the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Mar 2022
AWESOME: Archive for Water Erosion and Sediment Outflow MEasurements
Abstract. Soil erosion is a major threat to soil resources, causing environmental degradation and contributing to poverty in many parts of the world. Many field experiments have been performed over the past century to study spatio-temporal patterns of soil erosion caused by surface runoff under different environmental conditions. However, these data have never been integrated together in a way that can inform efforts to understand and model soil erosion at different spatial and temporal scales. Here, we designed a database titled AWESOME: Archive for Water Erosion and Sediment Outflow Measurements (Jian et al., 2022). The AWESOME database compiles field measurements of annual soil erosion and sediment yield caused by surface runoff, with data derived from sites around the globe. It includes four soil erosion-related indicators (surface runoff, annual erosion, annual sediment yield, and soil nutrient loss) and more than sixty variables for meta-data that describe the location, climate, soil properties, experimental design (e.g., soil erosion measurement method, field scale, replication), and bibliographic information (e.g., author name and year of publication). Currently, measurements from 1985 geographic sites with unique combinations of longitude and latitude, representing 75 countries, have been compiled into AWESOME. We provide an example of linking AWESOME with an external climate dataset, and identify correlations between soil erosion and several environmental variables. Annual soil erosion rates were most influenced by vegetation type and soil texture group. Annual soil erosion rates exhibited significant negative relationships with plant coverage, soil clay content, soil pH, and soil organic carbon content, and significant positive relationships with annual precipitation and soil bulk density. AWESOME aims to be a freely available, and global framework for compiling field soil erosion and subsequent sediment yield measurements, and to provide data sources to support statistical evaluations, model validation and applications, as well as a better understanding of spatial and temporal patterns of soil erosion.
- Preprint
(1388 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on essd-2022-87', Pia Benaud, 06 Apr 2022
General Comments:
Soil erosion by water is a complex process, resulting from the interaction between a number of environmental variables and land management decisions, making it challenging to predict rates of soil erosion. Testing soil erosion models against empirical data is therefore essential to improving modelling accuracy. However, accurately measuring rates of soil erosion at local or national scales is incredibly costly. Collating existing data into an open access database is therefore a very useful endeavour and the authors have clearly gone to considerable effort compiling data, amassing results from 1985 different geographic sites, along with useful records of environmental and study variables. However, the usefulness of such a database for modelling applications will inevitably limited by the reliability of the data, so rigorous quality control is crucial. Consistently extracting data from studies is incredibly challenging, particularly when there is a lot of inconsistency in the methods used to collect the empirical data. Unfortunately, there needs to be some improvements to the database in that regard and revisions to the manuscript before the article and database should be accepted for publication.
For full transparency, I recently compiled and published a similar database, however it was focussed on soil erosion observations in the UK (Benaud et al., 2020). Accordingly, I focussed my assessment on the quality of the UK observational data, and noticed a number of discrepancies. Brazier et al., (2001), for example, is not a first-hand soil erosion study. Through not seeking the original research, they have mis-reported details. For example, records 3832-3848 are not runoff plot data, they were volumetrically estimated from overflight surveys of the regions listed. The work of Chambers and Garwood (2000) is also incorrectly reported – the measurement method was a field survey, not a gauging station, and the results are therefore not sediment yields. Some of Walling et al. (2002) is indeed based Ceasium-137 observations, but 4526-4538 are suspended sediment yields. More work needs to be done to properly interrogate the data sources to match the claim on line 250.
The manuscript needs more information on the decisions surrounding how results were standardised for inclusion in the database. For example, Walling et al. (2002) report ‘gross’ and ‘net’ erosion rates, as frequently done with Caesium-137 studies, and the database contains the gross rate. While, Chambers and Garwood (2000) report both the mean and median ‘net’ erosion rates, and the database contains the mean rate. The “UnitsConverter” tab suggests an assumed bulk density of 1.5 g cm-3 was used to convert volumetric measurements, why was this decision taken? Some of these types of discrepancies are inevitable when compiling such a large dataset, however there needs to be a clear description of the rationale behind these decisions in the manuscript. The manuscript would also benefit from the results being contextualised with other soil erosion studies – either modelling or other databases.
