the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 May 2024
Soil contamination in arid environments and assessment of remediation applying surface evaporation capacitor model; a case study from the Judean Desert, Israel
Abstract. Many of the globe arid areas are exposed to severe soil contamination events, due to the presence of highly pollutant industries in these regions. In this work a case study from the Ashalim basin, at the Judean desert, Israel was used to examine the nature of solutes and contaminants transport in sandy terraces of an ephemeral stream that was exposed to a severe pollution event.
In order to to shed new light on contaminants distribution along the soil profile and transport mechanisms, in arid environments, three complimentary approaches were used: (1) Periodic on-site soil profile sampling, recording the annual solute transport dynamics; (2) Laboratory analyses and controlled experiments in a rain simulator, to characterize solutes release and transport; and (3) Numerical simulation was used to define and understand the main associated processes.
The study highlights the stubborn nature of the pollutants in these natural setting that dictates they will remain near the soil surface, despite the presence of sporadic rain events. It was shown that a vertical circulation of the contaminates is occurring with soil wetting and drying cycles. The ‘surface evaporation capacitor’ concept of Or and Lehmann from 2019 was examined and compared to field measurements and numerical simulations, and found to be a useful tool to predict the fate of the contaminants along the soil profile.
- Preprint
(2031 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (extended)
-
RC1: 'Comment on egusphere-2024-1014', Anonymous Referee #1, 03 Jun 2024
reply
The article sheds light on contaminant transport in arid region using a combination of lab and field experiments and numerical simulations. The incident releasing contaminants in the Judean desert was a flash flood after the breaking of a dike in the year 2017. After this event, salts and contaminants are redistributed in the soil profile during rainfall infiltration and evaporation. The authors apply the surface evaporation capacitor concept to test if contaminants could percolate to deeper soil layers and are removed from the active evaporation layer.
The topic, the case study, and the applied methods are interesting, but the analyses must be extended and presented in more detail as explained below.
- surface evaporation capacitor: the percolation from the capacitor to deeper soil layers depends on the water content of the capacitor. The water content in the capacitor must be higher than the critical water content (as calculated in Assouline and Or, 2014, WRR WR015475, or Lehmann et al, 2019, GRL GL083932). The authors should expand the SEC-analysis by estimating the water content after the winter rainfall events to check if percolation to deeper soil layers can occur (for the calculated thickness of the capacitor, what is the water content after a certain rainfall event?).
- Hydrus-1D simulations: After the simulation of the time period Sept_20 to August_21 presented in figure 9, the solute is concentrated in soil layers close to the surface and is not redistributed to larger depths shown in the experimental findings (figure 7). This discrepancy can be partially related to the flash flood that cannot easily be simulated. Another effect that should be taken into account in the simulation is the repeated redistribution between the incident 2017 until 2021. How is the solute plume travelling with depth for this 4-5 years period?
- Soil water retention: The soil water retention curve was measured with the hanging column method resulting in shape parameters of alpha equal to 0.011/cm and n = 2.8 (Table 3). With such a small alpha the drainage occurs between 50 and 200 cm. Was this pressure range covered with the hanging water column method? The authors should show both the measured values and the fitted curve.
- Saturated hydraulic conductivity: In contrast to the soil water retention curve, the saturated conductivity was not measured but estimated with Rosetta implemented in Hydrus-1D. The predicted value for a rather dense packing of ~1.7 g/cm3, is about 25 cm per day; the other predicted parameters (probably about n=1.41 and alpha = 0.0268/cm) are quite different compared to the lab experiments (n = 2.8 and alpha = 0.011/cm). The combination of parameters obtained with different approaches in the SEC-model may lead to inconsistent values of the thickness of the capacitor layer. In addition, the predicted hydraulic conductivity is rather small compared to the irrigation rate applied in the rain simulator experiments (48 mm per hour or 115.2 cm/day). I would expect that a saturated hydraulic conductivity much smaller than the irrigation rate would result in more runoff than found in the experiments presented in Figure 5.
- Figure 7, there is an increase of chloride in the East Plot. What are the hypotheses for that increase? Could you add in figures D, F, H and J the concentrations in deeper layers as well (line for 30-40 or 50-60 cm according to line 315). Please provide more information on the rainfall rates and amounts and on the profile measurements (show in the figure when the samples were collected). Do you expect identical hydraulic properties in East and West plot? Could this be tested?
