the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Assessing methods for representing soil heterogeneity through a flexible approach within the Joint UK Land Environment Simulator (JULES) at version 3.4.1
Heather Suzanne Rumbold
Richard J. J. Gilham
Martin John Best
Abstract. The interactions between the land surface and the atmosphere can impact weather and climate through the exchanges of water, energy, carbon and momentum. The properties of the land surface are important when modelling these exchanges correctly especially with models being used at increasingly higher resolution. The Joint UK Land Environment Simulator (JULES) currently uses a tiled representation of land cover but can only model a single dominant soil type within a grid box. Hence, there is no representation of sub-grid scale soil heterogeneity. This paper introduces and evaluates a new flexible surface-soil tiling scheme in JULES. Several different soil tiling approaches are presented for a synthetic case study. The changes to model performance have been compared to the current single soil scheme and a high resolution 'Truth' scenario. Results have shown that the different soil tiling strategies do have an impact on the water and energy exchanges due to the way vegetation accesses the soil moisture. Tiling the soil according to the surface type, with the soil properties set to the dominant soil type under each surface is the best performing configuration. The results from this setup simulate water and energy fluxes that are the closest to the high resolution 'Truth' scenario but require much less information on the soil type than the high resolution soil configuration.
Heather Suzanne Rumbold et al.
Status: closed
-
CC1: 'Comment on gmd-2022-139', Dalei Hao, 05 Aug 2022
Interesting work! I have some minor comments for the authors' considerations:
1. Is the testbed grid box from real word or articifical assumptions? Please give more details.
2. Whether was the heterogeneity of soil organic matter considered?
3. How does JULES calculate soil albedo for different soil types?
4. Apart from the LSMs listed in Table 1, E3SM land model (ELM) can represent the soil heterogeneity at different topographic units under a novel topography-based sub-grid structure (Hao et al., 2022).
Hao, D., Bisht, G., Huang, M., Ma, P.-L., Tesfa, T., Lee, W.-L., et al. (2022). Impacts of sub-grid topographic representations on surface energy balance and boundary conditions in the E3SM land model: A case study in Sierra Nevada. Journal of Advances in Modeling Earth Systems, 14, e2021MS002862. https://doi.org/10.1029/2021MS002862
5. I am curious why the lines for SurfDom and SurfGB overlap with the HResTex run in all variables but HResTexAgg has some differences from the HResTex. Please explain it.
-
RC1: 'Comment on gmd-2022-139', Anonymous Referee #1, 23 Sep 2022
General comments
The paper looks at how sub-grid scale soil heterogeneity can be added to a complex land-surface model. Using a synthetic example, different configurations are tested, exploring both increased heterogeneity and computational efficiency.
It is an interesting development shown to have an important impact on model vegetation-soil moisture interactions, especially at high resolutions. The paper is well written and a good fit for GMD. My comments are mainly about clarifying parts of the manuscript.
Specific comments
How was the synthetic example created? As in, how were the different fractions of PFTs and soil textures chosen? Is it based on the UK site where the meteorological forcing was chosen?
What resolution is the full grid box meant to represent? 0.5 degrees like the forcing?
For HResTexAgg, how are the interactions distributed? For example, it is mentioned that moisture infiltrating from BLT is distributed between the clay and loam. Is this distribution even, i.e., 50:50? Or is it proportional to the fraction of soil texture? i.e., 16/26 to clay and 10/26 to loam in this case.
L180: Can the authors comment more on the fact that SurfGB and SurfDom match HRexTex in Fig. 3? I realise this is discussed more at the end of this section, but I think a sentence here explaining how they all have a one-to-one mapping would help the reader.
Throughout the plots and analysis, four layers are discussed. However, I don’t think the concept of layers is introduced. How many layers total make up the soil column in JULES? How deep and thick is each respective layer?
Since it is a synthetic example, it cannot be evaluated against observations. However, maybe the authors could comment in the conclusion on how future work could use observations. Furthermore, only one climate is tested (mid-latitude temperate). Could the authors comment on how the results would change for a different climate? For example, what does one might expect results to be for an arid site?
Technical corrections
Throughout: change quotes ' to `
Throughout: sub grid vs sub-grid
L18: in the last couple
L26: tiled
L30: Do you mean representative parameter values? Or additional parameters on top of the parameter set used in the mosaic approach?
L62: remove extra brackets around the citation
L102|L107|L146|L156: add missing , after i.e. to be consistent with the rest of the text
L126: is not
L184: autumn is not a proper noun
L194: Clay does not need to be capitalised
L204: “\” missing in the latex maths mode for beta
L225: Is “Fig.’s” the correct shortening for multiple figures?
