i.landsat.acca: add test cases #5270
Conversation
github-actions
bot
added
Python
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
| self.runModule( | ||
| "r.mapcalc", | ||
| expression=f"{self.output_cloud} = if({self.output_cloud} > 0, 1, 0)", | ||
| overwrite=True, | ||
| ) |
There was a problem hiding this comment.
What is the purpose of this?
There was a problem hiding this comment.
This is done to follow the standard Landsat approach where 1 means clouds and 0 means clear areas. It is used to turn the cloud layer into a simple binary map. If a pixel has a value greater than 0 (meaning it’s likely a cloud), it’s set to 1. If it’s 0 (no cloud), it stays 0. It also ensures that later calculations, like those using assertRasterFitsUnivar, match the expected values—such as a maximum of 1 and a sum of 2401. Without this step, the sum could be like 10,000
| 0, | ||
| "Cloud detection should identify at least some cloud pixels", | ||
| ) | ||
| if np.sum(cloud_array) > 0 and np.sum(cloud_array) < np.size(cloud_array): |
There was a problem hiding this comment.
The condition if np.sum(cloud_array) > 0 and np.sum(cloud_array) < np.size(cloud_array): is used to confirm that cloud detection is neither completely empty nor marking everything as a cloud. If the sum was 0, it would indicate that no clouds were detected. If the sum equals the total number of pixels, it would mean the entire area is classified as a cloud
| cloud_array = array.array(cloud_output) | ||
| self.assertGreater( | ||
| np.sum(cloud_array), | ||
| 0, | ||
| "Cloud detection should identify at least some cloud pixels", | ||
| ) |
There was a problem hiding this comment.
This is pretty much testing the same thing as in the first test
There was a problem hiding this comment.
This one specifically focuses on verifying the correlation between cloud pixels and thermal band values. The goal is to confirm that detected clouds align with expected thermal characteristics, beyond just checking if cloud pixels are detected.
This PR introduces a test suite for the i.landsat.acca GRASS GIS module, covering a range of scenarios, including cloud mask creation, cloud shadow detection, and the effect of different input parameters on cloud identification. The test cases ensure correct functionality across scenarios like thermal correlation, binary output validation, and the impact of the b45ratio threshold.
Key updates in this PR include:
Functionality: Validate cloud mask creation using i.landsat.acca and ensure proper cloud detection.
Edge Cases: Test cloud shadow detection and the impact of the b45ratio parameter on cloud detection.
Cloud Mask Validations: Ensure the cloud mask is binary (with only two unique values) and accurately represents cloud and non-cloud pixels, while validating that cloud pixels have higher thermal values than non-cloud pixels.
Test Case Additions:
Basic Functionality:
Edge Cases:
Cloud Shadow Detection: Test cloud shadow detection by ensure that shadow pixels are
detected, and the output is valid.
b45ratio Parameter Effect: Test the impact of the b45ratio parameter on cloud detection by adjusting it. Ensure that a higher b45ratio threshold results in fewer detected cloud pixels, indicating a more conservative approach to cloud identification.
Cloud and Thermal Correlation Validation: