Mosaicked ProductsMosaicking is the capability of assembling L1B, geocoded or DEMs images or strips into a common grid, in order to generate a large-scale map. When assembling geocoded images, mosaicking is a relatively simple process. However, images that have to be assembled generally do not perfectly match: an overlap between contiguous frames, or a gap, is present. Therefore there is the need to blend these data together (by a proper feathering on the overlapping zones), and choose which portion must be discarded, or how to fill the gaps. In order to have mosaicked products with the high resolution as the inputs one, no undersampling process is applied to the input images to mosaic, so the combination of a large area with a high resolution produces very large products. Nevertheless, it could be possible to define a mosaic product with a very large coverage (say, the coverage of an entire state), provided that the input images are adequately undersampled before being assembled. Mosaiking products can be originated starting from L1B or L1C or L1D products and also DEM (and associate height error map) and can start from product acquired in similar or different modes. In case of assembling Level 1B images, the resulting mosaiked product is kept into the same ground range/azimuth projection. When assembling geocoded products (L1C, L1D, DEM), the input tiles to be mosaicked must share the following features:
and the mosaiked product projection depends on the projection of the input products.
In the first case, the resolution is the same of the less resoluted tile (i.e. the largest value) for input tiles of L1B, L1C, L1D while the opposite (less resoluted images are interpolated to the finest spacing) is used when the input is a DEM. In the second case, resolution can be selected as a multiple of the input tiles resolution. When assembling L1B images, one of the following circumstances could be verified:
In the cases 1 and 2, as the input tiles share the same azimuth and range grid, the assembling process is similar to that one detailed for the geocoded product. In the cases 3 and 4, mosaicking among scenes that could be carachterized by different orientation angles is requested; in such case, after having identified the master tile (the azimuth and range directions of the output mosaick must be referred to), slave tiles are at first regridded on the same grid of the master one by a very coarse approach based on correlation to the master orbit and a warp model derivation and application; such approach will be allowed only supposing that input tiles had been processed with orbital data derived from POD facility (hence, not processed with state vectors embedded into the downlinked RAW data). Finally, the mosaicking process is completed applying feathering on the overlapping zones. However, feathering is a configurable parameter of the Processing Request File and can be turned off during the CalVal activities. In the case of mosaicking of GTC, only the SAR image is included into the output product, that is the GIM layer is not mosaicked. In the case of mosaicking of DTM, the mosaicked HEM is also included into the output product. |
|
Speckle Filtered ProductsThe Higher Level Speckle Filtered Product deals with the improvement of the radiometric resolution of the SAR images by means of the reduction of the intrinsic multiplicative-like speckle noise. Speckle is a multiplicative noise-like characteristic of coherent imaging systems (such as the SAR), which manifests itself in the image as the apparently random placement of pixels which are noticeably bright or dark. In fact the speckle is a real electromagnetic effect that originates from the constructive or destructive interference (within a resolution cell) of multiple returns of coherent electromagnetic radiation. The Speckle Filtering processor tries to cope with any generic application that could benefit of a speckle noise suppression, improving the radiometric resolution of the SAR Standard images thus allowing a better estimation of the radiometric quantities and minimizing, whenever possible, side effects (degradation of the spatial resolution, artefacts, feature alteration). As such, the Speckle Filtered Product is derived by post-processing of the SAR Standard L1A or L1B products. The filtered product is formally equivalent to a L1B standard product and may be further processed by the SAR Standard chain. Many types of filters are allowable, belonging to various classes (Non-Adaptive, Adaptive MMSE, Adaptive MAP, Morphological). Speckle filtered products can be generated starting from L1B products and hence originating a product at same level. |
|
Interferometric ProductsSynthetic Aperture Radar interferometry is an imaging technique for measuring the topography of the surface and its changes over time. A radar interferometer is formed relating the signals from two spatially separated antennas; the separation of the two antennas is called baseline. COSMO-SkyMed constellation can be used for interferometric applications, which allow to produce three-dimensional SAR images by combining two radar images of the same point on the ground (one “master” and the other one is the “slave” image) obtained from slightly different incidence angles. COSMO-SkyMed constellation offers two different possibilities to achieve the interferometric baseline, namely: (1) the “tandem-like” interferometry configuration (i.e. "one-day" of relative phasing between the satellite couple), and (2) the “tandem” interferometry configuration (i.e. two satellites flying in close proximity). As such, the Interferometric products are derived by processing SAR Level 1A co-registered data, taken in any acquisition mode (except PingPong), to generate in automatic way the following product classes:
Some constraints exists in the generation of the interferometric products, e.g. polarimetric products or products having different trasmit-receive polarizations, products acquired on different subswaths or different look side or orbit direction. |
|
Co-registered Products
Two different images covering the same area can be made superimposable by means of the coregistration which is the process of lining-up two images, one “master” and the other one is the “slave” image, such to fit exactly on top of each other without adding artifacts in image intensity and phase components. The input images are co-registered using the master as reference. Co-registered images can be taken from simultaneous illuminations of the same scene at different frequencies (multifrequency images), from acquisitions taken at different time using different sensors, from multiple passages of the same satellite (multi temporal images). In general, images have different geometry, thus to be superposed the slave image must be re-sampled into the master geometry. The images may be fully or only partly overlapped and even more than two images can be co-registered at the same time.
|
|
DEM ProductsThe Digital Elevation Model (DEM) products are derived by mean interferometric processing of the SAR Level 1A coregistered products, in any acquisition mode except Polarimetric (i.e. PingPong), in automatic way.
|