L266: does not
L293: These results
Fig.s 3&5: superscript is needed for the units
Fg. 4: beta as a symbol?
-
RC2: 'Comment on gmd-2022-139', Anonymous Referee #2, 29 Sep 2022
General comments
Rumbold et al. describes improvements to the soil tiling scheme at different levels of complexity in the land surface model JULES. The preferred soil tiling scheme is determined by balancing the available resolution of soil types/surface types in a grid box and computational time required. A synthetic example grid box located in the UK is used to illustrate the effect on energy and moisture fluxes using the four different improved soil tiling methods along with the original simplistic method. Conclusions about the required compexity of the soil tiling scheme is mode based on this synthetic example grid box.
The improvement to the model is a highly desirable one and the methods and experimental setup is descibed in a clear and logical manner. Some improvements to the text should be done, especially avoiding repetitions. To the reader not familiar with the detailed representation of vegetation in JULES, some added information in this regard would help to ascertain the theoretical maximum complexity of soil/surface connectivity.
Specific comments
The complexity of the vegetation in the model should, to some extent, determine the ideal level of complexity of the soil tiling. How is vegetation demography represented in JULES ? Would the different pixels within a surface type in Fig.1 have different age distributions or different land use histories ? If yes, the “mosaic” soil tiling approach would perhaps be the ideal alternative, not considereing data availability and computational limits. If no, the tiling approach would be enough. It would be helpful for the general reader if some information regarding this question was added.
Can the version used, 3.4.1, be related to newer versions, which seem to include e.g. managed forests (?) and cropland as well as land-cover/land-use change ? Does v.3.4.1 contain any of these capabilities ? As already mentioned. these features would seem to influence the necessary complexity of the surface/soil tilling scheme.
What is the temporal resolution of the processes in the model ? It is not described (or it evaded me) how the consumption of water by vegetation from a single soil tile works with multiple surface tiles as in the DC and HResTexAgg options. A naïve reader’s guess could be that the interactions with the different tiles are done sequentially in the code, which would make the question of temporal resolution relevant.
(Noticing that this is a special JULES issue, I realise that some of these questions are probably answered by other articles in this issue, so they may not be so critical.)
Why are not carbon fluxes considered ? Is it because the focus is on meterological rather than climate applications ? This should be mentioned.
I’m not sure the format of this journal requires it, but ideally, example grid cells from other climates would seem necessary. Unless the intended usage would be for the UK only, but this should be mentioned in that case.
Also ideally, real-case example sites using real land cover and soil type data at different resolutions would be beneficial to illustrate the significance of selecting one soil tiling scheme over the other.
L.203: “the addition of soil tiles (and therefore more soil columns) has allowed each surface tile to have different rooting profiles and rates of water extraction.” How does this harmonise with L.71 “The root density is assumed to follow an exponential distribution with depth, with the depth scale varying between the different PFTs.”, which I assume is independent of the soil tiling method ?
In section 3, the text should be pruned much more stringently to avoid repetitions (see some examples below). Section 4 seems to repeat a lot of section 3, but in a much more readable form. I wonder if section 3 can be shortened significantly, e.g. removing the explanations that are repeated in section 4 (keeping section 4 as is).
Technical corrections
L.41: “Due to the non-linear nature of soil processes, the dominant soil type is used for each grid box and soil parameters associated with this soil type are then used.” What would the alternative be when only using one soil tile ? A soil type with some sort of weighting of the different soil parameters ? Perhaps this is the “aggregated” soil properties used in the CLS and ISBA models, but it reads a bit obscure in the text before this is mentioned a few lines later.
L.69, 72, 191, Fig.4: β is called “soil moisture availability factor” on L.69 and in Fig.4 and “soil moisture stress factor” on L.72 and L.191
L.70, 72: The definition of β is split into two sentences, surrounding a description of root density. Can the first sentence be merged with the second ?
L.190: “are gradually become”
L.191-197 (and further on). description of the line colours and styles in the text seems a bit redundant
L.204: β written as “beta”
L.210-212, L.259-260: Repetition of the same information.
L.222-227: Seems to be a lot of redundant information in these sentences. Compress ?
L.235-237, L.248-249. Repeating more or less the same thing.
L.239-245, L.249-251: Repeating the same thing, but less detailed.
-
AC1: 'Comment on gmd-2022-139', Heather Rumbold, 01 Nov 2022
We would like to thank the two reviewers and Dalei Hao for their time to read and comment on this manuscript.
