control points for later georegistration

We estimate one flight with a flight time of 2h at an estimated data aquisition speed of 4km/min and a turn time betwen the lines of 3 minutes and 15 lines in total. The swath width for the Hawk would be 450m with an overlap of 100m (50m at each side).
The flight shall take place at 1000m altitude at "clear sky" conditions between 2 hours before and after local noon in mid April.
A lower altitude to get a higher spatial resolution would be desired. This would lead to more flight lines and more flight hours. This has to be discussed with the aircraft operator.
The flight lines are orientated within the solar plane to guarantee best illumination conditions. The North-South extension of the indicated area of study is the required size and should not be diminuished. The E-W extension is minimum 1km and maximum 5km. This covers Acacia longifolia invasion at different stages and in different habitat types within the Nature Reserve "Lagoas de Santo André e Sancha". It also incorporates important references for atmospheric corrections and dGPS-based ground control points.
The NS extension should not be altered, the EW extension of the area of study can be altered within the mentioned dimension (covering a 3km wide coastal dune strip) if the flight lines were not lying within the solar plane, but parallel to the coast to safe flight lines. In the latter case, one flight perpendicular to the coastline would be needed for BRDF corrections.

The remote sensing department of the Portuguese Geographic Institute of the Military (IGeoE) has been contacted about the flight campaign already. There should be no restrictions.
Access to field equipment (ASD Field Spectrometer) is still needed.
A senior researcher with available field equipment has been contacted. I will also contact the colleagues I know from the EUFAR ADDRESSS summer school in Hungary 2010. Equipment and expertise of parallel flight campaigns could be shared. Further help from within the EUFAR network would be desired.
From 11th to 13th April 2011, the 7th Workshop on Imaging Spectroscopy organized the European Association of Remote Sensing Laboratories (EARSeL) will take place. Some of the participating scientists will attend this meeting. A flight during those days is basically not preferred, however if it would be the only possible flight window a plausible solution.

ALS50-II instrument specification:
- FOV: 40 deg.
- elevation point cloud in LAS and ASCII file format
- Multi return height intensity signals
- point density: >5 points/m²; the more, the better as single tree/shrub detection is one of the main targets using lidar in this study;
As the points density is a very crucial point in this study, a no calculator for the point density could be found on the aircraft operator`s webpage as e.g. for the Eagle/Hawk instruments, the set-up for the Lidar instrument (FOV, flight altitude) has to be discussed with the aircarft operator.

Images from the RCD105 digital camera

Full-range Hyperspectral eagle and hawk data, post-processed (NASA level 1b) from the aircraft operator
spectral binning can be applied to improve the quality of the data
AISA Eagle specification
- Ground pixel size: 0.7m at 1000 altitude
AISA Hawk Specification
- Ground pixel size: 1.3m at 1000 altitude

A smaller ground pixel size is desired. As this incorporates a lower flight altitude and more flight lines, this has to be checked with the aircraft operator.
Generally, a fine spectral and a fine spatial resolution are required. Thus the aquisition of the Lidar and the hyperspectral data including the flight patterns need further discussion between the research and the aircraft operator team.

None

AISA Eagle/Hawk hysperspectral sensor
LEICA ALS50-II LIDAR
Leica RCD 105 Digital Photogrammetric Camera

Trimble dGPS (1 unit)
ASD Field Spectrometer (requested)
ENVI Image Processing Software
Polyworks (Lidar processing software)
eCognition (Imaging Processing Software)
ArcGIS (Geospatial Analysis software)
PcORD, CANOCO (ordination software for vegetation data)
Lab facilities including basic chemical analysis of the soil and plant leafs and stable isotopes and near-infrared reflectance spectroscopy calibrations.
Facilities to analyze the age of Acacia longifolia by tree ring analysis are available, a dendrochronological/ecological study about Acacia longifolia is in progress and a study to analyse more occuring shrubs and tress and to reveal size-age relationships is planned.

