Hyperspectral imagery for Environmental urban Planning – HYEP

WP1 is dedicated to data acquisition (ground data) and the definition and creation of a morpho-spectral database.
WP2 concentrates on the development of pre- and post-processing tools, taking into account the selected morpho-spectral data defined in WP1.
WP3 tackles the assessment of sensor capabilities for urban applications, according to the following topics: impervious soils, urban vegetation, roof and infrastructure material, and urban wetland. This assessment is planned in comparison with other, planned, hyperspectral sensors (EnMap, Shalom or PRISMA) and existing multispectral sensors like Pleiades or future ones like Sentinel-2.
WP4 deals with the integration of the previous results with the following two goals:
• HYPXIM mission: to provide inputs for mission technical specifications;
• End user or local authorities’ expectations: to provide insight for planning purposes (Trame verte, Plan Climat Energie Territoire, Schémas Régionaux de Cohérence Ecologique (SRCE), etc).

Tests of atmospheric algorithms; creation of a spectral library adapted to urban areas; defnition of adapted methods: fusion, unmixing, classification. Outcomes: Spectral data library for the scientific community and methods implementations on shared platforms. Complete design on Kaunas (Lituania), contacts with Kaunas University and urban planners. Proposal of a thematic summer school in 2017.

Creation of a spectral library; Realisation of specific deliverables usable by planners that have been setup in a collaborative way.

Roussel G., et al., 2016, Comparaison de deux méthodes de correction atmosphérique à différentes résolutions pour des applications de classification d’images hyperspectrales. SFPT Section Hyperspectrale Grenoble, May 2016. = first results of a novel atmospheric correction technique
Zinko S., Lampert T, Gançarski P et Weber C, 2016, HYEP: des bases de données spectrales à l’usage de la communauté. SFPT Section Hyperspectrale Grenoble, May 2016. = first results of varliability experiments using the spectral database
Ourghemmi W. et al., 2016, Generation of urban objects spectral database using laboratory hyperspectral imager in Kaunas city (Lithuania). SFPT Section Hyperspectrale Grenoble, May 2016. = description of the development of the work carried out in Kaunas (Lituania)

World population gathers in urban areas restructuring at the local and regional levels the urban territories. These spaces are characterized by paradoxical process on the one hand a sprawl at the expense of natural and agricultural ecosystems and on the other hand a densification of the urban fabric. These combined processes have impacted the climate characteristics of local and regional scales (Shafri et al. 2012), as well as biotic and abiotic processes (Voogt and Oki, 2003). Needs for more information followed these developments, reinforced by the emergence of logics of sustainable development at different spatial scales. Current satellite data provide limited information, features urban complicating their use, by the strong internal dynamic, spatial heterogeneity of the elements, geometric shapes (horizontal and vertical), variety of materials and presence of shadow. Mapping of surfaces, state of vegetation, follow-up of the ageing of materials, characterization of plant biodiversity (Miller & Small, 2003) covering strong fields of investigation.
Various works show the contribution of hyperspectral compared the multispectral imagery. For Platt and Goetz (2004) the performance of the classification of urban surfaces obtained with images acquired by the spectro-imager Aviris are superior to those obtained with Landsat ETM + data. More recently. Tan and Wang (2007) assess the gain made by the hyperspectral instrument CHRIS/PROBA on the classification and use of urban areas from ASTER (3 bands VISNIR) data. Several authors (Chen, 2008) illustrate the limitations of multi-spectral sensors for the characterization of the impervious surfaces from hyperspectral imagery. The identification of plant species and vegetation (Mc Kinney, 2002) health survey can be an asset for the greenways.
The richness and complexity of materials and forms in urban centres require specific spectral characteristics ranging from the visible to the SWIR (< 2.5 µm). Herold et al. (2003), has shown that spectrometers provides a high continuous spectral domain spectral resolution better identify (compared to usual sensors) and spatial information on soils, vegetation and materials (Chen, 2008, Pascucci et al 2010).
The average size of urban objects lies between 10 and 20 m (Cutter et al. 2004). Also, a resolution of 5 m or better is considered necessary for a representation of urban objects (buildings, roads), vegetation (Jansen et al. 2012), or the planning of the territory (Wania & Weber, 2007). The contribution of the EnMAP data in a context of development and urban planning was recently studied by Heldens et al. (2011), but the spatial resolution of 30 m provides inadequate information at the district level.
Taking into account geometric environment characteristics, the capacity of an instrument as HYPXIM combining high-resolution spectral (hyperspectral imaging) and spatial (panchromatic Imaging) information should allow to study objects of less than 5 m with a capacity to cover the spectral range 0.4 to 2.5 µm. Hyperspectral imaging may thus lift some of the identified locks but requires better understanding of the specificities of urban system and its elements (geometry spectral properties). This project is designed to justify the mission HYPXIM (french hyperspectral sensor) compared to the existing and future missions, define ways of treatments to use these images and constitute a morpho-spectral database adapted to these various missions.