[OSM-Science] Paper "Leveraging OSM and GEOBIA to Create and Update Forest Type Maps "

[mbrauchler at posteo.de]

Hello to all list members,

I am writing to share with you our recently published paper on forest
type stratification titled ""Leveraging OSM and GEOBIA to Create and
Update Forest Type Maps". We use OSM landuse=forest + leaf_type=mixed
information as bounding areas for region growing segmentation on aerial
imagery and the main idea is to derive additional leaf type information
on forests, which are tagged with leaf_type=mixed, especially on those
which have a large area. We also classify the segments based on training
data derived from OSM. We therefore only use OSM and open data from the
Grand Duchy of Luxembourg as well as GRASS GIS and R to carry out the
workflow, which makes the approach reproducible and transferable. 

Brauchler, M.; Stoffels, J. Leveraging OSM and GEOBIA to Create and
Update Forest Type Maps. ISPRS Int. J. Geo-Inf. 2020, 9(9), 499;
https://doi.org/10.3390/ijgi9090499 
https://www.mdpi.com/2220-9964/9/9/499 [1]

As this is my first scientific contribution, I am happy to start a
discussion and to get in touch with others who might be interested in
the topic or who are working on similar projects.
Kind regards,
Melanie

Abstract
Up-to-date information about the type and spatial distribution of
forests is an essential element in both sustainable forest management
and environmental monitoring and modelling. The OpenStreetMap (OSM)
database contains vast amounts of spatial information on natural
features, including forests (landuse=forest). The OSM data model
includes describing tags for its contents, i.e., leaf type for forest
areas (i.e., leaf_type=broadleaved). Although the leaf type tag is
common, the vast majority of forest areas are tagged with the leaf type
mixed, amounting to a total area of 87% of landuse=forests from the OSM
database. These areas comprise an important information source to derive
and update forest type maps. In order to leverage this information
content, a methodology for stratification of leaf types inside these
areas has been developed using image segmentation on aerial imagery and
subsequent classification of leaf types. The presented methodology
achieves an overall classification accuracy of 85% for the leaf types
needleleaved and broadleaved in the selected forest areas. The resulting
stratification demonstrates that through approaches, such as that
presented, the derivation of forest type maps from OSM would be feasible
with an extended and improved methodology. It also suggests an improved
methodology might be able to provide updates of leaf type to the OSM
database with contributor participation.

 

Links:
------
[1] https://www.mdpi.com/2220-9964/9/9/499
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/science/attachments/20200825/5b4e6b59/attachment.htm>