# Relative / Absolute Orbit number interpretation

Hi,
Does the same relative /absolute orbit numbers between two S1 products (S1A + S1B) means that they have the same geometry (the same view angle and local incidence angle ) with respect to the scene?
The two tiles that I use both have relative orbit number 139 and they exactly overlap on each other. This made me believe that the two products must have the same SAR geometry. am I right?

In general I would appreciate any reading material regaring the two concepts as well as what happens behind the orbit correction in SNAP

The relative orbit (also called track) is the path that each of the S1 satellites passes over an area. Two images of the same relative orbit (e.g. 139) have the same incidence angle and look direction and can be coregistered for multi-temporal analysis.
The absolute orbit simply tells how often the satellite has passed this track since its launch. Therefore, this number is largely higher and also unique.

The application of an orbit file in SNAP is done to get the precise position of the satellite at the time of image acquisition. It is estimated during operation but can be determined more accurate after 2-3 weeks, because the position of the satellite varies inside the orbital tube. This exact position is required when you want to know the perpendicular baseline of an image pair which is the distance between the satellites inside the orbital tube. For Sentinel-1 it coarsely ranges between 5-50 meters (good for differential interferometry)

1 Like

Thank you very much for this clarification.
In the first part of your answer you mentioned that two images (with the same relative orbit) need to be co-registered ?
I was thinking that two images that are taken from the same look angle (geometry) do not need co-registration and applying â€śgeocodingâ€ť and â€śradiometric calibrationâ€ť on each of them them is enough to compare them and do further processing on them (lets say change detection). Am I making mistake here ?
I was thinking that two images that represent the same scene but are taken from different view angles do need co-registration for joint analysis. am I wrong ?

The overall angle is common, but slightly varies form image to image because of smaller differences in the sensor positions within the orbit. This can cause shifts in the image, especially when topography is strong in your image. In many cases, radiometric calibration and terrain correction are sufficient to have both images overlaid correctly (you would see the shifts in two-date RGB composites), but if you want to go sure, you coregister the images before you analyze the changes ans then you apply terrain correction as a last step. This is ultimately required for radar interferometry, but also advisable if you have subtle changes (e.g. along roads or single pixels) and you want to avoid resampling of your data which is involved in terrain correction that blurs your results. Also, in areas with strong topography, coregistration of images from the same track is more accurate, because it locally matches patterns of both images instead of simply stacking them.
Youâ€™ll have to check what is best for your data.

1 Like

Thank you very much indeed for the detail and clarification.
I am studying an area whose altitude variation is in the range ~[200,600] meters. I think therefore the effects of topography needs to be taken into account.
In the S1-Toolbox Tutorial (Time-series analysis with Sentinel-1) you mentioned a pre-processing scenario as follow :

Thermal Noise Removal â†’ Orbit correction â†’ Radiometric Calibration â†’ Terrain-correction â†’ LinearToDB â†’ Subset

Base on what you said in previous message I think that I need to modify the scenario as :

Thermal Noise Removal â†’ Orbit correction â†’ Radiometric Calibration â†’ Co-registration â†’ Terrain-correction â†’ LinearToDB â†’ Subset

Am I right?
In addition as the AOI covers only 1/15 area of a full GRD tile, I wonder is it OK to do the last step (subset cropping) at the beginning and then apply the pre-processing scenario on a small subset ?

Iâ€™d say yes to both questions

1 Like

The downside is that apparently co-registration can not be done in batch mode (using SNAP Graph builder). I have 15 dates to co-register. and I need to do the co-registration manually !

Coregistration does not only work pairwise.
You can coregister all 14 images to one reference image.

1 Like

You are right, I just figured it out