Citation: https://doi.org/10.5194/egusphere-2024-1014-RC1 -
AC1: 'Reply on RC1', Uri Nachshon, 03 Sep 2024
reply
The article sheds light on contaminant transport in arid region using a combination of lab and field experiments and numerical simulations. The incident releasing contaminants in the Judean desert was a flash flood after the breaking of a dike in the year 2017. After this event, salts and contaminants are redistributed in the soil profile during rainfall infiltration and evaporation. The authors apply the surface evaporation capacitor concept to test if contaminants could percolate to deeper soil layers and are removed from the active evaporation layer.
The topic, the case study, and the applied methods are interesting, but the analyses must be extended and presented in more detail as explained below.
We thank the reviewer for the positive and constructive comments. We addressed each comment raised by the reviewer, please see replies below (in bold).
- surface evaporation capacitor: the percolation from the capacitor to deeper soil layers depends on the water content of the capacitor. The water content in the capacitor must be higher than the critical water content (as calculated in Assouline and Or, 2014, WRR WR015475, or Lehmann et al, 2019, GRL GL083932). The authors should expand the SEC-analysis by estimating the water content after the winter rainfall events to check if percolation to deeper soil layers can occur (for the calculated thickness of the capacitor, what is the water content after a certain rainfall event?).
This is a good point. Indeed, percolation below the capacitor will initiate once water content at the SEC goes beyond a critical water content value - θcrit. As detailed in the works mentioned by the reviewer and others, θcrit is soil water content retained at the soil for suction pressure Lc , that describes the soil characteristic evaporation length (hcrit=Lc). As stated by Lehmann et al (at various papers) water at depths greater than Lc will not be exposed to evaporation during first stage of evaporation and it may drain downward.
Once again - if water content at the SEC will go over θcrit – water leakage from the SEC will start.
For calculated Lc in our work (as appears in Figures 8 & 9, by the red contour), and based on the VG model and the hydraulic parameters as appear in Table 3, average θcrit is equal to 0.1776 with maximal values during mid-summer (=0.197) and minimal values during mid-winter (=0.1561). This means that for water contents higher than ~0.15, during winter, it is expected that water will drain from the SEC downward.
Mass soil water content measurements were done during winter after each rain event (1-2 days following the rain event). Measurements were done by sampling the moist soil from various depths, weighing of the wet soil, oven drying of the soil samples at 105oC (for 24hr), and weighing of the dry soil. Averaged measured mass water content was at the order of 0.02, with maximal values of 0.06 (rarely measured). For soil bulk density of ~1.67 (g/cm3), that is computed for the sandy soil at the site, upon its measured porosity, the equivalent volumetric water content is at the order of 0.033 which is much smaller than θcrit , hence it is likely to assume that leakage from the SEC is negligible at the test site, under the examined conditions.
This was discussed in the revised paper, where at the introduction we explained the relation between water content and leakage from the SEC (P5, L121-122). At the M&M section we elaborated about the procedure of water content measurements (P8, L187-194), and at the results and discussion section, we discussed it further, considering measured water content values (P15, L344-350).
- Hydrus-1D simulations: After the simulation of the time period Sept_20 to August_21 presented in figure 9, the solute is concentrated in soil layers close to the surface and is not redistributed to larger depths shown in the experimental findings (figure 7). This discrepancy can be partially related to the flash flood that cannot easily be simulated. Another effect that should be taken into account in the simulation is the repeated redistribution between the incident 2017 until 2021. How is the solute plume travelling with depth for this 4-5 years period?
For simplicity, and in order to clearly visualize and characterize the simulated solute transport processes and to correlate it to the SEC concept, the initial conditions of the model were set with contamination to be located only at the top10 cm of the soil profile.
The effect of flash flood is not being taken into account as it is assumed (and observed at the site) that the sandy terraces are above the mainstream flow channel and are not being flooded during normal and natural flash flood. The only (natural) sources of water that the terraces are being exposed to are the sporadic and minor rain events that may occur during winter. These rain events are included in the model. The flood of the catastrophic contamination event was much higher than the terraces, therefore they were highly polluted.
The initial conditions of the model are quite similar to the measured conditions of Cl concentration at Sep. 2020, as at this time Cl concentration was concentrated at the top 10 cm of the East plot and ~20 cm of the west plot. For Ca however, the solutes concentration distribution was wider to depths of ~30cm. This disparity is associated to the much higher quantities of calcium at contaminated effluent, and in particular, the fact that Ca may go through processes of absorption to the soil particles and it is less mobile than Cl. This is already mentioned in the text, yet we elaborated on that at the revised paper at the field measurement section (P15-16, L354-357) and the numerical model section (P17, L370-371).
