Skip to main content
zenodoopen

Map of built-up expansion ("nedbygging") over Norway 2017-2022 version 2

<p>Version 2 of the dataset https://zenodo.org/records/10566644</p> <p>Changes from first version include:</p> <ul> <li>added crowdsourced verification labels to the dataset gathered from the interactive app (link below) explained here: https://www.nina.no/Om-NINA/Aktuelt/Nyheter/article/kartlegg-nedbygging-av-natur-selv</li> <li>added the year of change crowdsourced labels</li> <li>added the type of built-up expansion as labelled by the NRK team</li> </ul> <p>Data can be viewed interactively here: <a href="https://nina.earthengine.app/view/nedbygging">https://nina.earthengine.app/view/nedbygging</a></p> <h4>(see Norwegian description below)</h4> <h2>&nbsp;1. Dataset Information</h2> <p>&nbsp;&nbsp; - Title: Map of built-up expansion over Norway 2017-2022</p> <p>&nbsp;&nbsp; - Author(s): Zander Venter (NINA), Mads Nyborg St&oslash;stad (NRK), Ruben Solvang (NRK), Anne Linn Kumano-Ensby (NRK), Su Thet Mon (NRK)</p> <p>&nbsp;&nbsp; - Contact Information: zander.venter@nina.no</p> <p>&nbsp;&nbsp; - Date of Data Generation: 06.01.2024</p> <p>&nbsp;&nbsp; - Version: 1</p> <p>&nbsp;&nbsp; - Description: This is the dataset used in the <a href="https://www.nrk.no/dokumentar/xl/nrk-avslorer_-44.000-inngrep-i-norsk-natur-pa-fem-ar-1.16573560">NRK article</a> published on 06.01.2024. The data contains polygons outlining potential &ldquo;nedbygging&rdquo; (hereafter translated to &ldquo;built-up expansion&rdquo; in English) events between 2017 and 2022 over Norway. The built-up expansion polygons were identified using a combination of Sentinel-2 satellite imagery, a fully convolutional neural network (a type of AI model) from Google called <a href="https://dynamicworld.app/">Dynamic World</a> and NINA&rsquo;s time series analysis thereof. The method to create the map will be published by NINA at a later date. The original map was created by NINA, but NRK performed some post-processing which included joining some polygons which were part of the same built-up expansion event (e.g. a long road). It is important to note that the map is a result of AI and has errors in it. Therefore, users are encouraged to read the sections on data quality and usage information below. &nbsp;Users can refer to <a href="https://www.sciencedirect.com/science/article/pii/S2212041624000056">Venter et al. (2024)</a>&nbsp;for details on the scientific best practice which the NRK journalists followed to ensure that their reported area estimates in the article were not biased. In summary, the map is wrong 18% of the time. Users should expect to find that on average 1 in 5 square meter is incorrectly identified as built-up expansion. There are also many instances of built-up expansion which will be missed in the map such as forestry road development, building of small cabins etc.&nbsp;</p> <h2>&nbsp;2. File Details</h2> <p>&nbsp;&nbsp; - Format: Shapefile (.shp, .shx, .dbf, .prj)</p> <p>&nbsp;&nbsp; - Size: 13.27 MB</p> <h2>&nbsp;3. Geospatial Information</h2> <p>&nbsp;&nbsp; - Coordinate System: EPSG:32632, UTM zone 32N</p> <p>&nbsp;&nbsp; - Spatial Resolution: 10m</p> <p>&nbsp;&nbsp; - Geographical Coverage: Norway mainland (excludes Svalbard)</p> <p>&nbsp;&nbsp; - Temporal Coverage: 2017 to 2022</p> <h2>&nbsp;4. Data Content</h2> <p>&nbsp;&nbsp; - Attributes Included:</p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *id*: unique identity number for each polygon</p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *unders&oslash;kt*: whether the polygon