October 28, 2014

Level 1 product

simulation of Level 1 with Formosat-2


Muret data set (Southwest France)	Muret data set (Southwest France)	Cestas data set (Southwest France)
These Images are simulations of Venµs level 1 products (click on image for full resolution), from Formosat-2 data sets. These data are orthorectified, registered, calibrated and converted to Top of atmosphere reflectance. Note the excellent registration between successive images. For each data set, the same color table is used for the whole time series.

Description of level 1 product

Product Content

The main feature of Venµs products is the repetitivity of observations; it is also the main driver of Venµs product definition. Since the basic use of Venµs data is done with multiple images at different dates, the basic level 1 products must be geometrically registered and radiometrically calibrated. As a result, the Venµs level 1 products will be equivalent to Spot level 3 products (that do not follow the CEOS product level definitions).

The Venµs level 1 will thus provide:

Geolocated top of atmosphere reflectances (possibility to use a different geographic projection for each site, but only one per site), with a subpixel (objective 3m) multi-date registration.
A cloud mask at a fine resolution

The level 1 geometric resolution is 5 m.

Algorithm description

The level 1 processing combines radiometric and geometric corrections as follows:

See below for more information

Radiometric correction

The radiometric correction concerns:

Normalisation: dark currents and detector sensitivity corrections
Restoration: deconvolution for MTF correction (frequential processing) and if necessary (noise criteria) denoising based on a wavelet processing.

Geometric correction

The geometric process aims at registrating all the images taken over a given site. To reach this objective, a well located reference image is used for each site. The correlation with this reference image allows correcting the physical geometric model of the processed image; it is done for the stereoscopic bands (620nm) only, since these two bands are located at both extremities of the focal plane.

The geometric model provides a viewing direction for each point in the image, combining orbit, attitude, datation and the camera viewing directions. All these parameters are known with a certain accuracy: "refining the geometrical model" means correcting errors in the model parameters using ground control points obtained via correlation with the reference image.

A Digital Elevation Model is necessary to compute the location of the intersection of the viewing direction with the earth surface.

Resampling at 5 m

The level 1 product is provided at 5m resolution in a cartographic projection selected by the PI of the site.

The resampling step combines 2 processes:

A geometric process computes the inverse grid that gives for each point of the output image its location in the camera focal plane.
A radiometric process computes for each point of the grid its radiometry using Splines interpolation functions.

TOA Reflectance

The Top Of Atmosphere reflectance is computed from numerical counts, knowing the camera sensitivity and the sun elevation at the center of the image.

Cloud Detection

The cloud detection is based on two criteria:

their elevation deduced from the slight parallax between B5/B6 stereoscopic bands,
their high reflectance in a blue band (443nm).

The cloud mask is provided on the same sampling grid as the level 1 image.

Reference image

The reference image is nominally a monospectral Venµs image (620nm) which is well located; this means that its physical geometric model has been refined. This refinement is done

either through DEM correlation according to the following process:

Venµs DEM production using B5/B6 (620nm) duplicated bands
Venµs DEM matching with a well located DEM (SRTM, scientific,...)
Geometric model refinement using homologous points deduced from the previous matching