- Soil water retention: The soil water retention curve was measured with the hanging column method resulting in shape parameters of alpha equal to 0.011/cm and n = 2.8 (Table 3). With such a small alpha the drainage occurs between 50 and 200 cm. Was this pressure range covered with the hanging water column method? The authors should show both the measured values and the fitted curve.
The hanging column test went down to suction of 160 cm. Soil drainage occurred in between the suctions of 20 to 60 cm. At suction of ~150cm water content of the sample was already equal to the residual water content and no more water went out of the sample. The WRC data is added to the data presented at the Repository.
- Saturated hydraulic conductivity: In contrast to the soil water retention curve, the saturated conductivity was not measured but estimated with Rosetta implemented in Hydrus-1D. The predicted value for a rather dense packing of ~1.7 g/cm3, is about 25 cm per day; the other predicted parameters (probably about n=1.41 and alpha = 0.0268/cm) are quite different compared to the lab experiments (n = 2.8 and alpha = 0.011/cm). The combination of parameters obtained with different approaches in the SEC-model may lead to inconsistent values of the thickness of the capacitor layer. In addition, the predicted hydraulic conductivity is rather small compared to the irrigation rate applied in the rain simulator experiments (48 mm per hour or 115.2 cm/day). I would expect that a saturated hydraulic conductivity much smaller than the irrigation rate would result in more runoff than found in the experiments presented in Figure 5.
This is a very good point. Thanks to this comment we did a Darcy test to determine saturated hydraulic conductivity of the soil. Indeed, as speculated by the reviewer, Ks is much higher and it is at the order of 250 cm/d. We repeated all simulations with this value. Moreover, we did more repetitions for the hanging column test and we refined the V.G. parameters (not a big difference from previous values). Now all physical properties of the soil were determined based on physical measurements and the numerical simulations were changed accordingly. Fortunately, these changes did not change the processes, mechanisms and trends, which were discussed and examined in the paper. In the M&M section we mentioned that Ks was measured by a Darcy test (P6, L140-141).
- Figure 7, there is an increase of chloride in the East Plot. What are the hypotheses for that increase? Could you add in figures D, F, H and J the concentrations in deeper layers as well (line for 30-40 or 50-60 cm according to line 315). Please provide more information on the rainfall rates and amounts and on the profile measurements (show in the figure when the samples were collected). Do you expect identical hydraulic properties in East and West plot? Could this be tested?
Unfortunately, it is impossible to present deeper information. We presented all available data. In part of the samplings, we managed to reach deeper levels than others. Complete information about rainfall and evaporation is presented in panels A and B. As mentioned by the figure caption, the triangles at the upper X axis of panels A and B present times of soil sampling. Nothing is identical in nature, but it is believed / assumed that the sands in both terraces are alike. Texture measurements, which were done for samples from both sites support this assumption.
As for the first comment regarding increase in Cl concentration: this could be attributed to upward flow of Cl from levels deeper than ~60 cm, which were not sampled in Sep. 2020. However, this is not very likely as these depths are greater than the lower boundary of the ESC. We believe that the net change in solutes mass was negligible and that the reason for changes in concentrations is due to accumulation of the solutes in narrower layer in 2021. It is seen that in Sep. 2020 the “salt bulb” went down to depth of ~20 cm, whereas at Sep. 2021 it was concentrated in a narrower layer with a thickness of ~10 cm. In practice, it means that in 2021 the concentration is ~2 times higher than measured concentrations in 2020. This is mentioned at the revised manuscript (P15, L337-339).
Citation: https://doi.org/10.5194/egusphere-2024-1014-AC1 -
AC2: 'Reply on AC1 - FILES', Uri Nachshon, 04 Sep 2024
reply
Attached is the reply letter and the revised MS. All changes are marked in red.
Data sets
FIGS_PKUS_DATA_RESULTS_SECTION_GOLAN_ET_AL_2024 Rotem Golan et al. https://doi.org/10.6084/m9.figshare.25534285
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 181 | 65 | 70 | 316 | 29 | 23 |
- HTML: 181
- PDF: 65
- XML: 70
- Total: 316
- BibTeX: 29
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1