the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Sep 2026
Technical note: Monitoring diel soil water variations below trees using high-resolution electrical resistivity tomography
Abstract. Urban trees provide many important ecosystem services like cooling or CO2-sequestration. For planting, maintenance and long-term health it is essential to understand their water system and supply. While soil moisture sensors and soil sampling provide only point-scale information and are often invasive, geophysical methods provide spatially continuous information, capture subsurface structures non-invasively and thereby improve process understanding of water dynamics below trees. In this study, electrical resistivity tomography (ERT) was used to monitor subsurface moisture dynamics at a tree site in an urban park over a period of one month continuously and at a high temporal resolution. The ERT measurements were complemented by soil water content (SWC) and sap flow measurements. We focus our analysis on precipitation-free periods to capture small SWC variations caused by soil water redistribution. During these periods, both the SWC data and the ERT data showed a clear diel pattern, with zones of decreasing SWC during the day and increasing SWC at night. At the same time, sap flow measurements show increased tree activity during the daytime drying phase. We explain these observations with root water uptake during the day and a recovery of subsurface moisture during the night following the redistribution of water in the subsurface. In addition, we found indications for an additional conductivity signal that is linked to transpiration and not captured by the soil moisture sensors. Our results show that high-resolution ERT, in combination with SWC and sap flow measurements, is a useful approach to investigate tree–soil water interactions in urban environments given that artefacts and temperature effects are carefully handled.
- Preprint
(2961 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 23 Oct 2026)
- CC1: 'Comment on egusphere-2026-4818', Sivarajah Mylevaganam, 16 Sep 2026 reply
-
CC2: 'Comment on egusphere-2026-4818', Sivarajah Mylevaganam, 16 Sep 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-4818/egusphere-2026-4818-CC2-supplement.pdf
-
CC3: 'Comment on egusphere-2026-4818', Sivarajah Mylevaganam, 16 Sep 2026
reply
4. The authors demonstrate a strong correlation between the measured SWC and the ERT-derived SWC at depths of 1.0 m and 2.0 m using Pearson’s r. However, the reported values call into question whether Pearson’s r is the most appropriate statistical measure for this data. Specifically, subplot d shows a severe deviation between the measured SWC and ERT-derived SWC after June 13, yet it still yields a deceptively high Pearson’s r of 0.74. In contrast, subplot c visually displays a much closer fit and better representation of the trend, but yields a lower Pearson’s r of 0.649. Because Pearson’s r evaluates linear trends rather than absolute differences, it may be masking systematic offsets or phase shifts in subplot d. The authors should consider reporting some other statistical metrics that capture absolute agreement.
5. In Figure 4, the data in subplots c and d are based on the SWC derived from ERT measurements and SWC recorded by the sensors. Although the exact derivation steps are not explicitly shown, it is reasonable to expect the same downward trend over time across both methods. Discrepancies here would otherwise imply instrumental or methodological errors, given that the measurements originate from the same location. However, a critical point requires clarification: the ERT values represent an integrated depth interval of 0.75–1.25 m, whereas the point-sensor SWC is measured strictly at 1.0 m. Should we expect these values to align perfectly, or is a higher degree of variance expected from the ERT data due to its volume-averaging nature over the 0.75–1.25 m interval compared to the discrete 1.0 m SWC sensor?
6. In section 3.1, the authors note that soil moisture was monitored at depths of 0.15 m, 0.3 m, 0.5 m, 1 m, and 2 m. However, the subsequent data analysis, figures, and comparisons are strictly limited to the 1.0 m and 2.0 m, leaving the shallower details completely unaddressed or masked. The authors need to provide a clear justification for excluding these data points. Incorporating the shallow-depth profiles is highly critical, as it would provide a complete picture and help clarify the anomalous lag-time behaviors observed deeper in the soil profile.
