the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Model-based analysis of erosion-induced microplastic delivery from arable land to the stream network of a mesoscale catchment
Raphael Rehm
Peter Fiener
Abstract. Soils are generally accepted as sinks for microplastic (MP), but at the same time might be a MP source for inland waters. However, little is known regarding the potential MP delivery from soils to aquatic systems via surface runoff and erosion. This study provides for the first time an estimate of the extent of soil erosion-induced MP delivery from an arable-dominated mesoscale catchment (390 km²) to its river network within a typical arable region of Southern Germany. To do this, a soil erosion model was used and combined with potential particular MP load on arable land from different sources (sewage sludge, compost, atmospheric deposition and tyre wear) since 1950. The modelling resulted in an annual mean MP flux into the stream network of 6.33°kg° in 2020, which was dominated by tyre wear (80 %). Overall, 0.11–0.17 % of the MP applied to arable soils between 1950 and 2020 was transported into the stream network. In terms of mass, this small proportion was in the same range as the MP inputs from wastewater treatment plants within the test catchment. More MP (0.5–1 % of input between 1950 and 2020) was deposited in the grassland areas along the stream network, and this could be an additional source of MP during flood events. Most (5 % of the MP applied between 1950 and 2020) of the MP translocated by tillage and water erosion was buried under the plough layer. Thus, the main part of the MP added to arable land remained in the topsoil and is available for long-term soil erosion. This can be illustrated based on a ‘stop MP input in 2020’ scenario, indicating that MP delivery to the stream network until 2100 would only be reduced by 14 %. Overall, arable land at risk of soil erosion represents a long-term MP sink, but also a long-term MP source for inland waters.
- Preprint
(1180 KB) - Metadata XML
- BibTeX
- EndNote
Raphael Rehm and Peter Fiener
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2023-882', Anonymous Referee #1, 04 Jul 2023
Distinguished authors,
thank you for this interesting submission on modeling MP inputs from arable lands to surface waters. The topic is of high relevance for both communities, scientists as well es colleagues involved in practical water resource management.
In your manuscript you are combining a well established methodology for modeling long-term erosion and sediment transport with MP emissions from various sources, which have the potential to pollute surface waters. The regional study aims at the mesoscale Glonn catchment in Bavaria.
As most input data for MP emissions are of coarse resolution or originating from literature and due to the impossible validation, the results of your study show theoretical inputs into the soil and from the upper soil layer into surface waters. Therefore, the value of your study is less in the absolute emissions as much as of showing a potential magnitude and demonstrating systems behaviour for the pathway “water erosion”, which is underpinned by scenario analysis.
I have a couple of questions and remarks.
What is missing right from the beginning is a definition of MP: MP consists out of a wide range of different polymers with highly differing physical and chemical properties. Which polymers and which of their particle sizes are considered in your study?
Lines 34/35: “Arable soils in particular experience increased MP inputs as a result of agricultural management (Brandes, 2020).”
“increased” compared to what? Urban areas?
Lines 100-109: Information is missing on the average erosion risk. You are stating the average degree of slope and maximum erosion rates of higher than 10 t/(ha*a), but the understanding of the Glonn catchment would benefit from a map of erosion rates / USLE results. Could you please provide such map?
Line 130: The ktc parameter is crucial for determining the sediment delivery and concurrently the MP export pathway. How has it been calculated / derived? In line 159 you state a value of 150 m for ktc: Why do you think is this an appropriate setting for the Glonn catchment? Which measurements in the Glonn catchment did you use to derive the 150 m?
Line 162: “mean annual precipitation”: What is your model period? Are all input data referring to the same period? If not, why?
Line 167: The title of your manuscript deals with delivery from arable land, why do you aim at erosion modeling in forest, grassland and settlements?
Line 172: ktil is mentioned for the first time but with no background information. Where does this coefficient come from? How is it used in your model? Could you provide a relevant formula? How has it been determined? Why is the value of ktil transferable to the Glonn catchment?
Lines 205 ff: Averaging the sewage sludge amounts from the reports over all Bavarian fields including those in the Glonn catchment bears large uncertainties and is practically not a valid method.