has been investigated manually using visual interpretation of orthophotos. &ldquo;ja&rdquo; = &ldquo;yes&rdquo; and &ldquo;nei&rdquo; = &ldquo;no&rdquo;</p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *<span><span>unders&oslash;kt_source*: whether the data was collected by the NRK team or the crowdsourcing effort</span></span></p> <p><span><span>&nbsp; &nbsp; &nbsp; - *</span></span><span><span>kategori_1*: the type of built-up expansion labelled by the NRK team - see Google Translate for translations</span></span></p> <p><span><span>&nbsp; &nbsp; &nbsp; - *year*: the year in which the built-up expansion occurred as defined by the crowdsourcing volunteers</span></span></p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *ai_feil*: whether the AI model method correctly (&ldquo;riktig&rdquo;) or incorrectly (&ldquo;feil&rdquo;) identified natural habitat conversion to built-up surface. Values where *unders&oslash;kt* == &ldquo;nei&rdquo; are labelled as &ldquo;ikke_verifisert&rdquo;&nbsp;</p> <h2>&nbsp;5. Data Quality</h2> <p>&nbsp;&nbsp; - Accuracy: As described above, the false positive rate of the map was 18% based on 500 locations used for map validation and accuracy assessment. We did not quantify a false negative rate and balanced accuracy estimates because this would have required a denser sample for manual verification. Therefore, it is likely that there are many instances of built-up expansion that our map does not capture. After the formal accuracy assessment using the 500 stratified random points, NRK verified additional polygons (total of 3875) in the dataset during their investigative journalism workflow. Although these were not collected in a systematic manner, then can still be useful for some downstream tasks such as exploring what causes the AI model to misidentify built-up expansion.</p> <p>&nbsp;&nbsp; - Validation Methods: A design-based approach was used to quantify map accuracy and estimate uncertainty around the resulting area estimate reported in the NRK article. The details of this method are reported in <a href="https://www.sciencedirect.com/science/article/pii/S2212041624000056">Venter et al. (2024)</a>. This approach quantifies the error in the AI-derived map, and corrects for this using a stratified area estimator. Therefore, the total built-up expansion of 208 km&lt;2&gt; reported in the NRK article has been bias-corrected. We also quantified 95% confidence intervals around this are estimate of 9.8 km&lt;2&gt;. It is important to note that the validation approach was conducted on individual Sentinel-2 pixels of 10x10m and not at the polygon level. Therefore, we did not quantify the error in the precision of the polygon shape in terms of capturing the full extent of a given built-up expansion event.</p> <h2>&nbsp;6. Usage Information</h2> <p>&nbsp;&nbsp; - Use Limitations: Considering the map error described above, users should proceed with caution when analysing the map to derive area statistics or overlays with other maps. As described in <a href="https://www.sciencedirect.com/science/article/pii/S2212041624000056">Venter et al. (2024)</a>, simply adding the areas of the polygons (or &ldquo;pixel counting&rdquo; with maps formatted as images) without accounting for the error in the map will lead to incorrect area statistics. We recommend that users validate the map for their municipality or study area before proceeding with analysis. It is likely that the margin of error is highly variable between municipalities. For example, although we have not quantified it, we noticed many AI mistakes in mountainous regions due to snow and ice interference and therefore high-altitude municipalities might have more errors than low-altitude ones.