Specific comments:
Introduction
The introduction does a good job explaining the background and need for the database, but is currently too long – paragraphs 3 and 4 could be condensed. At the risk of promoting citation of my own work, I’d suggest it would be appropriate to cite Benaud et al. (2020) as a key aim of the paper was to collate all available, empirically-derived soil erosion datasets into a spatially explicit and open access resource, albeit with a UK focus. AWESOME is the first database that I know of that does so on a global scale, with as much detail. It would be useful to outline the aims/objectives of the paper.
Methods
This contains lots of useful information, however, as above, more information is needed on the decisions taken when selecting or standardising soil erosion measurements. I assume you log-transformed (line 284) the data to account for the skewed nature of soil erosion observations?
Results
Figure 2 needs to be updated – it’s trying to display too much information. Number of observations binned by region (see Fig1 Garcia-Ruiz et al., 2015) or whichever factor you think is most important would be more appropriate.
Figure 3 – boxplots would be more appropriate here to show the true distribution of the data. I alsondon’t think it is fair/useful to compare erosion pins with runoff plot measurements here.
Line 341 – Given how skewed soil erosion data is, it would be appropriate to also report median rates.
3.4 – This section seems a strange inclusion in the manuscript – it’s not introduced elsewhere in the manuscript. Details need to be added to the methods, and you need to explain why you have done this work. You also need to quantify “the two datasets matched each [other] well”
Discussion
The discussion covers some interesting points, but needs more work to improve its quality/value. At present, it misses the context provided important existing soil erosion research papers. Some key points:
400 – Plot studies are known to create bias due to their short length and boundaries (see Parsons et al., 2006, and subsequent papers)
415 – I suggest you have a look at the original papers in more details, and consider the environment where they were tested.
Paragraphs between 425 and 435 – it’s also important to consider channel bank erosion.
435 – vegetation cover also plays a significant role and the distribution of slopes in your data.
440 – this needs a better explanation
450 – Yes, the database is great for integrating with spatial datasets. The thing to consider is what the climatic data in the database represents i.e. is it say long-term average precipitation or is it precipitation during the erosion experiment.
Section 4.2 – How do your results match global models? A lot of the content here is not relevant.
Table 6 should be in the results
4.3 Great ambition! Though this section could do with being a little more formal.
Thanks for sharing the data. The R code could use some editing to make it fully reproducible – some of the data links are to local drives, for example.
The conclusion would benefit from stated aims/objectives in the introduction.
Author contributions needs some clarification.
In summary, the database represents significant effort from the authors, and could be a very useful and gratefully received resource for the soil erosion modelling community. However, there needs to be greater quality control carried out on the data – particularly looking at the original sources – and the manuscript needs to be improved to better support the database and its application.
References:
Benaud, P., Anderson, K., Evans, M., Farrow, L., Glendell, M., James, M. R., Quine, T. A., Quinton, J. N., Rawlins, B., Jane Rickson, R., & Brazier, R. E. (2020). National-scale geodata describe widespread accelerated soil erosion. Geoderma, 371, 114378. https://doi.org/10.1016/j.geoderma.2020.114378
Parsons, A. J., Brazier, R. E., Wainwright, J., & Powell, D. M. (2006). Scale relationships in hillslope runoff and erosion. Earth Surface Processes and Landforms, 31(11), 1384–1393. https://doi.org/10.1002/esp.1345
Citation: https://doi.org/10.5194/essd-2022-87-RC1 -
RC2: 'Comment on essd-2022-87', Anonymous Referee #2, 16 May 2022
This could be an interesting article. The authors restricted the authorship of such an article in few contributors ignoring the soil conservation community. For such an effort, you have to be inclusive and make a call for all scientist who have contributed in such data collection.