The “Klärschlammverordnung” (AbfKlärV) is a very restrictive instrument to manage the transport and distribution of sewage sludge in Germany. According to §6 (1) AbfKlärV from 1992, which is relevant for the time frame you are looking at, arable land can receive up to 5 tonnes per hectare sewage sludge in dried form (“TM”). In practice, only a few parcels receive sewage sludge and most of them don’t take sludge. When the sludge is being applied, the parcels receive the full load. After three years a minority receive sludge again, but mainly other parcels are being used then. All this is overstamped by the results from prescribed soil analyses to allow sludge application for on these target parcels.
Therefore, the distribution of sewage sludge is spatially highly variable and your averaging approach does not reflect the real situation. When you combine this sludge (and MP) average with your erosion rates and sediment deliveries, which are also highly variable in space, then the outcome on mixing in the soil, delivery to streams etc. is very theoretical. A validation of the model results could have revealed this, but it is not possible due to the lack of long-term MP measurements in surface waters and from point and other sources. So, there is no evidence for the validity of your model results.
Table 2 should be shifted to the end of section 2.3.5 as it summarizes the inputs from all sources considered.
Lines 270-274: The MP input from atmospheric deposition is coupled to the general plastic production in Germany. I don’t understand this approach as it bears another source of huge uncertainty.
MP consists out of a wide range of different polymers with highly differing physical and chemical properties. I miss a justification, why the range of MP in your deposition measurements should be equal to the range of plastic polymers being produced in Germany. How are you dealing with this uncertainty?
Line 288: “No emissions from unpaved roads and agricultural machinery were considered.”
Why not? I think, this would be of importance as it represents a direct input to arable lands.
Lines 348-351: Please explain in greater detail, why you think that “Given the ongoing increase in plastics production…this may even be a conservative estimate of a business-as-usual scenario pathway.”
Most of your MP input comes from tyre wear of adjacent roads. Tyre production is not the same as plastic production and the tyre wear is dependent on traffic density, population numbers and the situation of buffer strips and distances.
Line 374: You state correctly, that the estimated MP inputs contribute significantly to model uncertainty.
How are you dealing with these uncertainties in the model approach and application? Shouldn't the model structure be adapted to these uncertainties?
What are the consequences for the reliability of study results and the usability for practical water resources management?
Figure 6: Inserting the road network more clearly would help to understand the spatial distribution of polluted areas. Maybe this is a visibility problem with the resolution of the graphics.
Please explain, why you cannot see in the map the higher MP load in those parcels close to roads, which receive the high input from the tyre wear in comparison to parcels farer away.
Line 493: Sediment transport modeling is of course a difficult topic. But you should rethink, if you evaluate a R² of 51 % as “perfectly”? In line 366 you are talking about “satisfactorily”.
What is true now: perfection or satisfaction?
A formal point: Years in the references are lacking a closing bracket quite often throughout the manuscript. Please correct.
I would suggest to implement two additional aspects, maybe in the discussion:
- Arable land is contributing to MP pollution in surface waters, BUT the major inputs into arable soils are coming from non-agricultural sources. What can be done to reduce inputs from tyre wear?
- Plastic pollution in all environmental compartments is a major challenge. By far more measurements and basic research in this field are required to foster process understanding and to improve model applications significantly.
Citation: https://doi.org/10.5194/egusphere-2023-882-RC1 -
AC1: 'Reply on RC1', Raphael Rehm, 13 Aug 2023
Dear Reviewer,
We thank you for your kind and constructive feedback on our manuscript, which we greatly appreciate.
Thank you very much for pointing out that we missed stating right from the beginning which kind of microplastic we address. We had some information later in the model description, but we will undoubtedly give this earlier in the text. In principle, the model does not distinguish between different plastic sizes, types, shapes, etc. but transports everything added to the soil. Therefore, we should have included aspects such as microplastic enrichment during erosion and transport, as we could not address more specific microplastic properties based on the rough estimate of MP inputs to soils.
You mentioned needing average background erosion risk data and an erosion map. This information will be provided in a revised manuscript.
Reply to comments regarding model parameterization and model input data:
- The transport capacity coefficient in SPEROS-MP is taken from Van Oost et al. (2003) . Based on data from the Belgium Loess Belt, it was calibrated for a 5 m x 5 m grid resolution.
