We thank the reviewer for the comments and suggestions. We have revised our text accordingly and will upload a modified version of the manuscript, including the changes listed below. 

Response to major comments

> My other concern goes to the GEANT4 simulation, particularly its low-energy aspects. A more detailed discussion of the geometry, materials, and limitations would be beneficial. 

Generally, the GEANT4 simulation accounts for low-energy aspects, and the detector geometry is based on sketches provided to us by the manufacturer, including information regarding the materials. We provide the code used for the simulation on Zenodo to avoid any ambiguity. To accommodate the request for additional details in the main text,

we will include further information.  

> Finally, I would advise to round values according to their uncertainty, there are generally accepted rules.

Thank you for pointing this out. This comment likely refers to Table 2: We had decided to use the uncertainty of the HP-Ge detector as a benchmark and therefore we show two digits in the table consistently for all methods. However, for the final results, we will remove the digit as this number is indeed not meaningful given the precision of our approach. 

> L169-170: I agree that the dead time for environmental radioactivities should not influence the results. However, the sensitivity change with the energy might be a significant factor. Could you elaborate more why it can be neglected or what precisely you refer to when referring to sensitivity?

This is indeed a valuable comment because our writing obviously caused misunderstandings. What we meant was not the simulation 

but potential photon-energy related detection sensitivities of the detector system. Our GEANT4 modelling accounts for different

photon energies, but we wanted to indicate that there might be other effects we simply ignore. However, after an internal

discussion, we decided to rephrase this part of the manuscript because it sits on speculation on our part and invites misunderstandings.

The new phrase will read:  

"We did not consider dead times because we assume that this phenomenon has a low impact on 

determining the count/energy threshold."

> L297-305: I was wondering how accurate is the low energy part of the simulation. 

> How well are mimicked detector enclosure, was the material and it thickness available? 

> Is the detector response linear in tens of keV part?

We wrote in the figure caption of Fig. 2, "Information kindly provided by the Kromek Group plc".  

We will update the figure caption to include details that we had access to that information 

(except for details of the electronic). Therefore, we believe that the simulations are fairly accurate. 

Response to minor comments

We will address all minor comments as suggested in a new version of our manuscript, below you will find

a record of the changes applied  as documentation. 

> L23: Asserting or assessing?

Corrected to assessing; thank you!

> L42: Appeal?

We replaced this by "are indicated"

> L45: I would use “statistically agree” instead of “agree within uncertainties”. Usually expanded uncertainty is not used.

Done.

> L79-82: Those sentences do not make sense when they are together.

Indeed, thank you for pointing this out. We separated the sentences better. 

> L213: “..(2019) .” please remove space

Done. 

> Figure 7. Please set y axis from 4 to 6 cps.

Done.

> L281-284: Is there instability above what would be expected from Poisson distribution? If there is no such instability detected would not use term stable unstable words. They usually refer to phenomenon that arise from the detector system itself not from counting statistic.

Thank you for pointing this out. The effect is purely stochastic and not related to any instability of the measurement system. We will rephrase the sentence. 

> L292-295: Cold you explain the reasoning behind this rule. 

We will rephrase the sentence as follows to provide a better explanation: 

"Our unpublished observations, employing the threshold method, 

suggest that the threshold shifts towards higher energies for larger detectors 

of the same material. This phenomenon is likely attributed to the increased 

proportion number of photons registered from $^{40}$K, relatively to photons 

from the U- and Th-series. To compensate for this larger contribution distribution of $^{40}$K

photons, the threshold shifts towards higher energies to and ensures that the 

total count rate is proportional to the absorbed dose. Given the 

small volume of our CZT detector, we would position the threshold in 

the low-energy portion of the spectrum, not exceeding the values 

reported in the literature."

> L314: “We wil l later show” 

Corrected, thank you.

> L326: Fig. 8E - missing dot

Corrected. 

> L329: missing dot

Done.

> L332: missing space “99keV”

The line includes a half-space; which is correct (but indeed difficult to spot).

> L336: CZY?

Corrected. 

> L344: “... (see data on Zenodo: (Kreutzer et al., 2024)).” - missing ”)”

Done. 

> Fig. 2 - 11. Could you unify as much as reasonably possible the font size?

Thank you for pointing this out. However, all figures were already produced consistently "print ready" for a two-column layout.

In the single-column layout required by Copernicus for the manuscript, the used `\textwidth` command, 

however, scales the figures differently. This will look better in the production process if the manuscript

becomes accepted. Still, depending on the final decision of the typesetter about the figure placement, 

we may have to scale the fonts again. 

> Fig. 2 - 11. Could you use it systematically: Title Case, Sentence case or lowercase.

Thank you, we corrected the writing. 

> Fig. 2. How big was the area taken for the MC simulation?

The area is 2.65 m², the volume 4.4 m³. We will add those information to the figure caption. 

> Fig. 8. On all subplots it now is difficult to assess what is on the left and right axis.

Thank you, we will add arrows to each of the four figures to indicate the corresponding scale.

> CaCO3 (subscript)

We corrected the figure (not visible yet until we can upload a new manuscript version). 

> Fig. 11. The uncertainties might be somewhat misleading. From my understanding the error bars have the same component and therefore for other measurements I am wondering about the interpretation. 

The error components are governed by the calibration and will not become significantly smaller for other measurements unless additional sites are utilized for calibration purposes.

This section is indeed systematic, as it applies to all measurements, and the random component is negligible.  

Our objective here was to demonstrate that we cover the anticipated dose rate range even for a non-ideal threshold. 

In other words, even if the threshold was not estimated with the utmost precision, the results will be accurate within the uncertainties. 

This is the inherent limitation of the method.