We have addressed their comments in the attached document, where reveiwer comments are in plain black text while our responses are in blue text.
Status: closed
-
CC1: 'Comment on gmd-2022-139', Dalei Hao, 05 Aug 2022
Interesting work! I have some minor comments for the authors' considerations:
1. Is the testbed grid box from real word or articifical assumptions? Please give more details.
2. Whether was the heterogeneity of soil organic matter considered?
3. How does JULES calculate soil albedo for different soil types?
4. Apart from the LSMs listed in Table 1, E3SM land model (ELM) can represent the soil heterogeneity at different topographic units under a novel topography-based sub-grid structure (Hao et al., 2022).
Hao, D., Bisht, G., Huang, M., Ma, P.-L., Tesfa, T., Lee, W.-L., et al. (2022). Impacts of sub-grid topographic representations on surface energy balance and boundary conditions in the E3SM land model: A case study in Sierra Nevada. Journal of Advances in Modeling Earth Systems, 14, e2021MS002862. https://doi.org/10.1029/2021MS002862
5. I am curious why the lines for SurfDom and SurfGB overlap with the HResTex run in all variables but HResTexAgg has some differences from the HResTex. Please explain it.
-
RC1: 'Comment on gmd-2022-139', Anonymous Referee #1, 23 Sep 2022
General comments
The paper looks at how sub-grid scale soil heterogeneity can be added to a complex land-surface model. Using a synthetic example, different configurations are tested, exploring both increased heterogeneity and computational efficiency.
It is an interesting development shown to have an important impact on model vegetation-soil moisture interactions, especially at high resolutions. The paper is well written and a good fit for GMD. My comments are mainly about clarifying parts of the manuscript.
Specific comments
How was the synthetic example created? As in, how were the different fractions of PFTs and soil textures chosen? Is it based on the UK site where the meteorological forcing was chosen?
What resolution is the full grid box meant to represent? 0.5 degrees like the forcing?
For HResTexAgg, how are the interactions distributed? For example, it is mentioned that moisture infiltrating from BLT is distributed between the clay and loam. Is this distribution even, i.e., 50:50? Or is it proportional to the fraction of soil texture? i.e., 16/26 to clay and 10/26 to loam in this case.
L180: Can the authors comment more on the fact that SurfGB and SurfDom match HRexTex in Fig. 3? I realise this is discussed more at the end of this section, but I think a sentence here explaining how they all have a one-to-one mapping would help the reader.
Throughout the plots and analysis, four layers are discussed. However, I don’t think the concept of layers is introduced. How many layers total make up the soil column in JULES? How deep and thick is each respective layer?
Since it is a synthetic example, it cannot be evaluated against observations. However, maybe the authors could comment in the conclusion on how future work could use observations. Furthermore, only one climate is tested (mid-latitude temperate). Could the authors comment on how the results would change for a different climate? For example, what does one might expect results to be for an arid site?
Technical corrections
Throughout: change quotes ' to `
Throughout: sub grid vs sub-grid
L18: in the last couple
L26: tiled
L30: Do you mean representative parameter values? Or additional parameters on top of the parameter set used in the mosaic approach?
L62: remove extra brackets around the citation
L102|L107|L146|L156: add missing , after i.e. to be consistent with the rest of the text
L126: is not
L184: autumn is not a proper noun
L194: Clay does not need to be capitalised
L204: “\” missing in the latex maths mode for beta
L225: Is “Fig.’s” the correct shortening for multiple figures?
L266: does not
L293: These results
Fig.s 3&5: superscript is needed for the units
Fg. 4: beta as a symbol?
-
RC2: 'Comment on gmd-2022-139', Anonymous Referee #2, 29 Sep 2022
General comments
Rumbold et al. describes improvements to the soil tiling scheme at different levels of complexity in the land surface model JULES. The preferred soil tiling scheme is determined by balancing the available resolution of soil types/surface types in a grid box and computational time required. A synthetic example grid box located in the UK is used to illustrate the effect on energy and moisture fluxes using the four different improved soil tiling methods along with the original simplistic method. Conclusions about the required compexity of the soil tiling scheme is mode based on this synthetic example grid box.
The improvement to the model is a highly desirable one and the methods and experimental setup is descibed in a clear and logical manner. Some improvements to the text should be done, especially avoiding repetitions. To the reader not familiar with the detailed representation of vegetation in JULES, some added information in this regard would help to ascertain the theoretical maximum complexity of soil/surface connectivity.
