Mapping Lupinus nootkatensis in Iceland using SPOT 5 images

Abstract (Swedish)

Frön och rötter från Lupinus nootkatensis infördes till Island år 1945 med syftet att undersöka artens potential att bidra till återställandet av eroderad mark. Arten, som ursprungligen kommer från Alaska, visade sig vara mycket effektiv i samband med detta och dessutom kostnadseffektiv. Resultaten beror framförallt på Lupinus nootkatensis betydande biomassa, som förbättrar jordens bördighet, men artens snabba spridningsmönster bidrar också dess efektivitet. Under senare år har kunskapen om Lupinus nootkatensis ökat och den goda spridningsförmågan har identifierats som ett problem, eftersom den hotar den biologiska mångfalden på Island. Arten anses idag av många vara en främmande art, som kan invadera och ta över områden, som är täckta med naturlig isländsk flora. Användandet av Lupinus nootkatensis för återställande av skadade markytor är därför mycket omdebatterad.
Omfattande spridning av främmande arter kan ha negativ påverkan på ekosystem och naturliga processer samt leda till ekonomiska förluster. God kunskap om utbredningen av Lupinus nootkatensis är grundläggande för planering av åtgärder för begränsning av växtens spridning.
Syftet med denna studie är att utveckla en fjärranalysbaserad metod, som kan användas för att observera och övervaka Lupinus nootkatensis utbredning på Island. Fjärranalysbaserade metoder har inte tidigare använts för detta ändamål på Island.
Fältarbetet inom ett 45 km2 stort studieområde nära Reykjavik på sydvästra Island omfattade lokalisering av prov ytor (plots) för kartläggning respektive validering av noggrannhet. Studien omfattade utveckling av metodik som inkluderade användning av fältarbetets resultat för att identifiera Lupinus nootkatensis inom studieområdet. För detta används okontrollerade och kontrollerade (unsupervised och supervised) klassningsmetoder tillämpade på jordresurssatellitdata (SPOT 5). Dessa klassningsmetoder undersöktes även för att identifiera förekomsten av Lupinus nootkatensis inom ett större geografiskt område på Reykjaneshalvön på sydvästra Island.
De båda klassningsmetoderna (okontollerad och kontrollerad) gav likartade resultat. Den högsta klassningsnoggrannheten erhölls med en kontrollerad klassningsmetod maximum likelihood klassning, baserad på de spektrala banden närainfrarött, rött, grönt och mellaninfrarött. Denna klassningsprocedur resulterade i 94-95% producent- och 90-97% användarnoggrannhet och ett Kappavärde på 0.88 för studieområdet. Inom det större geografiska området erhölls liknande resultat utanför tätbebyggda områden men inom tätbebyggda områden var klassningsnoggrannheten lägre.
Studien visar att okontollerad och kontollerad klassning av SPOT 5 data kan användas på sydvästra Island för att identifiera och lokalisera områden tättbevuxna med Lupinus nootkatensis.

