Earthdata Forum

Dear NASA CERES Satellite team,

I’m writing to you for further guidance at the suggested email address provided in the NASA Quality Summary document - ERBE-like Monthly Geographic Averages (ES4) – Edition 4. It would be much appreciated if I could get some clarification from someone that is intimately familiar with the hardware measurements beyond what I could find on the specification sheet found here: https://ceres.larc.nasa.gov/instruments/

I’ve been doing personal research into climate and the Earth energy budget for a few years. In late 2024, I downloaded a few datasets from Earthdata. In particular, the most useful ones for my studies have been three all skies datasets of gsolar_mon, gtoa_sw_all_mon, and gtoa_lw_all_mon. I’m extremely impressed with the rich sets of data available for people doing research – it is very impressive what the instrumentation and data processing people at NASA have done.

Here are the questions I’m hoping someone can help me with:

Q1 – Sample size for radiometric accuracy spec - When I look at the specifications page, for radiometric accuracy, it says 1% (SW) k=1; 0.5% (LW) k=1; 0.5% (TOT) k=1. I’m assuming that is for a single measurement and k=1 is one standard deviation. Am I understanding that correctly?
Q2 – Resolution of hardware for one measurement - When I look at the data from the downloads, each month has a very precise number with four significant digits after the decimal – xxx.xxxx. This suggests to me CERES hardware has a lot of resolution in taking a measurement. What is the resolution of a single measurement in terms of significant digits after the decimal point?
Q3 – Number of measurements that are averaged - In a 31 day month, approximately how many measurements are taken for the average
Q4 – Dataset maintenance when drift detected - If NASA realizes drift every few years, I presume the datasets would be readjusted to the detected drift and my whole dataset has been re-aligned as best as possible. Is that correct?
Q5 – Goal of 2-sigma uncertainty on monthly data - Is there anything else that you can tell me as what I’m trying to do is put error bars on my plots for the monthly readings for a 2-sigma uncertainty? Perhaps you have them but either they aren’t on the documents I’ve looked through or I’ve just missed them. So my goal is to be able to put an uncertainty on your monthly values: xxx.xxxx +/- y.yyy to a 95% certainty.

I’d really appreciate it if you can help me if that isn’t too much trouble. I’m expecting I’ll need to wait until early January when people return from their holidays – not an issue for me as likely my questions need a specialist with that knowledge.his question was posted on behalf of the user.

Hello,

Q1 – Sample size for radiometric accuracy spec - When I look at the specifications page, for radiometric accuracy, it says 1% (SW) k=1; 0.5% (LW) k=1; 0.5% (TOT) k=1. I’m assuming that is for a single measurement and k=1 is one standard deviation. Am I understanding that correctly?

That is correct. K=1 is one standard deviation.

Q2 – Resolution of hardware for one measurement - When I look at the data from the downloads, each month has a very precise number with four significant digits after the decimal – xxx.xxxx. This suggests to me CERES hardware has a lot of resolution in taking a measurement. What is the resolution of a single measurement in terms of significant digits after the decimal point?

The footprint size of ERBE scanning radiometers is about 30 km by 45 km at nadir.
The footprint size of CERES instruments is about 20 km at nadir.

Q3 – Number of measurements that are averaged - In a 31 day month, approximately how many measurements are taken for the average

I believe there are 100,000 CERES instrument footprints per hour, hence about 2.5 million footprints per day globally in the Level 2 SSF product. But all footprint measurements are not independent. If you take monthly mean irradiances and compute autocorrelation with one month time lag, the coefficient is no 0, especially for longwave. For regional irradiances, however, we had assumed daily mean irradiances were independent before.

Q4 – Dataset maintenance when drift detected - If NASA realizes drift every few years, I presume the datasets would be readjusted to the detected drift and my whole dataset has been re-aligned as best as possible. Is that correct?

That is correct. Overpass time of spacecrafts that onboard ERBE and CERES instruments are maintained.


Q5 – Goal of 2-sigma uncertainty on monthly data - Is there anything else that you can tell me as what I’m trying to do is put error bars on my plots for the monthly readings for a 2-sigma uncertainty? Perhaps you have them but either they aren’t on the documents I’ve looked through or I’ve just missed them. So my goal is to be able to put an uncertainty on your monthly values: xxx.xxxx +/- y.yyy to a 95% certainty.

For ERBE and CERES data, a random error in monthly mean values is negligible because random noise from instantaneous radiance measurements is very small after the averaging. But biases remain, especially biases caused by instrument calibration. This is a problem because we do not know their signs. Bias errors in CERES global mean irradiances are summarized in Table 2 of Loeb et al. (2009). Instrument calibration uncertainties (k=2 in Wm-2)) appear under filtered and unfiltered radiances in the Table. ERBE instrument calibration is not as good as CERES instrument calibration. Similar uncertainties are listed in Table 6 of Kopia (1986). The values are k=3, the Table shows in radiance. Because the radiance to irradiance conversion is approximately 3, you can use these values as k =3 in Wm-2. When you compare the Table 2 and Table 6, you find that CERES instrument accuracy is about twice as good as ERBE’s. These are all uncertainties in absolute irradiances. If you are plotting, for example, anomaly time series, the uncertainty in changes in irradiances is much smaller. It is hard to quantify the uncertainty in irradiance changes. Depending on what you are doing, but Section 3.2 of Loeb et al. (2024) might be useful.

References
Kopia, L.P, 1986: Earth Radiation Budget Experiment scanner instrument, Review of Geophysics, 24, 400-406.

Loeb, N.G., and coauthors, 2009: Toward optimal closure of the Earth’s top-of-atmosphere radiation budget, J. Climate, 22, 748-766.

Loeb, N.G., and coauthors, 2024: Continuity in top-of-atmosphere earth radiation budget observations, J. Climate, 37, 6093-6108.