HELP_Clarification on MODIS Cloud Effective Radius Uncertainty

[aakashmodis]

Hello, I am Aakash, a researcher working on aerosol–cloud interactions. I have been extensively using MODIS aerosol and cloud products for my research.

I would like to seek clarification regarding the cloud variable "Cloud_Effective_Radius_Uncertainty" in the MYD06_L2 product. In the documentation, it is described as: "Cloud Effective Particle Radius (from band 7) Relative Uncertainty (Percent) from both best points and points identified as cloud edge at 1 km resolution or partly cloudy at 250 m."

Based on this, I would like to confirm whether this represents the relative measurement uncertainty (%) of the cloud effective radius retrieved from the 2.1 µm channel. For example, if Cloud_Effective_Radius = 16 µm and Cloud_Effective_Radius_Uncertainty = 25, does this imply an uncertainty of
16 ± (0.25 × 16) = 16 ± 4 µm?

I am also confused when comparing this with the Level-3 product MYD08_D3. The variable "Cloud_Effective_Radius_Liquid_Mean_Uncertainty", which is derived from the aggregation of the L2 uncertainty, is reported with units of microns (µm), suggesting an absolute uncertainty.

I would be grateful if you could kindly clarify the following:

1. Do these uncertainty variables represent measurement uncertainty of the retrieved cloud effective radius?

2. Is it correct to interpret that the L2 uncertainty is relative (%), while the L3 uncertainty is absolute (µm) after aggregation?

3. Are the uncertainties in CER retrieved from different channels (e.g., 2.1 µm and 3.7 µm) for the same pixel or grid box independent, or are they correlated due to shared retrieval assumptions and radiometric inputs?

I would greatly appreciate your guidance on this matter.


Best regards,
Aakash Kumar
IIST, India

[LAADS_UserServices_M]

Dear Aakash Kumar:
See our response in blue:

I would like to seek clarification regarding the cloud variable "Cloud_Effective_Radius_Uncertainty" in the MYD06_L2 product. In the documentation, it is described as: "Cloud Effective Particle Radius (from band 7) Relative Uncertainty (Percent) from both best points and points identified as cloud edge at 1 km resolution or partly cloudy at 250 m."

Based on this, I would like to confirm whether this represents the relative measurement uncertainty (%) of the cloud effective radius retrieved from the 2.1 µm channel. For example, if Cloud_Effective_Radius = 16 µm and Cloud_Effective_Radius_Uncertainty = 25, does this imply an uncertainty of
16 ± (0.25 × 16) = 16 ± 4 µm?
Yes. "Cloud_Effective_Radius_Uncertainty" is a relative uncertainty (%) associated the retrieved CER from band 7 (2.1 µm channel). It is pixel-level relative (%).
If Cloud_Effective_Radius = 16 µm and Cloud_Effective_Radius_Uncertainty = 25 (%), then the absolute uncertainty = 16 µm × 0.25 = 4 µm.
You can express the results as: 16 ± 4 µm.

I am also confused when comparing this with the Level-3 product MYD08_D3. The variable "Cloud_Effective_Radius_Liquid_Mean_Uncertainty", which is derived from the aggregation of the L2 uncertainty, is reported with units of microns (µm), suggesting an absolute uncertainty.
In MYD08_D3, "Cloud_Effective_Radius_Liquid_Mean_Uncertainty" is in µm (absolute unit). It is L3 grid-cell average.

I would be grateful if you could kindly clarify the following:

1. Do these uncertainty variables represent measurement uncertainty of the retrieved cloud effective radius?
These uncertainty variables are the retrieval-based estimate of how sensitive the CER is to the input and model uncertainties, not just the instrument measurement noise.

2. Is it correct to interpret that the L2 uncertainty is relative (%), while the L3 uncertainty is absolute (µm) after aggregation?
Yes, it is correct.

3. Are the uncertainties in CER retrieved from different channels (e.g., 2.1 µm and 3.7 µm) for the same pixel or grid box independent, or are they correlated due to shared retrieval assumptions and radiometric inputs?
No, they are not fully independent. They are partially correlated. They share the same instrument calibration errors, same cloud mask and phase classification, same LUTs, and same assumptions (cloud vertical homogeneity, particle size distribution, surface reflectance).
They are different in the absorption strengths, sensitivity to different cloud layers, and radiative effects.

[aakashmodis]

Dear LAADS User Services Team,

Thank you for your response and for the clarification. I sincerely appreciate your assistance and the information provided.

Best regards,
Aakash
IIST, Thiruvananthapuram