</p> <p>&nbsp;</p> <h1>&nbsp;Norwegian description:&nbsp;</h1> <h2>&nbsp;1. Datasettinformasjon</h2> <p>&nbsp;&nbsp;&nbsp; - Tittel: Kart over nedbygging over Norge 2017-2022</p> <p>&nbsp;&nbsp;&nbsp; - Forfatter(e): Zander Venter (NINA), Mads Nyborg St&oslash;stad (NRK), Ruben Solvang (NRK), Anne Linn Kumano-Ensby (NRK), Su Thet Mon (NRK)</p> <p>&nbsp;&nbsp;&nbsp; - Kontaktinformasjon: zander.venter@nina.no</p> <p>&nbsp;&nbsp;&nbsp; - Dato for datagenerering: 06.01.2024</p> <p>&nbsp;&nbsp;&nbsp; - Versjon: 1</p> <p>&nbsp;&nbsp;&nbsp; - Beskrivelse: Dette er datasettet som brukes i <a href="https://www.nrk.no/dokumentar/xl/nrk-avslorer_-44.000-inngrep-i-norsk-natur-pa-fem%20-ar-1.16573560">NRK-artikkelen</a> publisert 06.01.2024. Dataene inneholder polygoner som skisserer potensiell nedbygging mellom 2017 og 2022 over Norge. Nedbyggingsomr&aring;dene ble identifisert ved hjelp av en kombinasjon av Sentinel-2 satellittbilder, et fullstendig konvolusjonelt nevralt nettverk (en type KI-modell) fra Google kalt <a href="https://dynamicworld.app/">Dynamic World</a> og NINAs tidsserie-analyse av dette. Metoden for &aring; lage kartet vil bli publisert av NINA p&aring; et senere tidspunkt. Det originale kartet ble laget av NINA, men NRK utf&oslash;rte en del etterbehandling som inkluderte sammenf&oslash;yning av noen polygoner som var en del av den samme oppbygde utvidelseshendelsen (f.eks. en lang vei). Det er viktig &aring; merke seg at kartet er produsert ved hjelp av kunstig intelligens og inneholder feil. Derfor oppfordres brukere til &aring; lese avsnittene om datakvalitet og bruksinformasjon nedenfor. Brukere kan referere til&nbsp;<a href="https://www.sciencedirect.com/science/article/pii/S2212041624000056">Venter et al. (2024) </a>for detaljer om den vitenskapelige beste praksisen som NRK-journalistene fulgte for &aring; sikre at deres rapporterte arealstatistikk i artikkelen er korrekt. Oppsummert er 18 % av arealet i kartet feil. Brukere b&oslash;r forvente &aring; finne at i gjennomsnitt 1 av 5 kvadratmeter er feilaktig identifisert som nedbygging. Det er ogs&aring; mange tilfeller av nedbygging som som ikke vil vises i kartet, som skogsveiutbygging, bygging av sm&aring;hytter mm.</p> <h2>&nbsp;2. Fildetaljer</h2> <p>&nbsp;&nbsp;&nbsp; - Format: Shapefil (.shp, .shx, .dbf, .prj)</p> <p>&nbsp;&nbsp;&nbsp; - St&oslash;rrelse: 13,27 MB</p> <h2>&nbsp;3. Geospatial informasjon</h2> <p>&nbsp;&nbsp;&nbsp; - Koordinatsystem: EPSG:32632, UTM-sone 32N</p> <p>&nbsp;&nbsp;&nbsp; - Rolig oppl&oslash;sning: 10m</p> <p>&nbsp;&nbsp;&nbsp; - Geografisk dekning: Norges fastland (ekskluderer Svalbard)</p> <p>&nbsp;&nbsp;&nbsp; - Tidlig dekning: 2017 til 2022</p> <h2>&nbsp;4. Datainnhold</h2> <p>&nbsp;&nbsp;&nbsp; - Attributter inkludert:</p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *id*: unikt identitetsnummer for hver polygon</p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *unders&oslash;kt*: om polygonet er unders&oslash;kt manuelt ved bruk av visuell tolkning av ortofoto.