Assessment of OLCI and SLSTR misregistration
- Tonio Fincke
- Marco Peters
- Former user (Deleted)
Goals
- Facilitate to assess the spatial misregistration between OLCI and SLSTR bands. The misregistration can be assessed in two ways: In the first way, SLSTR L1B (FR) products are compared to SLSTR L1B products. In the second way, bands from a SYN L1C product are compared.
Background and strategic fit
By this manner, the Synergy processing can be validated.
Assumptions
Either OLCI L1B FR and SLSTR L1B or SYN L1C product available.
Requirements
| # | Title | User Story | Importance | Status | Notes |
|---|---|---|---|---|---|
| 1 | Assessment using reference points | The user opens an OLCI L1B and a SLSTR L1B product. He/she continues to load in a list of reference points with well-known geolocations. These points are placed on the image grids. The user can then assess how well the points were placed. The primary format for the geolocations are plain text, CSV files: <lat>, <lon>, <label>. Useful are also Google Earth KMZ and ESRI Shapefiles. | MUST | Implemented. This is basically the import of placemarks. |
|
| 2 | Assessment by comparing geolocations | The user places a pin in an OLCI L1B product at some distinct feature (e.g., the tip of an island). The user places another pin at the same feature in a SLSTR L1B product. Both pins will have geolocations assigned to them. These geolocations can be compared. | COULD | Not implemented. We are unsure about how and whether to integrate the uncertainty. The plan is for someone to look at the displacement bands created from olci and slstr pixelgeocodings and to determine whether the displacement is so large it should be considered. | This would also require form of uncertainty information: What is the error which is introduced by the application of the inverse geocoding? |
| 3 | Assessment by transferring pins | The user places a pin in an OLCI L1B product. A pin is automatically set in a SLSTR product on base of the pin's geolocation. The displacement of the pin and its expected position can be assessed visually. In addition, the geocoding from the expected position in the SLSTR product can be derived and compared to the pin's geoposition. | MUST | Not implemented, but there is a new feature: When pins are transferred from Product A to Product B a second time (or multiple times), the user is asked whether to update the pins. If the answer is yes, the corresponding pins from an earlier transfer are removed and replaced by new pins. This didn't work once without me being able to find out why. | This would make use of a global or synchronized pin. |
| 4 | Assessment in a SYN L1C product | First, any "OLC_RADIANCE" band. The user places pins in one of the camera images or imports pins from an external file. From a "MISREGIST_OLC" file, the SceneRasterTransform (forward and inverse) is created. "OLC_RADIANCE_O17" is the reference band and has the SceneRasterTransform.IDENTITY. Using "MISREGIST_SLSTR", SceneRasterTransforms (only inverse!) are created and set to SLSTR bands. A pin set in an OLCI image can thereby be placed at the corresponding position in a SLSTR image.The position of the pixels can be compared visually or by comparing the geo-locations. | MUST | Implemented. Users can set pins or draw shapes in any OLCI band. These figures are depicted on all OLCI or SLSTR bands. It is not possible to draw on SLSTR bands, as there is no forward transform. When a figure is set outside of the area where the misregistration has valid values, the figure is distorted. |
User interaction and design
For Requirement 2:
Define how to visualize displacement uncertainty. Using the displacement bands could be a start.
Also, consider the following formula where GC is a geocoding, G is a geolocation, and P is a position in image coordinates:
G = GC_olci (P_olci)
P_slstr = GC^-1_slstr (G)
G' = GC_slstr (P_slstr)
P'_olci = GC^-1_olci (G')
delta_G = G - G'
delta_P_olci = P_olci - P_slstr
delta_G and delta_P_olci would then be displacement measures.
Another attempt to be used on SLSTR or OLCI grid:
- Determine four corners and center of a pixel
- Determine the resulting five geolocations
- From these, determine relative displacement
For Requirement 3:
Use a global/synchronized pin. This pin would be placed in one product, and then, based on either its geoposition or its image coordinates, be placed in all (or some) other products which are currently opened in the Sentinel-3 Toolbox. There could already be uncertainty information attached to the pin ("uncertainty pin"), maybe in the form of an ellipse. See "For Requirement 2" for ideas on how to arrive at this information.
Also, there is the open issue how such a synchronized pin shall be saved.
For Requirement 4:
Requirement 4 consists of two phases: Transferring a pin to another grid ("OLC_RADIANCE_O17" -> "MISREGIST_SLST_NAD_Oref_S3") and then, based on the information at the "MISREGIST_SLST_NAD_Oref_S3", placing it in "SLST_NAD_RADIANCE_S3". These steps could easily be automated.
When the case for other olci or slstr bands is considered, the procedure is different in that the "MISREGIST_OLC" bands are required.
Questions
Below is a list of questions to be addressed as a result of this requirements document:
| Question | Outcome |
|---|---|