Abstract

Seeds and roots of Lupinus nootkatensis were imported to Iceland from Alaska in 1945 for the purpose of testing it as a re-vegetation plant for eroded land. It has turned out to be an effective and relatively inexpensive plant for re-vegetation, mainly because of its fast growth, high biomass contribution to soil fertility, and its capability to spread. Its ability to spread has however become an increasing worry and is currently seen by many as a disadvantage. Lupinus nootkatensis has been recognized as an alien invasive plant in Iceland that can invade well vegetated areas including native heathland. Thus the use of Lupinus nootkatensis in re-vegetation of eroded land has become debatable. Invasive species may not only have negative impact on ecosystems and natural processes but also cause economic cost. For management and any planning of controlling the spread of Lupinus nootkatensis in Iceland the knowledge of its distribution is fundamental. The aim of this study is to develop a remote sensing based methodology for mapping and monitoring the distribution of Lupinus nootkatensis in Iceland. Remote sensing has not been used for this purpose in Iceland before. Field work was carried out in a 45 km2 study area near Reykjavik in SW Iceland. Within the area both land cover plots for mapping and plots for evaluation of accuracy were sampled. The study involved designing a methodology to map Lupinus nootkatensis within the study area based on field sampling. This was done by use of supervised and unsupervised classification, applied on SPOT 5 imagery. These classification procedures were also used to map the occurrences of Lupinus nootkatensis within a larger area within the Reykjanes peninsula in SW Iceland. Whereas the two classification procedures (unsupervised and supervised) gave similar results, the absolute highest classification accuracies were obtained by means of the maximum likelihood supervised classification based on the NIR, red, green and MIR bands of the image. This classification procedure gave 94-95% producer accuracy and 90-97% user accuracy, and an overall Kappa of 0.88 within the study area. Similar classification results were obtained within the larger area outside urban areas, but within urban areas low classification accuracies were obtained. The study shows that both unsupervised and supervised classification procedures of SPOT 5 images may be used in SW Iceland to detect dense Lupinus nootkatensis patches.

Abstract

Popular science: Mapping Nootka lupine in Iceland.
The Nootka lupine (Lupinus nootkatensis) was imported to Iceland from Alaska in 1945 with the purpose of testing it as a re-vegetation plant on eroded land. Many characteristics of the Nootka lupine make it an effective and relatively inexpensive plant to use in re-vegetation. The plant is fast growing, produces a large amount of biomass, contributed to soil fertility and has the capability to spread. However, the ability to spread has become an increasing worry and is currently seen by many as a disadvantage. The Nootka lupine has now been recognized as an alien invasive plant species in Iceland that can invade vegetated areas including native heath land. Thus the use of Nootka lupine in re-vegetation of eroded land has become debatable.

The aim of this study was to develop a methodology for mapping the distribution of Nootka lupine in Iceland. Knowledge like that is fundamental for good management practises. Studies aiming to improve active inputs related to the re-vegetation of eroded surfaces may also benefit from improved knowledge of the distribution of Nootka lupine.

The methodology that was developed was based on the use of field samplings and an image from the SPOT 5 satellite.

Based on the spectral reflectance of the sunlight from the earth surface, the satellite sensor produces the satellite image. Because different surface and vegetation types have different reflectance the image can be used to provide a map of the earth surfaces.

By field observation several lupine patches were recognised and located on the satellite image and their reflectance value were categorised as Nootka lupine reflectance. A classification computer program was then used to recognise a similar reflectance elsewhere on the satellite image and thereby providing a tool for mapping the distribution of Nootka lupine.

The classification gave a map of the distribution of the Nootka lupine within a study area. The overall accuracy, when comparing classification map with field observations was 94 %.

This study has provided a tool that can be used for detecting the distribution of Nootka lupine country-wide in Iceland with good and accurate results. Through this methodology for mapping it will be possible to monitor the distribution. Being able to detect the spreading or withdrawing of the Nootka lupine from particular areas, by using satellite images taken years apart, will provide long needed information on the spreading behaviour of the Nootka lupine.

Abstract (Icelandic)