Specific comments
The complexity of the vegetation in the model should, to some extent, determine the ideal level of complexity of the soil tiling. How is vegetation demography represented in JULES ? Would the different pixels within a surface type in Fig.1 have different age distributions or different land use histories ? If yes, the “mosaic” soil tiling approach would perhaps be the ideal alternative, not considereing data availability and computational limits. If no, the tiling approach would be enough. It would be helpful for the general reader if some information regarding this question was added.
Can the version used, 3.4.1, be related to newer versions, which seem to include e.g. managed forests (?) and cropland as well as land-cover/land-use change ? Does v.3.4.1 contain any of these capabilities ? As already mentioned. these features would seem to influence the necessary complexity of the surface/soil tilling scheme.
What is the temporal resolution of the processes in the model ? It is not described (or it evaded me) how the consumption of water by vegetation from a single soil tile works with multiple surface tiles as in the DC and HResTexAgg options. A naïve reader’s guess could be that the interactions with the different tiles are done sequentially in the code, which would make the question of temporal resolution relevant.
(Noticing that this is a special JULES issue, I realise that some of these questions are probably answered by other articles in this issue, so they may not be so critical.)
Why are not carbon fluxes considered ? Is it because the focus is on meterological rather than climate applications ? This should be mentioned.
I’m not sure the format of this journal requires it, but ideally, example grid cells from other climates would seem necessary. Unless the intended usage would be for the UK only, but this should be mentioned in that case.
Also ideally, real-case example sites using real land cover and soil type data at different resolutions would be beneficial to illustrate the significance of selecting one soil tiling scheme over the other.
L.203: “the addition of soil tiles (and therefore more soil columns) has allowed each surface tile to have different rooting profiles and rates of water extraction.” How does this harmonise with L.71 “The root density is assumed to follow an exponential distribution with depth, with the depth scale varying between the different PFTs.”, which I assume is independent of the soil tiling method ?
In section 3, the text should be pruned much more stringently to avoid repetitions (see some examples below). Section 4 seems to repeat a lot of section 3, but in a much more readable form. I wonder if section 3 can be shortened significantly, e.g. removing the explanations that are repeated in section 4 (keeping section 4 as is).
Technical corrections
L.41: “Due to the non-linear nature of soil processes, the dominant soil type is used for each grid box and soil parameters associated with this soil type are then used.” What would the alternative be when only using one soil tile ? A soil type with some sort of weighting of the different soil parameters ? Perhaps this is the “aggregated” soil properties used in the CLS and ISBA models, but it reads a bit obscure in the text before this is mentioned a few lines later.
L.69, 72, 191, Fig.4: β is called “soil moisture availability factor” on L.69 and in Fig.4 and “soil moisture stress factor” on L.72 and L.191
L.70, 72: The definition of β is split into two sentences, surrounding a description of root density. Can the first sentence be merged with the second ?
L.190: “are gradually become”
L.191-197 (and further on). description of the line colours and styles in the text seems a bit redundant
L.204: β written as “beta”
L.210-212, L.259-260: Repetition of the same information.
L.222-227: Seems to be a lot of redundant information in these sentences. Compress ?
L.235-237, L.248-249. Repeating more or less the same thing.
L.239-245, L.249-251: Repeating the same thing, but less detailed.
-
AC1: 'Comment on gmd-2022-139', Heather Rumbold, 01 Nov 2022
We would like to thank the two reviewers and Dalei Hao for their time to read and comment on this manuscript.
We have addressed their comments in the attached document, where reveiwer comments are in plain black text while our responses are in blue text.
Heather Suzanne Rumbold et al.
Data sets
WATCH Forcing Data methodology applied to ERA-Interim (WFDEI) data (Meteorological Forcing Data) G. P. Weedon, G. Balsamo, N. Bellouin, S. Gomes, M. J. Best, and P. Viterbo ftp://rfdata:forceDATA@ftp.iiasa.ac.at
Assessing methods for representing soil heterogeneity through a flexible approach within the Joint UK Land Environment Simulator (JULES) at version 3.4.1 (Data sets). Heather Rumbold, Richard Gilham, Martin Best https://doi.org/10.5281/zenodo.6954142
Model code and software
Joint UK Land Environment Simulator (JULES) model code branch at version 3.4.1 Heather Rumbold, Richard Gilham, Martin Best https://code.metoffice.gov.uk/trac/jules/browser/main/branches/dev/heatherashton/vn3.4.1_soil_tiling/configurations?rev=23611
Heather Suzanne Rumbold et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 571 | 104 | 16 | 691 | 7 | 6 |
- HTML: 571
- PDF: 104
- XML: 16
- Total: 691
- BibTeX: 7
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1