On the opposite, you preferred to have few data contributors and it is un clear how you go the data from many many scientists who have made such experiments.
In addition, you miss colleagues from Latin America, Australia, Africa and many European colleagues.
Finally, I found many speculations in the paper which should not be the case for such a high profile article.
Finally, the authors did not address at all the use of sediment data and nutrient losses.
For all those major reasons, I do not consider appropriate the publication of the article.
L79: I do not believe that those data have lead to 30,000 papers. For sure you mean something else or you try to speculate here.
L112: There are other forms of erosion: loss by harvest crops, gully, etc.
L139: Why only English and Chinese and not Spanish or French?
Table 2: why the maximum soil erosion rate is different than the annual one (which should be mean annual soil erosion rate)? It is better to have the same unit for making comparisons.
Figure 3: The results in 3a are really too high. With such mean values in croplands, you will not have harvest not for the next 60 years but only for the next 20 years. I do not think that such assumptions and results are realistic if you do not put them in the right context.! The same applies for lines 340-342. The rates are too far un realistic.
The worst stuff is that you try to extrapolate such results for the whole globe.
L331: It is impossible that the third more correlative parameter on soil erosion is pH. For any geomorphologist or soil conservationist this has no sense.
Another important aspect is the complete absence of management and soil erosion conservation practices.
Also in figure 4, it is impossible to correlate the soil erosion with Bulk density. The annual precipitation is not the most appropritate driver and should be replaced by erosivity indexes.
The relationships in figure 5 are very poor.
The high uncertainty of such data is discussed briefly in 414-415 . however , very briefly.
Therefore, with such low correlations , you cannot produce statistical equations as in Table 6.
Forecasting global soil erosion is not a right term.
If we follow the European example with the publication of Cerdan and your objective to develop something similar at global scale , I am really skeptical of such efforts. It is known the high problematics of the Cerdan dataset where few plot data were extrapolated to the whole Europe. This had as a result to have Denmark with higher erosion rates than Spain or Italy.
Such data can be used to calibration and validation of data and not for developing spatial data with extrapolations.
Citation: https://doi.org/10.5194/essd-2022-87-RC2
Status: closed
-
RC1: 'Comment on essd-2022-87', Pia Benaud, 06 Apr 2022
General Comments:
Soil erosion by water is a complex process, resulting from the interaction between a number of environmental variables and land management decisions, making it challenging to predict rates of soil erosion. Testing soil erosion models against empirical data is therefore essential to improving modelling accuracy. However, accurately measuring rates of soil erosion at local or national scales is incredibly costly. Collating existing data into an open access database is therefore a very useful endeavour and the authors have clearly gone to considerable effort compiling data, amassing results from 1985 different geographic sites, along with useful records of environmental and study variables. However, the usefulness of such a database for modelling applications will inevitably limited by the reliability of the data, so rigorous quality control is crucial. Consistently extracting data from studies is incredibly challenging, particularly when there is a lot of inconsistency in the methods used to collect the empirical data. Unfortunately, there needs to be some improvements to the database in that regard and revisions to the manuscript before the article and database should be accepted for publication.
For full transparency, I recently compiled and published a similar database, however it was focussed on soil erosion observations in the UK (Benaud et al., 2020). Accordingly, I focussed my assessment on the quality of the UK observational data, and noticed a number of discrepancies. Brazier et al., (2001), for example, is not a first-hand soil erosion study. Through not seeking the original research, they have mis-reported details. For example, records 3832-3848 are not runoff plot data, they were volumetrically estimated from overflight surveys of the regions listed. The work of Chambers and Garwood (2000) is also incorrectly reported – the measurement method was a field survey, not a gauging station, and the results are therefore not sediment yields. Some of Walling et al. (2002) is indeed based Ceasium-137 observations, but 4526-4538 are suspended sediment yields. More work needs to be done to properly interrogate the data sources to match the claim on line 250.