</p> <p>&nbsp; &nbsp; &nbsp; &nbsp;- *unders&oslash;kt_source*: om dataene er samlet inn av NRK-teamet eller crowdsourcing-innsatsen</p> <p>&nbsp; &nbsp; &nbsp; &nbsp;- *kategori_1*: typen nedbygging merket av NRK-teamet&nbsp;</p> <p>&nbsp; &nbsp; &nbsp; &nbsp;- *year*: &aring;ret hvor nedbygging skjedde som definert av crowdsourcing</p> <p>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - *ai_feil*: om AI-modellmetoden var &ldquo;riktig&rdquo; eller &ldquo;feil&rdquo;. Verdier der *unders&oslash;kt* == &laquo;nei&raquo; er merket som &laquo;ikke_verifisert&raquo;</p> <h2>&nbsp;5. Datakvalitet</h2> <p>&nbsp;&nbsp;&nbsp; - N&oslash;yaktighet: Som beskrevet ovenfor var andelen falske positive punkter i kartet 18 % basert p&aring; 500 steder (pr&oslash;veflater) brukt for kartvalidering og n&oslash;yaktighetsvurdering. Vi kvantifiserte ikke andelen falske negative punkter og balanserte n&oslash;yaktighetsestimater, fordi dette ville ha krevd en tettere stikkpr&oslash;vedensitet for manuell verifisering. Derfor er det sannsynlig at det er mange tilfeller av nedbygging som kartet v&aring;rt ikke fanger opp. Etter den formelle n&oslash;yaktighetsvurderingen ved bruk av 500 stratifiserte tilfeldige pr&oslash;veflater, verifiserte NRK ytterligere polygoner (totalt 3875) i datasettet i l&oslash;pet av deres journalistiske unders&oslash;kelser. Selv om disse ikke ble samlet inn p&aring; en systematisk m&aring;te, kan de fortsatt v&aelig;re nyttige for noen oppf&oslash;lgingsanalyser som &aring; utforske hva som f&aring;r AI-modellen til &aring; feilidentifisere nedbygging.</p> <p>&nbsp;&nbsp;&nbsp; - Valideringsmetoder: En designbasert tiln&aelig;rming (&laquo;design-based area estimation&raquo; p&aring; engelsk) ble brukt for &aring; kvantifisere kartn&oslash;yaktighet og estimere usikkerhet rundt det resulterende arealestimatet rapportert i NRK-artikkelen. Detaljene ved denne metoden er forklart i <a href="https://www.sciencedirect.com/science/article/pii/S2212041624000056">Venter et al. (2024)</a>. Denne tiln&aelig;rmingen kvantifiserer feilen i det KI-avledede kartet, og korrigerer for dette ved &aring; bruke en stratifisert arealestimator. Derfor er den totale bebygde utvidelsen p&aring; 208 km&lt;2&gt; som er rapportert i NRK-artikkelen, skjevhetskorrigert. Vi kvantifiserte ogs&aring; 95 % konfidensintervaller rundt dette er estimat p&aring; 9,8 km&lt;2&gt;. Det er viktig &aring; merke seg at valideringstiln&aelig;rmingen ble utf&oslash;rt p&aring; individuelle Sentinel-2 piksler p&aring; 10x10m og ikke p&aring; polygonniv&aring;. Derfor kvantifiserte vi ikke feilen i presisjonen til polygonformen n&aring;r det gjelder &aring; fange opp hele omfanget av en gitt oppbygget ekspansjonshendelse.</p> <h2>&nbsp;6. Bruksinformasjon</h2> <p>&nbsp;&nbsp;&nbsp; - Bruksbegrensninger: Med tanke p&aring; kartfeilen beskrevet ovenfor, b&oslash;r brukere fortsette med forsiktighet n&aring;r de analyserer kartet for &aring; beregne omr&aring;destatistikk eller overlegg med andre kart. Som beskrevet i <a href="https://www.sciencedirect.com/science/article/pii/S2212041624000056">Venter et al. (2024)</a>, ved &aring; legge til omr&aring;dene til polygonene (eller &laquo;pikseltelling&raquo; med kart formatert som bilder) uten &aring; ta hensyn for feilen i kartet vil f&oslash;re til feil arealstatistikk. Vi anbefaler at brukere validerer kartet for sin kommune eller studieomr&aring;de f&oslash;r de g&aring;r videre med analyse. Det er sannsynlig at feilmarginen varierer mye mellom kommuner. For eksempel, selv om vi ikke har kvantifisert det, la vi merke til mange AI-feil i fjellomr&aring;der p&aring; grunn av sn&oslash;- og is, og derfor kan kommune med mye fjell ha flere feil enn de langs kysten (for eksempel).</p>

ShareScore

40/100

Overall dataset sharing score

Score breakdown

These five areas show where the dataset supports — or may limit — practical reuse.

Stewardship
8
Harmonization
4
Access
16
Reuse readiness
8
Engagement
4

Topics