General Comments

                  Voigt and coauthors investigate the water balance and chemistry of a small lake in an arid environment with dynamic, seasonal hydrology. This is an interesting and timely sibject because small arid lakes are prone to changes in the face of anthropogenic climate change and are rarely in hydrologic steady state. Voigt and coauthors characterize the hydrologic conditions in the lake with the following modeling and empirical efforts: 1) collection of water samples for isotopic (d18O, d17O, and d2H) and anion/cation analysis, 2) simulations to match the water isotopes, and 2) isotope mass balance considering lake levels, and 4) model water in the lake via satellite imagery, bathymetry, and an estimate of input water from soil moisture. The lake water balance is controlled by evaporation, precipitation, and basin discharge. There is pronounced seasonality in the water isotopes, including in triple oxygen isotopes.  Evaporation and relative humidity are two main controls on the isotopic composition of the lake water.

                  This paper is an impressive combination of empirical data (isotope and environmental monitoring) and modeling approaches. Combined, these results yield a comprehensive description of the dynamic hydrological processes in a small, arid lake.  Their use of triple oxygen isotopes to identify non-steady state hydrological processes is exciting and demonstrates the value of this novel technique in applications to modern hydrology. Furthermore, I think that while it is not surprising that water isotopes vary in a small catchment throughout a year, it continues to be important to point this out to the paleoclimate community. I recommend publication of this manuscript with very minor revisions. Below I offer some comments and suggestions for improvement.

Specific Comments

Line 135: How are you estimating measurement precision? Is this the S.D. of the seven injections for a single vial, or is it the S.D. of multiple replicates of the standards run over time? I would recommend using the latter, and also incorporating an estimate of error in your normalization scheme, to arrive at an accurate estimate of error. See Hutchings and Konecky (2023).

Line 330: the spatial variation in lake water isotopes is a very interesting finding. I would like to see this emphasized for paleoclimate applications. Paleoclimate workers often sample lacustrine sediments at a single location (one outcrop, one core). This result implies that, for small lakes in the geologic record, we should be sampling horizontal transects.

Line 365: this finding points to a strong need to measure triple oxygen isotopes in water vapor. You may consider highlighting this result in the conclusions.

Line 385: Do you have any thoughts on why the C-G evaporation model is unable to match the isotope data given the measured parameters?

Line 430 - 455: this discussion text mostly answers my above question. This data highlights that diurnal fluctuations are important to semi-arid lake hydrology, perhaps more so than seasonal or annual conditions. This finding could be useful for understanding which/how anthropogenic climate changes will impact arid environments, and which parameters should be considered when examining model predictions.

Line 444: Is there a citation showing water vapor build up above lakes during periods of high evaporation that could support this idea? This set of sentences is confusing - why does more wind correspond to lower turbulence? This is the opposite of what I would expect.  

Line 468 - Could the "end of the dry season" gray dots also be explained by evaporation from a different source? The points do not quite match up with the concave up prediction from the model - they form more of a cluster, not a trend. While the concave-up/looping prediction matches with previous data (Voigt et al., 2023), the data in this paper do not strongly support that prediction. I would suggest modifying the text to describe this slight disagreement, and possible offer an explanation for the offset.

Furthermore, why do you think that the January 2022 samples do not 'complete' the cycle, and end up agreeing with the Jan 2021 samples? Are the conditions (or antecedent conditions) different comparing the two Januaries?

Figure 5 caption: there is only a central panel, rephrase

Figure 6- It might be useful to add a gradient of color within your time blocks to demonstrate that these isotopic values are evolving towards the steady state. This might be a challenge, though, with the rainbow in the background.

Table S3: Please report d17O and d18O to the third decimal point as this information is needed to calculate D17O in per meg.

Technical corrections

Line 26 (and elsewhere): Minor grammar error.  The triple oxygen isotope system allows to identify non-steady state conditions --> The triple oxygen isotope system allows us to identify non-steady state conditions. Alternate grammatically correct structure: The triple oxygen isotope system allows the identification of non-steady state conditions.