7. As indicated in Lines 175–178, the instrumentation deployed varies by depth: an FDR Wet150 sensor was used at 1.0 m, while a TDT SMT100 sensor was used at 2.0 m. Mixing entirely different measurement technologies (FDR vs. TDT) introduces distinct variations in the operations. Therefore, the authors must address whether this discrepancy in sensor instrumentation influenced the monitored data outcomes, particularly the contrasting trends, sharp variations, and anomalous behaviors reported between the 1.0 m and 2.0 m depths.
8. Comparing subplot d with the precipitation data in Figure 3 (plot a) raises a critical question: what physical mechanism caused the soil moisture content at 2.0 m to suddenly spike to approximately 3.2 around 9:00 PM on June 13? This sudden increase contradicts expected drainage and drying behaviors. In fact, the moisture levels abruptly rebounded to match the peak baseline seen on June 11. When cross-referenced with sap flow measurements, this spike seems to point toward an unaccounted-for hydrological impact or artifact.
Notably, this phenomenon occurs exclusively at the 2.0 m depth and is completely absent at the 1.0 m depth. Is it possible that this delayed spike was triggered by the precipitation event on June 10? That specific rain event appears to have been cleanly captured by the 1.0 m sensor as a sharp spike on June 11, where moisture levels climbed well above their original baseline. Could preferential flow paths, perched water tables, or a delayed groundwater response explain why this wetting front bypassed or cleared the 1.0 m sensor but heavily saturated the 2.0 m sensor three days later?
Citation: https://doi.org/10.5194/egusphere-2026-4818-CC3 -
CC4: 'Comment on egusphere-2026-4818', Sivarajah Mylevaganam, 16 Sep 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-4818/egusphere-2026-4818-CC4-supplement.pdf
-
RC1: 'Comment on egusphere-2026-4818', Andre Revil, 16 Sep 2026
reply
This is a well-written manuscript on an exciting topic. I have only major comments, which are provided below without specific order of importance. I am sure the authors will correct them easily with some work and the paper could be published afterward. At this stage however, the paper contains a lot of mistakes and errors that should be avoided in the published paper.
1. The use of the word artifact at the end of the abstract is unclear.
2. Line, this is not really the clay content that controls conductivity and complex conductivity in general but rather clay mineralogy and clay content because different clay minerals have very different CEC, cation Exchange Capacity.
3. I wonder why ERT alone has been done while induced polarization is a much more advanced technique able to separate the surface conductivity from the bulk conductivity component. I think using IP would have been a much better option.
4. The factor a is NOT a tortuosity factor in equation (1). This error has plagued the literature and we mistake appears again and again. The bulk tortuosity is the product of the formation factor (corrected for surface conductivity) and the (connected) porosity. Please remove this error. There is no a-factor in Archie’s law and read Archie’s paper.
5. It has been shown again and again that surface conductivity exists even in abstence of clay minerals, see a paper published in Geophysics on the Fontainebleau sandstone for instance. It just a matter of salinity.
6. Just below equation 2, what are these additional conductivities? I am puzzled. Do you mean Stern and diffuse layers contributions?
7. Equation was developed much before Haley et al. so please as a rule of basic ethics cite the oldest references on the subject…
8. You forgot to mention that surface conductivity depends on the water content itself. I don’t understand why this has been lost in translation while this is a well-known and established fact.
9. This sentence “From these quantities and the geometry of the electrode layout, the apparent resistivity of the subsurface can be calculated using Ohm’s law.” Is of course wrong. The geometrical coefficient requires (in general) the use of both a constitutive equation (ohm’s law) and a continuity equation for the charge. This is just basic physics teach in any 101 of geophysics.
10. Provide perhaps a motivation regarding the use of the time lapse ERT approach with respect to the various methods published in the literature. See fr instance the papers of Karaoulis.
11. I don’t believe this sentence is right: “Reciprocal measurements provide a widely used and reliable way to estimate 200 data errors”. Reciprocity measure reciprocity, not error level. Reciprocity has to do with the causal and linear aspect of the physics and the properties of the Green function in this case. If you inject too much current, Ohm’s law becomes non linear and reciprocity is not valid. This has nothing to do with error level. This is another myth of the recent literature associated with a strong decrease of the level of understanding of the physics. If you want to measure error level in the data, just use a repeat of the measurements and compute their standard deviation or probability density on the measurements. But fine, I am asking for any changes here but the authors should think a bit by themselves on this point.