Árið 1945 voru nokkrar rætur og fræ af alaskalúpínu (Lupinus nootkatensis) flutt inn til Íslands frá Alaska til að kanna möguleika á að nota tegundina í landgræðslu. Lúpínan hefur verið talin áhrifarík og ódýr uppgræðslutegund. Það skýrist helst með hraðri og mikilli uppbyggingu lífmassa sem eykur frjósemi jarðvegs við rotnun, og einnig með mikilli dreifingarhæfni plöntunnar. Lúpínan hefur dreifst hratt út og verið skilgreind sem ágeng framandi tegund á Íslandi. Hún getur jafnvel yfirtekið vel gróin svæði þar sem fyrir er náttúrulegur gróður. Notkun lúpínu til uppgræðslu hefur því orðið umdeild á síðari árum.
Ágengar framandi tegundir geta haft neikvæð áhrif á vistkerfi og ýmsa ferla í náttúrunni og valdið fjárhagslegum skaða. Þekking á útbreiðslu lúpínu er grundvallaratriði í skipulagningu á stjórnun aðgerða gegn útbreiðslu plöntunnar og ágengni hennar.
Markmið þessarar rannsóknar er að þróa aðferð byggða á fjarkönnun, sem má nota til að rannsaka og fylgjast með útbreiðslu alaskalúpínu á Íslandi. Fjarkönnun hefur ekki verið notuð í þessum tilgangi á Íslandi áður.
Vettvangsvinna fór fram innan 45 km2 rannsóknarsvæðis í nágrenni Reykjavíkur. Hún fól í sér að staðsetja tvennskonar reiti, annarsvegar til notkunar við kortlagningu og hins vegar við mat á nákvæmni kortlagningar. Skráð var yfirborðsgerð reitanna í fyrra tilvikinu og í því seinna var skráð hvort reitur var með eða án lúpínu. Þróuð var aðferðafræði, þar sem niðurstöður vettvangsmælinga voru notaðar til að kortleggja útbreiðslu lúpínu af SPOT 5 gervitunglamynd innan rannsóknarsvæðisins. Til þess voru notaðar tvær flokkunaraðferðir, annars vegar sjálfvirk flokkun (unsupervised classification) og hins vegar stýrð flokkun (supervised classification). Sú síðarnefnda var einnig notuð til að kortleggja lúpínu á stærra svæði, umhverfis rannsóknarsvæðið, sem náði yfir Reykjanesskaga, hluta Ölfuss og höfuðborgarsvæðið.
Sjálfvirk og stýrð flokkun gáfu svipaðar niðurstöður, en mest greiningarnákvæmni náðist með stýrðri flokkunaraðferð, maximum likelihood, sem var beitt á nærinnrautt, rautt, grænt og miðinnrautt band gervitunglamyndarinnar, innan rannsóknarsvæðisins. Nákvæmni flokkunar var reiknuð út á þrjá vegu. Í fyrst lagi hvernig vettvangsreitir með og án lúpínu flokkuðust (producer accuracy). Þá fékkst mest 94-95% nákvæmni. Í öðru lagi hvernig flokkun í hverri einingu gervitunglamyndarinnar (pixel) samsvaraði greiningu á sama stað á jörðu niðri (user accuracy). Þá fékkst mest 90-97%. Í þriðja lagi var reiknaður Kappastuðull, sem var hæstur 0,88 sem þýðir að viðkomandi flokkun er 88% betri en tilviljunarkennd flokkun ("true" agreement versus "chance" agreement). Svipuð greiningarnákvæmni fékkst innan stærra svæðisins utan þéttbýlis, en nákvæmni greiningarniðurstaðna á var hins vegar lág innan þéttbýlissvæða.
Rannsóknin sýnir að bæði má nota sjálfvirku og stýrðu aðferðina til að greina þéttar lúpínubreiður á SPOT 5 gervitunglamyndum á suðvesturlandi.

• author: Thorsteinsdottir, Arna Björk
• supervisor: Karin Hall-Könyves
• organization: Dept of Physical Geography and Ecosystem Science
• course: NGEM01 20111
• year: 2011
• type: H2 - Master's Degree (Two Years)
• subject: Earth and Environmental Sciences
• keywords: alien invasive species, physical geography, remote sensing, unsupervised classification, maximum likelihood classification, geography
• publication/series: Student thesis series INES
• report number: 229
• funder: Soil Conservation Service in Iceland
• funder: Rannis, The Icelandic Centre for Research
• language: English
• id: 2224334
• date added to LUP: 2011-12-09 14:44:02
• date last changed: 2012-03-20 11:47:11