- We only varied the annual rainfall erosivity as we did not have suitable other input data back to 1950. This is one reason why the model underestimates the annual sediment delivery dynamics. However, as we intended to use the model to show the potential magnitude of MP delivery and also use the model to study the system behavior, we did not focus more on the input variables (since, by far, the most significant uncertainty in simulated MP delivery results from the estimate of soil contamination based on large-scale input data estimates).
- The tillage erosion coefficient is used to estimate tillage erosion rates. The coefficient represents a literature mean for conventional tillage typically applied in the catchment. We will give more details in the revised version of the paper.
- We agree with the reviewer that using an average sewage sludge input derived from the Bavarian average does not represent the complex actual situation. However, we do not have parcel-specific information since the 1950s (assuming all associated MP inputs would be more or less stable over time). However, as stated above and in the paper, the intention of the modeling exercise was not to exactly reproduce the MP delivery in the Glonn catchment but to use the model in combination with available input data to perform system analysis. We will try to make this more evident in der revised version of the manuscript.
- Other estimates of MP inputs: We will try to be more explicit about how we made our MP input estimates but will also discuss the associated uncertainties in more detail (e.g., our estimate of atmospheric MP deposition since the 1950s using current data and long-term plastic production is very uncertain. However, based on our modeling analysis, it should be clear that the atmospheric input is of minor importance anyway, which we will discuss in our updated manuscript).
Reply to comments regarding modeling results and their discussion:
- We will provide more detail regarding our scenarios and their implications.
- We will add more discussion regarding the uncertainty of our input data and focus on ignored input pathways, e.g., tire wear from unpaved roads.
- In general, the idea was that the uncertainty of the model results is mainly related to the estimates of the soil MP contamination resulting from different sources. Therefore, we always used potential contamination from other sources as model inputs. This should show the main uncertainties, which are much larger than any uncertainties resulting from the erosion modeling-related input data, model parameters, and model structure, which is also underlined by the satisfactory validation of the modeled sediment delivery.
- We will update the maps, including the main roads, to make identifying highly polluted fields along the main roads easier.
- We will discuss in more detail how MP pollution could be reduced and where scientific improvements are essential to improve model applications.
Best regards
Raphael & Peter
Van Oost, K., Govers, G., & Van Muysen, W. (2003). A process‐based conversion model for caesium‐137 derived erosion rates on agricultural land: An integrated spatial approach. Earth Surface Processes and Landforms: The Journal of the British Geomorphological Research Group, 28(2), 187-207.
Citation: https://doi.org/10.5194/egusphere-2023-882-AC1
-
RC2: 'Comment on egusphere-2023-882', Anonymous Referee #2, 26 Sep 2023
This study addresses an interesting question by evaluating the contribution of erosion processes to microplastic delivery to the river network. I applaud the authors to collate a substantial amount of information on different sources of plastic inputs into soils. Overall, the paper is well written, the results are clearly described and an in-depth analysis of the results is provided. However, I do feel that the methods are not fully appropriate: the use of a complex spatially- and temporally explicit model in a context where no model-constraining data is available is questionable. All plastic pathways are proportional to sediment fluxes and can therefore be inferred directly from sediment pathways. There are very large uncertainties associated with the estimation of the plastic input (info on spatial and temporal patterns is absent) and several key processes are not constrained or represented in the model (eg size selectivity of detachment, transport and deposition, plastic weathering, ….). Together, this makes its application to a specific case-study in a low-data context challenging.
I understand that a scoping study can rely on several assumptions and simplifications. However, I suggest that the authors evaluate to what extent their complex approach is justifiable under the absence of sufficient quality input and validation data. An alternative, and likely more robust approach is to use a simple (spatially lumped) mass-balance or accounting model that allows to cover both a range of plastic input- and a range of water, tillage, sediment delivery- scenarios. This could provide a more comprehensive assessment of plastic pathways, residence times and delivery. I am confident that the authors can re-frame this study and maybe add a more robust approach.
Citation: https://doi.org/10.5194/egusphere-2023-882-RC2
Raphael Rehm and Peter Fiener
Raphael Rehm and Peter Fiener
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 246 | 73 | 14 | 333 | 5 | 6 |
- HTML: 246
- PDF: 73
- XML: 14
- Total: 333
- BibTeX: 5
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1