Dear,

I agree with your replies to the comments of the reviewers.
However, I have two main additional comments. The first is on the statistical evaluation of agreement between the measured and EMI derived ECe’s. This was to my opinion not clearly described and I propose including the equations that were used to calculate the RMSEs, R² and other indices. The reason for including these is that there is an ambiguity about what you use as predicted value when you calculate these indices. The ECe that you obtained from the EMI estimated sigma and a regression equation, or the prediction of the measurement from a regression between the EMI derived ECe and the measured ECe?
The second main (and more critical) comment is that you do not address the real issue of this paper, namely, can you use EMI to MONITOR ECe, i.e. to evaluate whether it changes over time. You are only discussing that you observe changes over time. But, are the changes you observe in the soil sample measurements consistent with the changes that you observe in the EMI measurements? You do not address this issue and I think this is crucial. A negative answer to this question will not lead to a rejection of your paper. But, you need to answer this question using appropriate statistical analyses. If this question is not answered appropriately, the work that you present is not meeting the objectives that are suggested in the title, the introduction and the conclusion and can therefore not be accepted.

Detailed comments:

Ln 15: You could maybe include here why the soil faces a risk of salinization in this area.

To avoid confusion, I propose to use the same units for ECa, sigma, and ECe.

Ln 23: ‘This study aims to evaluate the potential of time-lapse EMCI and the regional calibration to predict the spatiotemporal variability of soil salinity in the study area.’ After having read the article, I think that this statement is not sufficiently supported by the results that were presented and discussed. You indeed evaluated whether time lapse EMI measurements can explain the total variation in a dataset of measured ECe values at different locations and depths and at different times. But, the question whether EMI can be used to monitor changes in ECe over time was not addressed and to my opinion it should be. You should evaluate whether the changes in ECe that are observed over time correspond with changes in EMI estimated ECe over time. When the answer to this question is negative, this paper shows that the spatial variations in ECe in the studied area are much larger than the temporal variations so that EMI can still be used to map spatial variations but not the temporal variability due to irrigation and seasonal leaching. Evaluating the temporal variability of ECe within a year might require to have also information about variation of soil water content, groundwater table depth and soil temperature, which also vary over time and modulate the variations of ECe over time.

Ln 27: ‘revealed salinity fluctuations related to the input of salts and water either through irrigation, precipitation or groundwater level and salinity.’ You best reformulate this sentence. ‘Salinity fluctuation related to salinity’ is a circular explanation.

Ln 100: How was reference ET obtained?

Ln 119: Can you give some details about the soil samples? Soil rings or bulk samples? Which volume, mass of soil?

Ln 139: ‘The RMSE is the square root of the mean of the squared differences between the measured and predicted ECe, indicating how concentrated the data is around the linear regression’ I am confused here. Which linear regression line are you referring to here? The linear regression between sigma and ECe that is based on the regional dataset and that you use to obtain ECe estimates from EMI measurements, which I would call ECep? Or the linear regression between ECep and the measured ECe from the soil samples? In case of the latter, you are ‘correcting’ the regional regression equation. At ln 140, you write: ‘Lin's CCC measures the agreement between the measured and predicted ECe evaluating how close the linear regression is to the 1:1 relationship’ This makes me wonder which regression equation you are talking about here. The regression between ECep and ECe measured? If that is the case, then the RMSE that you mention in ln 139 is not the mean of the squared differences between the measured and predicted ECe but it is the mean of the squared differences between the measured and the ECe’s which are predicted using two sequential regressions: first you predict ECe form sigma to obtain ECep and then you make a linear regression between ECep and ECe measured and calculate the RMSE of the deviations between this regression and the measured ECe’s. In order to avoid all these confusions, you should write the equations that you used to calculate the RMSE and the R² and you need to define clearly what are the ‘predicted’ ECe’s: the ones obtained using the linear regression based on the regional calibration between sigma and ECe or the predictions by the regression between these predicted ECe’s and the measured ECe’s?

Ln 154: for clarity, give the depths of ‘lower subsoil’, ‘intermediate subsoil’.

Ln 165: Figure 3: use the same horizontal axis for the different figures.

Ln 222: ‘At location 4 the trend of increasing salinity with depth is accurate in all dates, but it tends to be slightly underestimated. The salinity maps show that salinity increases from non-saline in topsoil to severely-saline in lower subsoil.’ How can I see in figure 5 that the trend is underestimated? I do not see that.

see added file

Dear authors,

Based on the reports of the reviewers and the TE, Im happy to inform you that we can accept your paper for publication in SOIL subject to technical corrections. However, I agree with the final suggestions made by the TE in relation to the quantification of the performance (r2) and the impact of location time. If you are able to implement these changes, we can publish your work.
We are looking forward receiving your final version and thanks again for submitting your work to SOIL.

Kristof Van Oost