La densité de bâti est calculée par maille de 150 mètres de côté et sur la base d'une extraction du bâti à partir d'imagerie très haute résolution spatiale (1.5m) SPOT 6/7, pour les années 2015 à 2019.

Le projet ArtiSols s’inscrit dans une démarche d’évaluation et de qualification du processus d’artificialisation des sols en région Occitanie. L’accentuation de ce phénomène face à l’accroissement des enjeux liés au logement, à l’alimentation des populations, à la consommation énergétique et des ressources naturelles pose la question des impacts sur les sols, essentiels à la régulation des écosystèmes mais également à la production agricole. C’est pourquoi la caractérisation de l’artificialisation constitue un point de départ dans la maîtrise de politiques d’aménagement adaptées aux enjeux des territoires. Ce projet régional s’articule de façon complémentaire avec le projet national piloté par le Ministère de la Transition Écologique et Solidaire (DGALN) dans le cadre du Plan Biodiversité (Action 7, visant la mise en place d’un observatoire national sur l’artificialisation des sols).

Les cartographies des espaces bâtis sur la région Occitanie résultent d'une extraction automatique par méthode d'apprentissage profond (deep learning) à partir d'imagerie très haute résolution spatiale (1.5m) SPOT 6/7, pour les années 2015 à 2019. Fichiers fournis sous forme vectorielle. (2021-09-09)

Les taches artificialisées sont calculées sur la base d'une extraction du bâti à partir d'imagerie très haute résolution spatiale (1.5m) SPOT 6/7, pour les années 2015 à 2019. Deux distances de connexion sont proposées, à 50m et 100m.

Full hyperspectral VNIR-SWIR ENVI standard image obtained from the coregistration of both VNIR and SWIR ones through a signal aggregation process that allowed to obtain a synthetic VNIR 1.6 m spatial resolution image, with pixels exactly corresponding to natif SWIR image ones. First, a spatially resampled 1.6 m VNIR image was built, where output pixel values were calculated as the average of the VNIR 0.8 m pixel values that spatially contribute to it. Then, ground control points (GCP) were selected over both images and SWIR one was tied to the VNIR 1.6 m image using a bilinear resampling method using ENVI tool. This lead to a 1.6 m spatial resolution full VNIR-SWIR image.

This dataset provides georeferenced polygon vectors of individual tree canopy geometries for dryland areas in West African Sahara and Sahel that were derived using deep learning applied to 50 cm resolution satellite imagery. More than 1.8 billion non-forest trees (i.e., woody plants with a crown size over 3 m2) over about 1.3 million km2 were identified from panchromatic and pansharpened normalized difference vegetation index (NVDI) images at 0.5 m spatial resolution using an automatic tree detection framework based on supervised deep-learning techniques. Combined with existing and future fieldwork, these data lay the foundation for a comprehensive database that contains information on all individual trees outside of forests and could provide accurate estimates of woody carbon in arid and semi-arid areas throughout the Earth for the first time.

Hyperspectral data were obtained during an acquisition campaign led on Toulouse (France) urban area on July 2015 using Hyspex instrument which provides 408 spectral bands spread over 0.4 – 2.5 μ. Flight altitude lead to 2 m spatial resolution images. Supervised SVN classification results for 600 urban trees according to a 3 level nomenclature: leaf type (5 classes), family (12 and 19 classes) and species (14 and 27 classes). The number of classes differ for the two latter as they depend on the minimum number of individuals considered (4 and 10 individuals per class respectively). Trees positions have been acquired using differential GPS and are given with centimetric to decimetric precision. A randomly selected subset of these trees has been used to train machine SVM and Random Forest classification algorithms. Those algorithms were applied to hyperspectral images using a number of classes for family (12 and 19 classes) and species (14 and 27 classes) levels defined according to the minimum number of individuals considered during training/validation process (4 and 10 individuals per class, respectively). Global classification precision for several training subsets is given by Brabant et al, 2019 (https://www.mdpi.com/470202) in terms of averaged overall accuracy (AOA) and averaged kappa index of agreement (AKIA).

Cette donnée raster résulte d'une classification par méthode d'apprentissage profond à partir d'imagerie très haute résolution spatiale (1.5m) SPOT 6/7. Des post-traitements ont été effectués afin de mieux caractériser les classes relatives à l'artificialisation.

ObjectivesThe hyperspectral images (HI) is at the moment still too poorly considered; nevertheless its specificities make a weighty auxiliary for the monitoring of the elements of the urban area. The HYEP project has for objective to propose a panel of methods and processes designed for hyperspectral imaging. We take into account other existing sensors in order to compare the performances. If the IH is complementary to the sensors HRS and VHRS due to its better radiometric richness, it allows to identify and to characterize the natural or anthropogenic elements in a complementary perspective. To this end methods for the extraction of information had to be adapted, created even. The methodological part of the project articulated at the same time in the solidification of the current approaches and the design of new methods. Results have been presented along the project duration to scientific community and local authorities. One of the milestones of the works was the comparison of the results to various spatial resolutions to specify the contribution of such a hyperspectral sensor with regard to those existing or to come. Data and data processing Methods: existing or adapted The methods were chosen among all the existing methods by adapting them to the signal, among spectral ranges and to the characteristics of urban areas. Since data arose from airborne platforms, the first developments were realized to counter the effects of the atmosphere on the IH (atmospheric correction - 3 tested methods) and a database of spectral signature for diverse elements of land use in town (roofs, roads, the vegetation etc.) was established. It allowed encircling better the spectral values of materials. It was set up based simultaneously on the literature, in situ and laboratory measurements. Its contributions in various classification processing were tested. Methods for information, extraction, pansharpening or classification purposes were used for various spatial and spectral resolutions to highlight its interest towards other sensors and also its benefits for a spatial mission. Classification and unmixing methods have been adapted and spatial pattern indicators for urban areas defined.Outcomes- 3 atmospheric correction methods have been tested; it leads to a specific code development by ONERA.- Methods adaptation : pansharpening and unmixing- Transfer: a complete design of the study has been transfered to Kaunas (Lituania) teams- Algorithms: Depository on http://openremotesensing.net/- One of the major results is the extraction and the identification of photovoltaic panel- CNRS Summer school 2017Scientific productionThe team has presented at ISPRS Geospatial Week 2015, GeoHyper (2015), Jurse 2017, IGARSS 2016-2017, SFPT or workshops TEMU, AFIGEO and to GdR Session (MaDics and ISIS) or within the framework of the Hypxim mission. The team organized special conference sessions at the national level, SFPT hyperspectral (2016) and international level IGARSS 2018 and WHISPERS 2018. A thematic CNRS summer school (2017 August 28 - September 1st - 25 participants) has been set up.The project gave rise to 10 publications Rang A and 38 communications, 4 chapters and a special issue for the RemoteSensing journal.