MODIS
-
giannismac
- Posts: 4
- Joined: Thu Jun 06, 2019 1:22 pm America/New_York
MODIS
A full answer to your question needs to consider how the level-3 mapped images are computed. Mapping always introduces distortions (which is why many different projections are available) and there are multiple ways to go from level-2 data to mapped images.
Level-2 products represent data from a single overpass of the sensor. Each level-2 pixel corresponds to a single sensor measurement, which collects light from an area on the surface where the contribution from the "pixel footprint" (an oval area on the ground) is greatest at the center and declines in a 2-D bell-shaped response away from the center. The footprint can be much larger near pass edges due to the earth's curvature. The quality of "edge" pixels is generally lower due to increased atmospheric path length, which introduces complications due to the vertical distribution of gases, clouds, and water vapor in the atmosphere. Sensors have a nominal resolution which is based on the distance between pixel centers directly under the sensor.
Level 3 mapped (L3M) products ideally combines data over space and time from multiple overpasses, but in practice missing level 2 data (clouds, etc.) means the may be no data or contributions from a small number of level-2 pixels. NASA L3M products are generated from binned (L3B) data. You can also create mapped (L2M) images by mapping each level-2 pass to a common projection and time-averaging the resulting images. With L2M data the large ground footprint of (lower quality) edge pixels means they can contribute multiple mapped values, and thus outweigh centerline pixels in time-averaged mapped images. Binning gives each level-2 pixel equal weight, so L3M files generated from L3B files represent higher quality data (and provide more meaningful estimates of variances) than mapped files generated by averaging L2M data.
If you have access to the OCSSW processing tools, comparisons between L3M files from NASA and from averaging L2M files are a useful exercise, and should include maps of the variances or standard deviations (using the l3mapgen program).
Level-2 products represent data from a single overpass of the sensor. Each level-2 pixel corresponds to a single sensor measurement, which collects light from an area on the surface where the contribution from the "pixel footprint" (an oval area on the ground) is greatest at the center and declines in a 2-D bell-shaped response away from the center. The footprint can be much larger near pass edges due to the earth's curvature. The quality of "edge" pixels is generally lower due to increased atmospheric path length, which introduces complications due to the vertical distribution of gases, clouds, and water vapor in the atmosphere. Sensors have a nominal resolution which is based on the distance between pixel centers directly under the sensor.
Level 3 mapped (L3M) products ideally combines data over space and time from multiple overpasses, but in practice missing level 2 data (clouds, etc.) means the may be no data or contributions from a small number of level-2 pixels. NASA L3M products are generated from binned (L3B) data. You can also create mapped (L2M) images by mapping each level-2 pass to a common projection and time-averaging the resulting images. With L2M data the large ground footprint of (lower quality) edge pixels means they can contribute multiple mapped values, and thus outweigh centerline pixels in time-averaged mapped images. Binning gives each level-2 pixel equal weight, so L3M files generated from L3B files represent higher quality data (and provide more meaningful estimates of variances) than mapped files generated by averaging L2M data.
If you have access to the OCSSW processing tools, comparisons between L3M files from NASA and from averaging L2M files are a useful exercise, and should include maps of the variances or standard deviations (using the l3mapgen program).
-
giannismac
- Posts: 4
- Joined: Thu Jun 06, 2019 1:22 pm America/New_York
-
OB.DAACx - SeanBailey
- Posts: 1519
- Joined: Wed Sep 18, 2019 6:15 pm America/New_York
- Been thanked: 9 times