None

A Trimble dGPS is available for ground truthing at the Institute of Landscape Ecology, University of Münster. Reference sites of the vegetation types including different stages of invasion by Acacia longifolia and different invaded habitats (canopy building in the open dunes and dominating the understorey in the forests) within the area of study were already mapped with the Trimble dGPS during a field campaign in September/October 2010. These previously identified reference sites will allow for effective field work since logistics can be optimized for reference checking.
Parallel to the flight, an ASD Field Specrometer will be used to collect field spectra for ground truthing of the different vegetation types including different stages of invasion by A. longifolia across different invaded habitats. Field spectra will also be taken of native species canopies close and far away from A. longifolia to analyze the enrichment of neighbouring plants by nitrogen-fixing A. longifolia. Plant samples will be collected for further chemical analyses in the labs of the participant group (N,P,K, stable isotopes, lignin, cellulose, tannin, chlorophyll).
A terrestrial laser scanner, Optech ILRIS-3D, will be used in the field to improve terrain modelling via airborne laser scanning data interpretation.
Digital hemispherical photographs will be applied to analyse and model the light conditions and the Leaf Area Index to add to the vegetation and canopy structure analysis with LiDAR data.

The flight at an altitude of 1000m or less will generate hyperspectral data with a spatial resolution of about 1m. The LiDAR data with its high point density (> 5 points/m²) for single tree selection will deliver fine resolution data about the vegetation and canopy structure. The ground truth data will consist of field spectral data for calibration, dGPS mapped Acacia longifolia individuals and populations, dGPS mapped native shrubs and trees including data about plant individual height, crown size, phenological state and growth form, and dGPS mapped vegetation types, the results of the chemical analysis of the leaf samples and the chemical analysis of sap flow including isotope analysis of selected Acacia and overstorey trees to characterize the canopy chemistry in addition to leaf analysis. The vegetation types will be characterized by plot-based description of species cover for every vegetation layer. The plots will be selected by stratified random sampling. The selection of the sub-units of the study area will be based on experience of ealiert field work and aerial photograph interpretation.
The raw data will be atmospherically corrected and then georeferenced. There are sufficient reference targets for the remote sensing data in the area of study (e.g. sandy plains, lagoons, car parkings, buildings, roads). Additional, artificial targets (coloured boards with standard reflection properties) will be laid out at site and mapped with the dGPS for referencing.
The key questions is that if by fusing fine resolution LiDAR, fine resolution hyperspectral and detailed ecological data a distribution map of A. longifolia in open dunes and dune forest can be generated, and if different stages of invasion can be identified. Furthermore, a question is if the nitrogen enrichment of neighbouring plants and the alteration of the vegetation structure and the above ground biomass by A. longifolia can be observed. Differences in biomass and diversity of invaded and non-invaded sites can be calculated.
Ground truth data have been retrieved in various projects of the participants such as historical photographs, digital vegetation maps, selected dGPS-mapped vegetation types and A. longifolia invasion stages, selected dGPS-mapped A. longifolia populations and selected analyses on A. longifolia and native species’ leaf chemistry. Digital hemispherical photography and ground-based LiDAR data Optech’s ILRIS 3D terrestrial laser scanner will be applied to improve the knowledge and the calibration of the vegetation and canopy structure (vegetation density, light conditions, leaf area index). A project to derive the age of A. longifolia by stem disk analysis and relating this to height, crown size and above ground biomass is in progress.
Analysis of the spatial distribution of A. longifolia including the historical landuse derived from historical aerial photographs can help to identify the triggers and the limits of its invasion, and lead to and map vulnerable sites susceptible to future invasions. In a possible future project, this study could be extended to other sites along the Portuguese coast to analyse the influence of the pronounced climatic gradient between northern and southern Portugal which is often discussed as a very important factor of A. longifolia invasions.
When regarding A. longifolia as a model for an invasive species, the results of this study can be related to invasive species with similar characteristics such as growth form, fixation of atmospheric nitrogen, invasion on nutrient poor soils, especially when occuring in the understorey of forests.
The data will be analysed using the appropiate software for image analysis (ENVI, eCognition), for LiDAR data modelling and visualization (Polyworks), for ordination of the vegetation data (CANOCO, PcOrd, R) and geospatial analysis (ArcGIS, R).

Preferred: April 2011
Acceptable: March, April, May 2011

If the time for review is too short, the flight could be delayed. The latest date would be March/April 2011. A flight in summer/autumn 2010 is possible, but would need to be discussed with the applicant and the aircraft operator. I just have been contacted by researchers who would like to apply for a LiDAR flight at a Long Term Ecological Research (LTER) site in Coruche, 70km northeast of Lisbon. They would need land surface structure data and calibrate these data with their isotope and carbon flux measurement from their Eddy covariance tower.

Yes

The group of participant consists of research institutions of different areas (Biology, Ecophysiology, Applied Landscape Ecology, Ecosystem Research). National authoratives of remote sensing, forestry and nature conservation in Portugal have asked to join the campaign.
There is a long lasting exchange of researchers including PhD, Master and Bachelor students between the Universities of Bielefeld, Lisbon, and since recently, the University of Münster. The exchange were and are mainly funded by the DAAD personal and travel grants, DFG and FCT projects and EU Erasmus student exchanges. Recently, there was a field trip of the work group of Münster with Master students to study Acacia longifolia invasion in Portugal advertising the possible upcoming EUFAR flight campaign.
It is intended to publish the results at upcoming international conferences about Ecology, Remote Sensing or Biological Invasions.
The flight campaign will also be a subject of the exchange network between the Universities of Bielefeld, Lisbon, Münster and Vicosa (Brazil) to study Acacia invasion worldwide. The University of Stellenbosch (South Africa) and the University of Coimbra (Portugal) have also been contacted and will be part of the network soon.
The data set can be used in master and bachelor courses to show up-to-date possibilities of the application of remote sensing data in ecological studies.
The results of this project may inspire new ideas to study A. longifolia invasion in Northern Portugal and worldwide (e.g. Brazil, South Africa). A. longifolia can be seen as a "model invader" and the knowledge can also inspire similar projects to study invasive species and ecosystem processes.
As I participated in the 1st FP7 EUFAR Training Course“ADvanced Digital Remote sensing in Ecology and earth Sciences Summer School” (ADDRESSS) in August 2010 in Hungary where the students were taught how to apply LiDAR and hyperspectral data in ecological studies, and where the students were repeatedly encouraged to apply for a flight within via “Transnational Access”, this proposed flight would directly connect to EUFAR’s commitment to promote young scientists. If the intended follow-up workshop of the ADDRESSS summer school will take place next summer, then the preliminary results could be presented there, too. I will contact my colleagues from the ADDRESSS summer school if they would like to participate in the proposed campaign.

In all existing projects, the focus is in the impact on the impact of Acacia longifolia on the diversity, structure and the ecosystem processes local scale.
This project will provide for the first time the required data to analyze A.longifolia's impact on landscape scale.
It fits in very well with my PhD project as I am funded in Germany so far by my working, as in Portugal I will be funded by the DAAD and in Brazil I will be funded by the EU IRSES Programm. My work group will apply together with our colleagues from the University of Bielefeld for a DFG project to sustain my funding (salary) and to install a counterpart at the Biological Faculty in Bielefeld. The EUFAR project would complement very well this funding: our field work is basically funded, but there is no funding for a flight campaign. This EUFAR project would fit it perfectly and complement the current projects, especially my PhD project, very well.

Yes

14

None

• Acacia longifolia invasion impacts vegetation structure and regeneration dynamics in open dunes and pine forests
• Combining vegetation indices, constrained ordination and fuzzy classification for mapping semi-natural vegetation units from hyperspectral imagery
• Competitive displacement or biotic resistance? Disentangling relationships between community diversity and invasion success of tall herbs and shrubs
• Estimating tree and stand variables in a Corsican Pine woodland from terrestrial laser scanner data
• On the use of phloem sap δ13C as an indicator of canopy carbon discrimination
• Project DeInVader: Map of general area