Data CitationPlease cite this dataset as follows:Vasquez, V., Cushman, K., Ramos, P., Williamson, C., Villareal, P., Gomez Correa, L. F., & Muller-Landau, H. (2023). Barro Colorado Island 50-ha plot crown maps: manually segmented and instance segmented. (Version 2). Smithsonian Tropical Research Institute. https://doi.org/10.25573/data.24784053This data is licensed as CC BY 4.0 and is thus freely available for reuse with proper citation. We ask that data users share any resulting publications, preprints, associated analysis code, and derived data products with us by emailing mullerh@si.edu. We are open to contributing our expert knowledge of the study site and datasets to projects that use these data; please direct queries regarding potential collaboration to Vicente Vasquez, vasquezv@si.edu, and Helene Muller-Landau, mullerh@si.edu.Note that this dataset is part of a collection of Panama UAV data on Smithsonian Figshare, which can be viewed at https://smithsonian.figshare.com/projects/Panama_Forest_Landscapes_UAV/115572Additional information about this research can be found at the Muller-Landau lab web site at https://hmullerlandau.com/All required code is freely available at https://github.com/P-polycephalum/ForestLandscapes/blob/main/LandscapeScripts/segmentation.py and it can be cited as:Vicente Vasquez. (2023). P-polycephalum/ForestLandscapes: segmentwise (v0.0.2-beta). Zenodo. https://doi.org/10.5281/zenodo.10380517Data DescriptionThis dataset is part of a larger initiative monitoring forests in Panama using drones (unoccupied aerial vehicles), an initiative led by Dr. Helene Muller-Landau at the Smithsonian Tropical Research Institute. As part of this initiative, we have been collecting repeat imagery of the 50-ha forest dynamics plot on Barro Colorado Island (BCI), Panama, since October 2014 (see Garcia et al. 2021a, b for data products for 2014-2019).Contained within this dataset are two sets of field-derived crown maps, presented in both their raw and improved versions. The 2021 crown mapping campaign was overseen by KC Cushman, accompanied by field technician Pablo Ramos and Paulino Villarreal. Additionally, Cecilia Williamson and KC Cushman reviewed polygon quality and made necessary corrections. Image data occurred on August 1, 2020, utilizing a DJI Phantom 4 Pro at a resolution of 4cm per pixel. A total of 2454 polygons were manually delineated, encompassing insightful metrics like crown completeness and liana load.The 2023 crown mapping campaign, led by Vicente Vasquez and field technicians Pablo Ramos, Paulino Villarreal, involved quality revisions and corrections performed by Luisa Fernanda Gomez Correa and Vicente Vasquez. Image data collection occurred on September 29, 2022, utilizing a DJI Phantom 4 Pro drone at a 4cm per pixel resolution. The 2023 campaign integrated model 230103_randresize_full of the detectree2 model garden (Ball, 2023). Tree crown polygons were generated pre-field visit, with those attaining a field validation score of 7 or higher retained as true tree crowns.The data collection forms are prepared using ArcGIS field maps. The creator of the data forms uses the spatial points from the trees in the ForestGeo 50-ha censuses to facilitate finding the tree tags in the field (Condit et al., 2019). The field technicians confirm that the tree crown is visible from the drone imagery, they proceed to collect variables of interest and delineate the tree crown manually. In the case of the 2023 field campaign, the field technicians were able to skip manual delineation when the polygons generated by 230103_randresize_full were evaluated as true detection.The improved version of the 2023 and 2021 crown map data collection takes as input the raw crown maps and the globally aligned orthomosaics to refine the edges of the crown. We use the model SAM from segment-anything module developed my Meta AI (Krillov, 2023). We adapted the use of their instance segmentation algorithm to take geospatial imagery in the form of tiles. We inputted multiple bounding boxes in the form of CPU torch tensors for each of the files. Furthermore, we perform several tasks to clean the crowns and remove the polygons overlaps to avoid ambiguity. This results in a very well delineated crown map with no overlapping between