12. Line 219, it is affected by the CEC, clay content, and texture, this is why ERT cannot be used as a stand-alone technique…
13. About the temperature correction,, what is the depth of the domain in which there is a temperature change affected by the daily T variations. This is strange that this is mentioned nowhere…
14. Line 300, I am puzzled by the choice of using n = 2 without any lab data indicating such a value? May we have a clue.
15. Lines 300-301. I was surprise that in the first part of the manuscript, the underlying physics of surface conductivity is never mentioned or expressed. Then, reading Lines 300-303, I understand why ! Surface conductivity depends on porosity and saturation BUT the authors decided to ignore this dependence by choosing to go for a surface conductivity that is saturation independent. There are tens perhaps hundreds of recent publications demonstrating that this assumption is totally wrong and of course ANY of them is cited in this study. Either the authors do not know the literature or either they hide some truths: their assumption are not valid. If the authors would have maude IP, this problem would have been solve. But like they ignore the physics of surface conductivity, they also decided to ignore induced polarization and,makes like of ERT exist and not IP. As a letter of fact IP can b done with the same equipment than ERT and in the same timeframe. So again, I am bit puzzled here. I think a far better discussion of the state-of-knowledge we have would be MORE than welcome. Making assumptions is not an issue. Trying to hide the truth below the carpet is a perhaps very serious issue. More ethic is required.
16. It is kind of sand that the soil in the area has not been sampled to conduct proper lab characterization (porosity, CEC measurements) etc. and to assess the n and m textural parameters. I think this is a major issue in this work. I would humbly remark that Figure 4 contains errors associated with the inversion approach itself and not accounted for by the reciprocal measurements. These errors are far bigger than all the discussion related to error in the measurements made in the paper. Here again, I am kind of surprise that there is no word on that. This can be easily fixed.
A.Revil
Citation: https://doi.org/10.5194/egusphere-2026-4818-RC1
Data sets
Dataset - Technical note: Monitoring diel soil water variations below trees using high-resolution electrical resistivity tomography Hoppenbrock et al. https://doi.org/10.5281/zenodo.21875392
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 227 | 93 | 38 | 358 | 23 | 22 |
- HTML: 227
- PDF: 93
- XML: 38
- Total: 358
- BibTeX: 23
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The global demand for development has driven rapid urbanization across the world. However, human infrastructure remains inherently bound to the natural climate system. To mitigate and pacify the climate alterations caused by rapid urban expansion, various counter-measures have been introduced. Among these, urban trees play a critical role, acting as both biological filters and natural fans that provide microclimate cooling and carbon dioxide sequestration. Despite their benefits, urban trees are frequently planted without a sufficient understanding of the underlying subsurface system, specifically within the complex unsaturated zone. To address this critical knowledge gap, the authors utilize Electrical Resistivity Tomography (ERT) measurements alongside sap flow and soil water content (SWC) monitoring. This experimental setup allows them to evaluate subsurface dynamics during non precipitation periods and contribute valuable data to the existing literature. Based on the information presented in the manuscript, the following points are brought forward for discussion:
1. In Figure 4, subplots (a) and (b) are based on Archie's law (Equations 1 and 2). The current scaling of the data creates the impression that the plots were tailored specifically to demonstrate that the observed data follows a y = mx^2 + c relationship. While the x-axis spans widely from 0 to 16, the actual data points are tightly clustered within a very narrow range: between 8 and 10 in subplot (a), and between 2 and 3 in subplot (b). Consequently, compressing the data into such restricted windows visually forces an artificial alignment with the trendline, masking the true distribution and variance of the data.
