maja issueshttps://gitlab.orfeo-toolbox.org/maja/maja/-/issues2023-05-22T09:19:24Zhttps://gitlab.orfeo-toolbox.org/maja/maja/-/issues/346False cloud detection on Venus images2023-05-22T09:19:24ZCarole AmiotFalse cloud detection on Venus imagesOn specific Venus sites (landscapes), cloud detection presents more false detection than on other sites.
This behavior is looked into. It could be linked to the spectral indicators used for cloud detection, which might need to be adapt...On specific Venus sites (landscapes), cloud detection presents more false detection than on other sites.
This behavior is looked into. It could be linked to the spectral indicators used for cloud detection, which might need to be adapted for desert landscape.
**Urgency/Criticity**
- Urgency (Select Low/Medium/High): Low
- Criticity (Select Low/Medium/High): Lowhttps://gitlab.orfeo-toolbox.org/maja/maja/-/issues/318Poor cloud detection over India (43PFP)2022-06-10T07:11:01ZJerome ColinPoor cloud detection over India (43PFP)---
name: General Issue (FA)
about: Use this template for reporting an anomaly.
---
<em>Please make sure that this is a bug. As per our [GitHub Policy](https://gitlab.cnes.fr/maja/maja/blob/develop/ISSUE.md), we only address code/doc b...---
name: General Issue (FA)
about: Use this template for reporting an anomaly.
---
<em>Please make sure that this is a bug. As per our [GitHub Policy](https://gitlab.cnes.fr/maja/maja/blob/develop/ISSUE.md), we only address code/doc bugs, performance issues and build/installation issues on GitHub.</em>
**System information**
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04): HAL
- Maja installed from (source or binary): binary
- Maja version (use command below): 4.5.4
- Maja commit sha (if Maja is installed from source):
- Python version (From binpkg or custom):
- Complementary info (i.e. GCC/Compiler version)
**Describe the current behavior**
While processing a time series over India (43PFP) from november 2021 to may 2022, I found several products with unexpectedly poor cloud mask quality, especially within the period of March-April 2022 (mostly clear with sparse clouds).
![Capture_d_écran_de_2022-06-02_16-52-29](/uploads/a8d88dda5275c24d3a97c25f346545e5/Capture_d_écran_de_2022-06-02_16-52-29.png)
Example for April 6th (SENTINEL2B_20220406-053529-254_L2A_T43PFP_C_V1-0)
![L2A_T43PFP_20220312_vs_20220406](/uploads/00dbf2e4fba9171e089e5c7fc2f410e6/L2A_T43PFP_20220312_vs_20220406.gif)
I'm wondering if the temporal criteria applies here, especially considering that another clear product was available less than a month before on March 12th.
If you look at the full quicklook, you'll notice that the mask is better on the eastern side of the product, and flags both clouds and cloud shadows. There are indeed much more products on the neighboring orbit. But still, the product from March should be enough to detect the clouds of April.
![SENTINEL2B_20220406-053529-254_L2A_T43PFP_C_V1-0_FRE_B2](/uploads/d79d4dc7fdbf153be47931a87220485e/SENTINEL2B_20220406-053529-254_L2A_T43PFP_C_V1-0_FRE_B2.jpeg)
I couldn't find an explanation in making diffs on GIPPs. `Cloud_Forgetting_Duration` is set to 45 which is correct. Could there be an issue here (or is the maximum threshold on surface reflectance variation with time to be tuned again) ?
**Describe the expected behavior**
Nice cloud/cloud shadow mask at least on the biggest cumulus.
**Urgency/Criticity**
- Urgency (Select Low/Medium/High): tempting to sigh High
- Criticity (Select Low/Medium/High): High
**Code to reproduce the issue**
Provide a reproducible test case that is the bare minimum necessary to generate the problem.
Include any logs or source code that would be helpful to diagnose the problem. If including tracebacks, please include the full traceback. Large logs and files should be attached.
A path to a context folder is also accepted. Please note the [Context policy](https://gitlab.cnes.fr/maja/maja/blob/develop/ISSUE.md#Contexts)
**Correction**
(Optional) Provide a version number (Major.Minor.Patch, e.g. 3.2.1) when the correction shall be introduced.Jerome ColinJerome Colin