tree crowns. Despite our diligent efforts in detecting, delineating, and evaluating all visible tree crowns from drone imagery, this dataset exhibits certain limitations. These include missing tags denoted as -9999, erroneous manual delineations or instance segmentation of tree crown polygons, duplicated tags, and undetected tree crowns. These limitations are primarily attributed to human error, logistical constraints, and the challenge of confirming individual tree crown emergence above the canopy. In numerous instances, particularly within densely vegetated areas, delineating polygons and assigning tags to numerous small trees posed significant challenges.MetadataThe dataset comprises four sets of crown maps bundled within .zip files, adhering to the naming convention MacroSite_plot_year_month_day_crownmap_type. As an illustration, a sample file name follows the structure: BCI_50ha_2020_08_01_improved.For a comprehensive understanding of variable nomenclature within each shapefile, exhaustive details are provided in the file named variables_description.csv. Additionally, our dataset incorporates visualization figures corresponding to both raw and refined crown maps.The raw crown maps contain:A GeoTiff-formatted raster image reflecting the image acquisition date during field data collection.The tiles folder housing all tiles utilized for instance segmentation.The most recent version of the raw crown map manually revised and retaining its original naming scheme.A reformatted iteration of the raw crown map, involving column renaming and the reprojection of its coordinate reference system.The improved crown maps contain:"_crownmap_segmented.shp" version: This subproduct has all polygons segmented via the SAM model from the segment-anything process."_crownmap_cleaned.shp" version: This subproduct features one polygon allocated per GlobalID, specifically the one with the highest segment-anything score."_crownmap_avoidance.shp" version: This subproduct is devoid of any overlapping polygons."_crownmap_improved.shp" version: The outcome of the instance crown segmentation workflow, incorporating all original crown map fields.Author contributionsVV wrote the code for standardized workflow for processing, alignment, and segmentation of the tree crowns. MG and MH led the drone imagery collection. HCM conceived the study, wrote the grant proposals to obtain funding, and supervised the research.AcknowledgmentsVicente Vasquez and KC Cushman created the field map forms and coordinated the 2023 and 2021 crown map field campaign. Milton Solano assistance with the ArcGIS platform. Field technicians Pablo Ramos, Paulino Villareal, and Melvin Hernandez delineated and evaluated tree crown polygons. Luisa Gomez-Correa and Cecilia Williamson assisted with quality assurance and quality control after field data collection. Milton Garcia and additional interns in the Muller-Landau lab assisted with drone data collection. Funding and/or in-kind support was provided by the Smithsonian Institution Scholarly Studies grant program (HCM), the Smithsonian Institution Equipment fund (HCM), Smithsonian ForestGEO, the Smithsonian Tropical Research Institute.ReferencesBall, J.G.C., Hickman, S.H.M., Jackson, T.D., Koay, X.J., Hirst, J., Jay, W., Archer, M., Aubry-Kientz, M., Vincent, G. and Coomes, D.A. (2023), Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R-CNN. Remote Sens Ecol Conserv. 9(5):641-655. https://doi.org/10.1002/rse2.332Condit, Richard et al. (2019). Complete data from the Barro Colorado 50-ha plot: 423617 trees, 35 years [Dataset]. Dryad. https://doi.org/10.15146/5xcp-0d46Garcia, M., J. P. Dandois, R. F. Araujo, S. Grubinger, and H. C. Muller-Landau. 2021b. Surface elevation models and associated canopy height change models for the 50-ha plot on Barro Colorado Island, Panama, for 2014-2019. . In Smithsonian Figshare, edited by S. T. R. Institute. https://doi.org/10.25573/data.14417933Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A. C., Lo, W.-Y., Dollár, P., & Girshick, R. (2023). Segment Anything. arXiv preprint arXiv:2304.02643.Scheffler D, Hollstein A, Diedrich H, Segl K, Hostert P. AROSICS: An Automated and Robust Open-Source Image Co-Registration Software for Multi-Sensor Satellite Data. Remote Sensing. 2017; 9(7):676.

