Doc

About

TwinCity is the result of a co-design effort between Mines PARIS - PSL and the City of Marseille, as part of the Destination Earth project funded by ESA. The goal of this application is to provide city officials with high-resolution Earth observation data useful for their work : - Urban nature management - Urban planning - Energy - ...

"FAIR" Philosophy

TwinCity adheres to the FAIR principles and relies on open-source tools and recognized standards to enable users to access data as easily as possible and derive maximum added value from it. Two interfaces are offered: - A simplified web interface for users without specific skills in Geographic Information Systems (GIS). This application allows data visualization and extraction of useful statistics. - A QGIS project allows direct access to raw data via a GeoServer implementing OGC standards. This interface is intended for mapping experts wishing to directly exploit the data with the powerful tools offered by QGIS.

Data

The data provided are derived from satellite data (Sentinel 2 and Landsat 8) and LIDAR (IGN). Satellite data provide temporal changes over a long period (10 years) at a medium resolution (8 to 40 meters). Data cubes have been pre-extracted over the Marseille area and optimized for fast display. LIDAR data provide static high-resolution information captured in 2022.

Name with Icon Source Type Spatial Resolution Temporal Resolution and Extent
🌿 Vegetation Index (NDVI) Sentinel 2 temporal 8 meters 2014-2017 (1/month)
💧 Moisture Index (NDMI) Sentinel 2 temporal 15 meters 2017-2024 (1/month)
🌡 Surface Temperature Landsat 8 temporal 46 meters 2014-2024 (17/year)
⛰️ Surface Model IGN LIDAR raster 20 cm 2022
☀️ Horizontal Irradiation IGN LIDAR raster 1 meter 2022
☀️ Tilted Irradiation IGN LIDAR raster 1 meter 2022
📊 Land Cover IGN LIDAR categorical 20 cm 2022
🌲 Tree Detection IGN LIDAR vector - 2022

🌿 Vegetation Index

The Vegetation Index (NDVI) is a dimensionless index between -1 and 1. It distinguishes vegetated areas (>0.2) from urban areas (<0.2) and monitors the health of vegetation. It is derived from Sentinel 2 data.

💧 Moisture Index

The Normalized Difference Moisture Index (NDMI) detects moisture levels in vegetation. It is a dimensionless index between -1 and 1. It is derived from Sentinel 2 data.

🌡 Surface Temperature

Surface temperature, in degrees Celsius, is derived from the infrared bands of Landsat 8 data. It is a value different from air temperature but provides interesting information on the heat difference between various locations.

⛰️ Surface Model

The surface model provides the elevation at any point, in meters, with a resolution of 20 cm. It is derived from IGN LIDAR data.

☀️ Solar Irradiation

The total annual solar irradiation (in kWh per meter) is calculated for all roofs in the city at a resolution of 1 meter, using CAMS Radiation data (Heliosat-4 model), taking into account shading calculated based on the high-resolution surface model above. Irradiation is calculated for both horizontal surfaces and tilted surfaces (following the roof slope).

📊 Land Cover

Land classification is derived from IGN LIDAR data. It is provided at a resolution of 20 cm.

🌲 Individual Trees

Individual trees are detected from IGN LIDAR data by searching for local maxima in the surface model in areas classified as "high vegetation" in the land cover. This layer is available only through the QGIS project.

Web Interface

This section describes the features of the Web Interface.

Main Screen

Main Screen

  1. The top bar allows navigation to documentation pages, access to the QGIS project, and changing the interface language.
  2. In a banner on the left side, a dropdown button allows choosing the data layer to display on the map.
  3. Below this button, a panel provides a detailed description of the current layer and allows downloading the raw data.
  4. The main map displays a base map and the current data layer.
  5. Map options allow choosing the opacity of the data layer and the base map from:
  6. An Open Street Map
  7. IGN orthophotos from 2023
  8. IGN orthophotos from 2020
  9. IGN orthophotos from 2017
  10. IGN orthophotos from 2014
  11. For temporal data, this button allows choosing between:
  12. Displaying data for a given month
  13. A map of the relative evolution of this data over the entire period
  14. For temporal data, this timeline allows selecting the date of the data to display.
  15. The map drawing tools allow extracting statistics for specific areas.

Statistics Extraction

Statistical calculation is done by drawing one or more areas of interest on the map using the tools located to the left of it (rectangle and polygon [8]). As soon as at least one area is selected, statistical data appears below the map.

Temporal Data

Here is an example of a result for a temporal layer (NDVI):

drawing

  1. Two distinct polygons are drawn on the map.
  2. The evolution of the data over time is extracted for each polygon. It is possible to combine the evolutions of other layers on the same graph.
  3. The normalization button removes strong seasonal variations by calculating the difference in values relative to the average over the entire city at the same time.
  4. The curves are displayed on a dynamic graph that can be saved in png format.
  5. The corresponding time series can be exported in CSV format.
  6. A selection button allows choosing the spatial statistic to calculate among: average, minimum, maximum.
  7. The chosen statistic is calculated for each pixel of the polygon.
  8. The result can be exported in GeoTiff format, ready to be imported into QGIS.

Raster Data

For non-temporal (raster) data, spatial statistics are presented as follows:

drawing

  1. The roof of a building is drawn on the horizontal irradiance layer.
  2. Global statistics are displayed for this area:
  3. Minimum value
  4. Maximum value
  5. Average
  6. Sum over the area
  7. The distribution of values is displayed. It can be exported in png format.

QGIS Project

The QGIS project is available at this address. It should be opened with a recent version of the open-source software QGIS.

The provided QGIS project is already configured to automatically connect to the GeoServer. QGIS retrieves data at runtime as needed.

Here is the interface of the opened project:

drawing

  1. The left panel provides a list of layers that the user can display/hide. The layers are organized into 4 groups:
  2. Vector data
  3. Temporal rasters
  4. Static rasters
  5. Base maps: satellite view (MapBox) or streets (OpenStreetMap)
  6. The temporal controller allows the user to select the date to display (for temporal rasters only).
  7. The map displays all selected layers.
  8. The right panel displays the toolbox, with a subset of the many operations provided by QGIS.