The manuscript needs more information on the decisions surrounding how results were standardised for inclusion in the database. For example, Walling et al. (2002) report ‘gross’ and ‘net’ erosion rates, as frequently done with Caesium-137 studies, and the database contains the gross rate. While, Chambers and Garwood (2000) report both the mean and median ‘net’ erosion rates, and the database contains the mean rate. The “UnitsConverter” tab suggests an assumed bulk density of 1.5 g cm-3 was used to convert volumetric measurements, why was this decision taken? Some of these types of discrepancies are inevitable when compiling such a large dataset, however there needs to be a clear description of the rationale behind these decisions in the manuscript. The manuscript would also benefit from the results being contextualised with other soil erosion studies – either modelling or other databases.
Specific comments:
Introduction
The introduction does a good job explaining the background and need for the database, but is currently too long – paragraphs 3 and 4 could be condensed. At the risk of promoting citation of my own work, I’d suggest it would be appropriate to cite Benaud et al. (2020) as a key aim of the paper was to collate all available, empirically-derived soil erosion datasets into a spatially explicit and open access resource, albeit with a UK focus. AWESOME is the first database that I know of that does so on a global scale, with as much detail. It would be useful to outline the aims/objectives of the paper.
Methods
This contains lots of useful information, however, as above, more information is needed on the decisions taken when selecting or standardising soil erosion measurements. I assume you log-transformed (line 284) the data to account for the skewed nature of soil erosion observations?
Results
Figure 2 needs to be updated – it’s trying to display too much information. Number of observations binned by region (see Fig1 Garcia-Ruiz et al., 2015) or whichever factor you think is most important would be more appropriate.
Figure 3 – boxplots would be more appropriate here to show the true distribution of the data. I alsondon’t think it is fair/useful to compare erosion pins with runoff plot measurements here.
Line 341 – Given how skewed soil erosion data is, it would be appropriate to also report median rates.
3.4 – This section seems a strange inclusion in the manuscript – it’s not introduced elsewhere in the manuscript. Details need to be added to the methods, and you need to explain why you have done this work. You also need to quantify “the two datasets matched each [other] well”
Discussion
The discussion covers some interesting points, but needs more work to improve its quality/value. At present, it misses the context provided important existing soil erosion research papers. Some key points:
400 – Plot studies are known to create bias due to their short length and boundaries (see Parsons et al., 2006, and subsequent papers)
415 – I suggest you have a look at the original papers in more details, and consider the environment where they were tested.
Paragraphs between 425 and 435 – it’s also important to consider channel bank erosion.
435 – vegetation cover also plays a significant role and the distribution of slopes in your data.
440 – this needs a better explanation
450 – Yes, the database is great for integrating with spatial datasets. The thing to consider is what the climatic data in the database represents i.e. is it say long-term average precipitation or is it precipitation during the erosion experiment.
Section 4.2 – How do your results match global models? A lot of the content here is not relevant.
Table 6 should be in the results
4.3 Great ambition! Though this section could do with being a little more formal.
Thanks for sharing the data. The R code could use some editing to make it fully reproducible – some of the data links are to local drives, for example.
The conclusion would benefit from stated aims/objectives in the introduction.
Author contributions needs some clarification.
In summary, the database represents significant effort from the authors, and could be a very useful and gratefully received resource for the soil erosion modelling community. However, there needs to be greater quality control carried out on the data – particularly looking at the original sources – and the manuscript needs to be improved to better support the database and its application.