Maanpeite 2 m 2022 -aineisto kuvaa maanpeitettä valtakunnallisesti. Aineisto on rasteripohjainen ja sen pikselikoko on 2 m. Aineistossa on keskitytty etenkin pinnoitetun (läpäisemättömän) ja pinnoittamattoman (läpäisevän) maanpeitteen tarkkaan erottamiseen, johon on hyödynnetty Scalgon kehittämää tekoälyyn (AI U-NET) pohjautuvaa NDVI-ilmakuvatulkintaa. Tuotannossa on hyödynnetty Maanmittauslaitoksen ilmakuvia ja maastotietokantaa sekä Väyläviraston Digiroad-aineistoa.

Aineistossa on seuraavat maanpeiteluokat:

Pikselin arvo / pixel value - Luokan selite / class value

• 111 päällystetty tie / paved road (Digiroad)
• 112 päällystämätön tie / unpaded road (Digiroad)
• 120 rakennus / building (Maastotietokanta)
• 130 muu vettä läpäisemätön pinta / other impervious land (AI)
• 211 pelto / field (Maastotietokanta)
• 212 muu avoin matala kasvillisuus / shallow-vegetation (AI)
• 220 korkea kasvillisuus / dense-vegetation (AI)
• 310 avokalliot / bare-rock (Maastotietokanta)
• 410 paljas maa / bare-land (AI)
• 510 vesi / water (Maastotietokanta)

Aineistoa on jatkojalostettu Suomen ympäristökeskuksessa (Syke) sisältämään tiedon kasvillisuuden korkeudesta Suomen metsäkeskuksen valtakunnalliseen laserkeilauksella tuotetettuun latvusmalliin perustuen. Jatkojalostuksen myötä syntyneitä uusia luokkia ovat:

• 213 kasvillisuus alle 2 m / vegetation below 2 m
• 231 kasvillisuus 2-5 m, sis. sähkölinjat / vegetation 2-5, inc. power lines
• 232 kasvillisuus 5-10 m, sis. sähkölinjat / vegetation 5-10, inc. power lines
• 233 kasvillisuus 10-15 m, sis. sähkölinjat / vegetation 10-15, inc. power lines
• 234 kasvillisuus 15-20 m, sis. sähkölinjat / vegetation 15-20, inc. power lines
• 235 kasvillisuus yli 20 m, sis. sähkölinjat / vegetation over 20 m, inc. power lines

Kaikki luokat excel-taulukkona: https://geoportal.ymparisto.fi/meta/julkinen/dokumentit/land_cover_classes.xlsx

Aineistot on tuotettu Scalgon ja Syken yhteistyönä osana Mammutti- ja TIIMA-hankkeita ja niitä voidaan käyttää paikkatietoanalyysien lisäksi myös taustakarttana.

Aineistot kuuluvat Syken avoimiin aineistoihin (CC BY 4.0).

Katselua varten esim. QGIS:ssä tai ArcGIS:ssä on käytössä WMS-rajapinta. Aineiston esittelyä varten on lisäksi laadittu karttapalvelu. Aineiston lataaminen esim. analysointia varten tehdään puolestaan WCS-rajapinnan kautta. Lataus on nopeinta tehdä selaimella koordinaattirajauksella (ETRS-TM35FIN) seuraavien esimerkkien mukaisesti:

Maanpeite 2m 2022 -aineiston lataaminen Helsingin Viikistä GeoTIFF-formaatissa https://paikkatiedot.ymparisto.fi/geoserver/syke_maanpeitescalgo/wcs?service=WCS&version=2.0.1&request=GetCoverage&coverageId=Maanpeite2022Scalgo&SUBSET=N(6677650,6679380)&SUBSET=E(389470,392190)&subsettingcrs=http://www.opengis.net/def/crs/EPSG/0/3067&format=image/tiff

Maanpeite 2m 2022 kasvillisuuden korkeudella -aineiston lataaminen Helsingin Viikistä GeoTIFF-formaatissa https://paikkatiedot.ymparisto.fi/geoserver/syke_maanpeitescalgo/wcs?service=WCS&version=2.0.1&request=GetCoverage&coverageId=Maanpeite2022kasvkorkScalgo&SUBSET=N(6677650,6679380)&SUBSET=E(389470,392190)&subsettingcrs=http://www.opengis.net/def/crs/EPSG/0/3067&format=image/tiff

