As a data archivist, I want to use day^-1 as a permissible value for unit

[murashinln]

Checked for duplicates

No - I haven't checked

🧑‍🔬 User Persona(s)

No response

💪 Motivation

...so that I can describe a sensitivity degradation rate for detector of an imager in the label.

📖 Additional Details

I'm preparing the PDS4 bundle for the Longwave Infrared Camera instrument (LIR) onboard the Venus Climate Orbiter Akatsuki spacecraft. I defined vco:lir_sensitivity_degradation_rate attribute for LIR calibrated data in VCO LDD, and value of the attribute has a unit of day^-1. It seems to me that Units_of_Rates is an appropriate class to define it, but it's ok defining it in other class if you think it is more suitable.

Acceptance Criteria

Given When I perform Then I expect

⚙️ Engineering Details

No response

[cgobat]

Seems like Units_of_Frequency might make sense for this, since that's where all the other inverse time units currently live. A stopgap measure could even be to multiply your value(s) by ~0.01157 and use the 'mHz' (i.e. $10^{-3}\ \text{s}^{-1}$) unit that already exists.

[matthewtiscareno]

The difference between Units_of_Rates and Units_of_Frequency appears to be a subtle one. A rate should specify what it is that is happening per unit time (the only current examples are counts/bin and kilobits/s). A frequency is understood to be some kind of cycle per unit time.

Both of these have units of one over time. For that matter Units_of_Angular_Velocity also has units of one over time.

@murashinln, can you be more specific about how exactly you are measuring sensitivity degradation? Is there some specific quantity whose rate you are reporting?

[murashinln]

Sensitivity degradation is estimated from measured radiance for deep space pixel which is decreasing in time. The sensitivity degradation is modeled as a linear function of time (Taguchi+2023), as described below:

A sensitivity degradation rate $\alpha$ is used in an equation for sensitivity degradation correction, $I_o(T, t) = (1+\alpha t')\{I(T) - I(T_s)\} + I(T_s)$, where $I_o(T, t)$ is an observed radiance for a target temperature $T$ at an observation time $t$, $t'$ is an operating time of the instrument at the observation time $t$, $I(T)$ is a corrected radiance for a target temperature $T$, and $I(T_s)$ is a radiance for a shutter temperature $T_s$.

[matthewtiscareno]

@murashinln: Okay, so it seems like α is the rate at which a dimensionless quantity is decreasing linearly with time. Correct?

In that case, perhaps Units_of_Rates makes the most sense, as you suggested in the OP. @cgobat, would you agree?

[murashinln]

@matthewtiscareno Correct. To be clear, we are thinking on radiance, but a sensitivity degradation is represented as a part of scaling factor $1+\alpha t'$. If there is an image with 20% sensitivity degradation, then $\alpha t' = -0.2$ and $1+\alpha t' = 0.8$.

[jordanpadams]

@murashinln @murashinln @matthewtiscareno do we have agreement that this will require an update to the PDS4 Standard? If so, we can get a ticket created with the DDWG to review this request.

[murashinln]

@jordanpadams I think we have agreement. @cgobat reacted with thumbs up emoji to https://github.com/NASA-PDS/pds4-information-model/issues/660#issuecomment-1681732496 .

[jordanpadams]

@rsjoyner would you mind creating an SCR for this with the DDWG?