2. In Equations 1 and 2, which are based on Archie’s law, bulk electrical conductivity is determined by scaling the pore fluid conductivity by a specific formation factor. This factor is typically expressed as a function of parameters such as soil water content (SWC) and porosity. Modern theoretical advancements have expanded this framework by incorporating surface conductivity as an additive term to the factored pore fluid conductivity. While Section 4.2 attempts to explain the findings presented in Figure 4, critical methodological details have been omitted from the text, thereby obscuring the authors' exact workflow and assumptions. To ensure transparency and reproducibility, the following points regarding the figures require immediate clarification:
2.1 In Figure 4, in subplot (a), the exponent m is set to 2, whereas subplot (b) uses a value of 1.81. Reviewing the equations, if m = 2 (as in subplot a) and the variable n is also set to 2, the exponent term (m – n) becomes 2 - 2 = 0. Consequently, the term porosity^{m-n} (or equivalent porosity) in equation 2 vanishes to 1, meaning this parameter exerts no influence on the fluid conductivity scaling.
Conversely, in subplot (b), m is set to 1.81. This implies that porosity does influence the bulk conductivity. However, because n remains fixed at 2, the exponent becomes negative (1.81 - 2 = -0.19), resulting in an inverse relationship for that factor. The authors need to justify the physical mechanism behind this inverse relationship and explain why different m values were applied across the subplots. It is understood that these subplots represent the experimental data collected at depths of 1 m and 2 m, respectively.
2.2 Clarification regarding the surface conductivity values must be included in the manuscript or technical note. Subplot (a) shows that the surface conductivity—acting as the additive term in equation 2—is 0.27 units. In contrast, this value increases nearly fivefold to 1.17 units in subplot (b).
Because subplot (a) corresponds to a depth of 1 m and subplot (b) corresponds to a depth of 2 m, the authors need to provide a clear physical or geochemical justification for this substantial fivefold jump in surface conductivity with depth. Factors such as changes in clay content, mineralogy, or soil compaction across these two horizons should be explicitly discussed to validate these parameters.
Furthermore, this justification must align logically with the porosity values assumed by the authors, which decrease from 0.50 at a depth of 1 m to 0.44 at a depth of 2 m. The authors must reconcile these physical parameters with the mathematical coefficients applied to the porosity term—specifically, an exponent of 0 in subplot (a) versus -0.19 in subplot (b). A cohesive explanation is required to show how these combined changes in surface conductivity, porosity, and structural exponents physically represent the soil conditions at their respective depths.
2.3 An evaluation of the values presented in subplots (a) and (b) reveals a remarkable shift in magnitude between the two depths. In subplot (a) (1 m depth), the bulk conductivity is bounded between 0.30 and 0.35, while in subplot (b) (2 m depth), it ranges between 1.5 and 2.0. More notably, the pore fluid conductivity jumps drastically from 7.23 in subplot (a) to 276 in subplot (b).
This represents a massive, multi-orders-of-magnitude increase in fluid conductivity over a mere 1-meter depth interval. Based on the information provided by the authors, the groundwater table is assumed to be situated much deeper, at 4.0 to 5.5 m below the ground surface. Given that both monitored intervals reside well within the unsaturated zone, such an extreme spike in fluid salinity or conductivity is highly anomalous. The authors must provide a thorough, decoded breakdown and physical justification for these presented values.
3. In Equation 3, the authors demonstrate that standard methodologies are available in the literature to normalize soil electrical conductivity (EC) to a reference temperature of 25°C. However, according to the variable descriptions, this temperature correction is explicitly applied only to the measured bulk EC.
This raises a critical methodological question: does this imply that Equations 1 and 2 cannot be fused with Equation 3 to extend Archie’s law framework and define bulk conductivity directly at the reference temperature of 25°C? From a reader’s perspective, the current presentation makes these equations appear completely disconnected from one another.
Moreover, it is unclear whether Equation 3 was actually utilized in the analysis or if it is merely highlighted as background theory. The authors need to explicitly state how Equation 3 integrates with the rest of their workflow.