Line 60: Anthropogenic climate change (delete "the")

Line 360: remove comma after Both

General Comments: Voigt et al. present a manuscript detailing the monitoring and modeling of a lake in semiarid southern Spain. Using a multi-systems approach, they model lake levels as a function of parameters that are commonly measured in modern systems as well as characterize potential proxy variables for the reconstruction of paleo-hydrological conditions. I find their manuscript carefully written and well-presented. I have a series of minor questions and inquiries as line comments below. In addition, I have one main desire which would be the inclusion of more detail on how their salts-based and isotope-based modelling might influence paleohydrologic studies. This linkage is teased in the front of the manuscript but perhaps could be expanded on somewhat in the discussion.

Figure 1: Is the extrapolation of lake SA and volume consistent with previous lake levels? Or simply an extension based on the established topography? In panel (a), what do the numerical values inset on each lake contour indicate? I assume total lake volume at that water level?

Lines 155-161: E_SA is estimated from potential evapotranspiration and lake surface area? Or is there additional treatment of evaporation-driven water losses from the lake? Aren’t there considerations needed based on wind fields, surface roughness, air-water temperature differences, etc. when estimating something like E_SA? This is not my expertise, so perhaps a lake of this size and particular setting make these less of an issue, but my general understanding is that E_SA is typically non-trivial.

Lines 173-184: This section also does not mention in any depth the relevance of the E_SA term. Obviously, from proper evaporation, we do not have any meaningful losses of salinity. However, depending on the seasonal strength of the wind field then we might expect some losses due to lake spray aerosols. Again, perhaps this is a non-issue due to the size of this lake and the environmental conditions, but some text on the E_SA term here could help more fully explain your methodology even if the processes I describe are completely negligible.

Line 212-217: Can you expand some on how these sensitivity experiments were handled? As Monte Carlo-type simulations? Were distributions for variables normal, uniform, etc.? Were all variables allowed to freely vary within their uncertainty bounds or was each sensitivity experiment the variation of a single parameter?

Line 221-223: What is the error window of the bathymetric model? Perhaps the uncertainty of the lake bathymetry is small enough that any volume error from bathymetry (as opposed to lake level) is negligible, but given all the other considerations, perhaps this is worth including.

Line 235-239: Another possible consideration is the difference between air temperature and lake temperature. Although this is a small lake and probably reasonably tracks monthly air temperature, diurnal variations in air temperature may result in larger-than-expected evaporation occurring during nighttime when air temperatures drop but the lake remains warm.

Line 247-255: How well-constrained is n, really? A margin of 0.1 (as a SD of a normal distribution?) may be adequate, but I wonder if a larger range should have been considered.

Line 270 / Figure 2: I wonder how many discrete rain events you have based on weather station data? Would it be useful (and legible) to include indicators for each rain event observed during the period of observation? You could perhaps limit indicators to rain events above some threshold value. This might be helpful in seeing the step-wise linkage between precipitation and lake level. I see the stepwise increases in cumulative precipitation at the top of Fig. 2. I wonder if a bar chart of each rain event (or binned 1 or 2-week cumulative values) would serve the visual explanation of this data better than a running accumulation timeseries? Just thoughts!

Line 385-392 / Figure 5: This is great work and shows the strength of the sensitivity of isotope data to the important tunable parameters. I wonder exactly how the authors arrived at this particular solution and whether or not an automated solution approach (perhaps via Markov-Chain Monte Carlo) could identify whether or not multiple possible solutions exist. Further, I wonder if the variability in each parameter is well constrained by the bounds selected here. Certainly, we can imagine n to vary more than between 0.4 and 0.6, but also if we consider things like diurnal variability it may be possible to imagine that a 5% +/- bound on relative humidity is too conservative. Obviously, you must strike a middle point between unbounded variables and exact values – I think the authors have done a good job here generally but should, perhaps, include some additional reasoning on their stated bounds.