the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Feb 2023
A generalised ecohydrological landscape classification for assessing ecosystem risk in Australia due to an altering water regime
Abstract. Describing and classifying a landscape for environmental impact and risk assessment purposes is a non-trivial challenge, as standard landscape classifications that cater for region specific impacts do not exist. Assessing impacts on ecosystems from extraction of water resources across large regions requires linking of landscape features to their water requirements. We present the rationale and implementation of an ecohydrological classification for regions where coal mine and coal seam gas developments may impact on water. Our classification provides the essential framework for modelling the potential impact of hydrological changes from future coal resource developments at the landscape level.
We develop an attribute-based system that provides representations of the ecohydrological entities and their connection to landscape features and make use of existing broad-level, classification schemes into an attribute-based system. We incorporate a rule-set with prioritisation, which underpin risk modelling and make the scheme resource efficient, where spatial landscape or ecosystem classification schemes, developed for other purposes, already exist.
A consistent rule-set and conceptualised landscape processes and functions allow combining diverse data with existing classification schemes. This makes the classification transparent, repeatable, and adjustable, should new data become available. We apply the approach in three geographically different regions, with widely disparate information sources for the classification and provide a detailed example of its application. We propose that it is widely applicable around the world for linking ecohydrology to environmental impacts.
- Preprint
(5264 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on hess-2022-408', Willem Vervoort, 09 Apr 2023
This manuscript describes a classification framework to classify landscapes into different categories which can subsequently be used for conceptual modelling.
I am afraid that after reading the manuscript I am still unsure how the framework actually works. In my opinion the methodology is not well described and it is unclear how decisions were made for different classes. At some point it seemed that the main framework that was being described was a GIS overlay, but how the different features were weighted is unclear, or I must have missed this. The manuscript regularly refers to other papers in the methods and the result without explaining how exactly this was integrated in the current paper and was used in the framework. As a result, it is not clear how the results were actually derived. From a reproducibility perspective, I think it would be hard to replicate the results.
I am left wondering how decisions were made about different classes, where these simply in the original data, or were those classes decided on in this study. It almost seems like this study was the culmination of a series of other studies, but these studies (while referenced) are not discussed in the paper. I think there needs to be a much clearer methodology and workflow to be able to reproduce the results and to make the paper easier to read and understand.
Here is an example of some of the unclear discussion (l408 and further):
“The modelling of risk to ecosystems at regional scale focuses on recognising which parts of the region are potentially impacted and which parts are unlikely to experience harm. Using our landscape classification as a crucial input, the modelling delineated impacted areas within each region, based on a zone of potential hydrological change.”
From this, I fail to understand how the classification was aa “crucial input” and how this assisted in delineating the impacted areas. There is an earlier reference to Hosack et al., is this the paper that describes the modelling? It would still be useful to help the reader understand what the modelling was (Summarising the earlier study) and highlighting how it was shown that the classification was a “crucial input”.
Another example from the start of the methodology, where essentially the overall approach is summarised (l184..):
“The purpose of this ecohydrological landscape classification is to characterise the landscape based on patterns in land use, ecology, geomorphology and hydrology, and from these, develop landscape classes of water-dependent, remnant and human-modified features. Existing spatial data for each region forms the basis for categorising the landscape features using a rule-set based on attribute features within the spatial datasets.”
The first problem I have is that why landscape classes of “water-dependent, remnant and human-modified features” are chosen doesn’t seem to be explained. I can see that this is a useful classification, but at least some rational for the choice (and why no other classes) should be presented. The second problem is the references to a “rule-set”. I presumed this was going to be discussed later in the paper, but either I have totally missed it, or it is never discussed. There is further reference to the “rule-set” in l200 with no further explanation, simply a listing of the features (and again no explanation why these features were chosen). There is subsequently mention of a “hierarchical approach, where hydrological features have priority…” (l215) but again no explanation how this priority is incorporated.
At the moment, I can only reject this paper, it really needs a much clearer explanation of what was actually done.
Many more comments on the attached pdf.
-
AC1: 'Reply on RC1', Alexander Herr, 21 Apr 2023
Publisher’s note: a supplement was added to this comment on 8 May 2023.
Dear Willem and Patricia,
Thank you Willem for taking the time to review this paper. In the following I provide a brief response to some of Willem’s main assertions and for Patricia’s considerations.
Willems first assertion is that “the main framework that was being described was a GIS overlay, but how the different features were weighted is unclear, or I must have missed this”.