Kuvaustekniikka löytyy tästä zip-paketista QML- ja SLD-formaateissa: https://wwwd3.ymparisto.fi/d3/wcs_kuvaustekniikka/syke_maanpeitescalgo.zip

Lisätietoja:

Mapping Land cover for Finland.pdf (Scalgo): https://geoportal.ymparisto.fi/meta/julkinen/dokumentit/land_cover_scalgo.pdf

Maanpeiteaineiston jatkojalostaminen (Syke): https://geoportal.ymparisto.fi/meta/julkinen/dokumentit/maanpeiteaineiston_jatkojalostaminen_syke.pdf

The data represents the land cover of the whole Finland. The data are in raster-format having pixel size of 2 m. The data are especially focused on the accurate separation of paved and unpaved land cover, for which the artificial intelligence-based (U-NET) NDVI orthophoto interpretation developed by Scalgo has been used. The data has been further processed at the Finnish Environmental Institute (Syke) in classifying high and low vegetation based on the canopy height model of the Finnish Forestry Center. In addition to this, the existing spatial data from the Finnish National Land Survey and the Finnish Environment Institute was used to supplement the land cover data.

The final dataset consists of 16 land cover classes. The data was produced in cooperation with Scalgo and Syke as part of the Mammutti- and TIIMA-projects and it can be used for various geospatial analysis and also as a background map.

More information: Mapping Land cover for Finland(Scalgo): https://geoportal.ymparisto.fi/meta/julkinen/dokumentit/land_cover_scalgo.pdf

Maanpeiteluokkien muodostamiseen on käytetty useita eri paikkatietoaineistoja, tärkeimpinä Maanmittauslaitoksen ilmakuvat (2017–2022) ja Maastotietokanta (2023), Metsäkeskuksen valtakunnallinen latvusmalli (0.5 p/m2 & 5 p/m2, 2009–2022) ja Väyläviraston digiroad (2023).

Maanpeiteaineiston tuotantovuosi on 2022, mutta käytettyjen lähtöaineistojen tuotanto on ajoittunut aineistosta riippuen vuosille 2009–2023, esim. lähtötietona käytetyt ilmakuvat ovat vuosilta 2017-2022. Ilmakuvat otetaan valtakunnallisesti 3 vuoden sykleissä (Lapissa vielä harvemmin), joten osa käytetyistä kuvista on useamman vuoden takaa ja maanpeite saattaa olla jo muuttunut. Tämän lisäksi kuvausajankohdat vaihtelevat kevät ja kesäkuvauksien välillä, joka on voinut aiheuttaa virheitä esimerkiksi kasvillisuuden tunnistamiseen. Laserkeilausaineisto perustuu osittain vanhempiin 2009–2019 keilauksiin, joten osa kasvillisuudesta voi olla tulkittu virheellisesti puustoiseksi alueeksi, vaikka alue voi olla todellisuudessa hakattu.

Metsäkeskuksen latvusmallia ei ollut saatavilla koko Suomesta (puuttui esim. Ahvenanmaalta), joten paikoin myös jatkojalosteessa on esitetty tekoälytulkintaisia kasvillisuusluokkia (212 ja 220).

Several different spatial datasets have been used to create the land cover classes, the most important being aerial orthophotos (2017 – 2022) and Topographic Database (2023) of the Finnish National Land Survey as well as the Digiroad from the Finnish transport infrastructure Agency (2023) and canopy height model of the Metsäkeskus (0.5 p/m2 & 5 p/m2, 2009–2022).

The production year of the land cover data is mainly 2022, but part of the input data represents years 2009–2023. Aerial orthophotos are taken nationally in 3-year cycles (even less often in Lapland), so some of the photos used are from several years ago and the land cover may have already changed. In addition to this, the shooting times vary between spring and summer shootings, which could have caused errors, especially in the identification of vegetation. The laser scanning data is partially based on older scans from 2009 to 2019, so some of the vegetation may have been mistakenly interpreted as high vegetation, even though the area may actually have been cut down.