References:
Benaud, P., Anderson, K., Evans, M., Farrow, L., Glendell, M., James, M. R., Quine, T. A., Quinton, J. N., Rawlins, B., Jane Rickson, R., & Brazier, R. E. (2020). National-scale geodata describe widespread accelerated soil erosion. Geoderma, 371, 114378. https://doi.org/10.1016/j.geoderma.2020.114378
Parsons, A. J., Brazier, R. E., Wainwright, J., & Powell, D. M. (2006). Scale relationships in hillslope runoff and erosion. Earth Surface Processes and Landforms, 31(11), 1384–1393. https://doi.org/10.1002/esp.1345
Citation: https://doi.org/10.5194/essd-2022-87-RC1 -
RC2: 'Comment on essd-2022-87', Anonymous Referee #2, 16 May 2022
This could be an interesting article. The authors restricted the authorship of such an article in few contributors ignoring the soil conservation community. For such an effort, you have to be inclusive and make a call for all scientist who have contributed in such data collection.
On the opposite, you preferred to have few data contributors and it is un clear how you go the data from many many scientists who have made such experiments.
In addition, you miss colleagues from Latin America, Australia, Africa and many European colleagues.
Finally, I found many speculations in the paper which should not be the case for such a high profile article.
Finally, the authors did not address at all the use of sediment data and nutrient losses.
For all those major reasons, I do not consider appropriate the publication of the article.
L79: I do not believe that those data have lead to 30,000 papers. For sure you mean something else or you try to speculate here.
L112: There are other forms of erosion: loss by harvest crops, gully, etc.
L139: Why only English and Chinese and not Spanish or French?
Table 2: why the maximum soil erosion rate is different than the annual one (which should be mean annual soil erosion rate)? It is better to have the same unit for making comparisons.
Figure 3: The results in 3a are really too high. With such mean values in croplands, you will not have harvest not for the next 60 years but only for the next 20 years. I do not think that such assumptions and results are realistic if you do not put them in the right context.! The same applies for lines 340-342. The rates are too far un realistic.
The worst stuff is that you try to extrapolate such results for the whole globe.
L331: It is impossible that the third more correlative parameter on soil erosion is pH. For any geomorphologist or soil conservationist this has no sense.
Another important aspect is the complete absence of management and soil erosion conservation practices.
Also in figure 4, it is impossible to correlate the soil erosion with Bulk density. The annual precipitation is not the most appropritate driver and should be replaced by erosivity indexes.
The relationships in figure 5 are very poor.
The high uncertainty of such data is discussed briefly in 414-415 . however , very briefly.
Therefore, with such low correlations , you cannot produce statistical equations as in Table 6.
Forecasting global soil erosion is not a right term.
If we follow the European example with the publication of Cerdan and your objective to develop something similar at global scale , I am really skeptical of such efforts. It is known the high problematics of the Cerdan dataset where few plot data were extrapolated to the whole Europe. This had as a result to have Denmark with higher erosion rates than Spain or Italy.
Such data can be used to calibration and validation of data and not for developing spatial data with extrapolations.
Citation: https://doi.org/10.5194/essd-2022-87-RC2
Data sets
Jinshijian/AWESOME: Archive for Water Erosion and Sediment Outflow MEasurements (v2.0.0) Jinshi Jian, Xuan Du, Juying Jiao, Xiaohua Ren, Karl Auerswald, Ryan D. Stewart, Zeli Tan, Jianlin Zhao, Daniel L. Evans, Guangju Zhao, Nufang Fang, Wenyi Sun, Chao Yue, & Ben Bond-Lamberty https://doi.org/10.5281/zenodo.6324809
Model code and software
Jinshijian/AWESOME: Archive for Water Erosion and Sediment Outflow MEasurements (v2.0.0) Jinshi Jian, Xuan Du, Juying Jiao, Xiaohua Ren, Karl Auerswald, Ryan D. Stewart, Zeli Tan, Jianlin Zhao, Daniel L. Evans, Guangju Zhao, Nufang Fang, Wenyi Sun, Chao Yue, & Ben Bond-Lamberty https://doi.org/10.5281/zenodo.6324809
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 896 | 342 | 70 | 1,308 | 68 | 83 |
- HTML: 896
- PDF: 342
- XML: 70
- Total: 1,308
- BibTeX: 68
- EndNote: 83
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1