I take the point that the document gives no description of how the different features are weighted. This is because we did not apply any weighting. The approach we took is one of prioritisation and this we describe in in lines 186 to 187: "Existing spatial data for each region forms the basis for categorising the landscape features using a rule-set based on attribute features within the spatial datasets" and lines 195-196 state: "Our approach uses a defined rule-set and priorities, which we apply to regionally available data sets to achieve a landscape classification for each of our regions. " We then detail the rule sets in lines 200 onwards in the remainder of the methodology section.
The resulting rule set we present in the Results as figures 2,4,5 for each of the regions and we outline the hierarchical approach in lines 214 – 221. This paragraph also provides the reason for the prioritisation, that is identifying hydrological connectivity (and causality) between water and landscape elements.
A second contention Willem lists is that the publication “refers to other papers in the methods and the result without explaining how exactly this was integrated in the current paper and was used in the framework. As a result, it is not clear how the results were actually derived. From a reproducibility perspective, I think it would be hard to replicate the results“.
I suggest that the latter statement is based on a false premise because (1) the citations in the Methods are supporting information, for example, to point to further details regarding the study areas and in lines 198-199 to contrast our method to what others have done. Tables 1-3 provide example datasets for the spatial data. These citations are not needed to understand the methods, but they are important for replicating the results, as they contain the attributes we use for prioritisation. (2) Citations in the Results are datasets only, which are also required for reproducing the work. Indeed, transparency and repeatability are one of the major features of the Bioregional Assessments, which developed an extensive datastore for just these reasons (see https://www.bioregionalassessments.gov.au/bioregional-assessment-program, and e.g. https://www.bioregionalassessments.gov.au/data/3-4-impact-and-risk-analysis-namoi-subregion)
Willem also asserts in a third contention that another “problem is the references to a ’rule-set‘. I presumed this was going to be discussed later in the paper, but either I have totally missed it, or it is never discussed”.
I suggest that Willem missed this, as it is clearly outlined in Figures 2,4,5, which summarise the resulting rule sets. Details for the rule sets are in the methods and specifically in lines 200-226; and the reasoning for prioritisations, which we state in lines 214-215: “For our work, where hydrological connectivity is the main reason for developing a new classification, the most important characteristics are the hydrological features”.
While I do not agree with Willem’s statement “In my opinion the methodology is not well described and it is unclear how decisions were made for different classes”, I take from this that there is need to better clarify in-text that the work is about a GIS overlay using prioritisation to obtain a hydrological landscape classification, and this classification then forms the crucial part for a subsequent risk assessment of hazards from coal resource development that lead to water related impacts on ecological entities.
I would assert that the paper already provides a clear outline of methods, results and application of the approach. While this is just my opinion, I would like to add that two previous reviewers (from a strictly hydrological focussed journal that suggested to find a journal with a broader water-landscape focus) stated: (R1) "In general, it is well written and clear structured, the reasons why it was developed were given and three aims were defined: characterize the system at regional level, develop the system and ensure that the new developed system is able to fulfil its purpose (aiding in formulating conceptual models and patterns of water dependency across the landscape)", and (R2) "The paper is clear about what has been done and why, and the outcomes".
What I take from Willems’ comment is that for understanding the paper, readers may require immersive attention and focus on the details within the methods. Only then is it possible to make sense of the Results and to understand the main thrust of the Discussion, which asserts that for determining the impact of coal resource developments on ecological elements spatially, our landscape classification was crucial. It formed one of the main elements (besides hydrology) that experts used to determine risk to a region’s ecology. I also take from this that, a high-level overview guide in the introduction, maybe with a visual summary, may aid in the process of gaining a detailed understanding.
-
AC1: 'Reply on RC1', Alexander Herr, 21 Apr 2023
-
RC2: 'Comment on hess-2022-408', Anonymous Referee #2, 24 Apr 2023
I found the paper to be confusing and difficult to read. This was probably due in part to the nature of the work which spans hydrological and ecological modelling, but it was also because of the style of writing, which is wordy and vague. These are a few examples:
line 17, and 109: "at the landscape level" What does this mean?
line 105: "It places the landscape classification within a common framework" I don't understand this. Common with what?
line 108: "conceptually describe" How is this different from "describe"?
line 217 “spatially complete”. I don’t understand this.
line 347: "conceptual understanding". How is different from "understanding"?
line 371: “to conceptualise and prioritise” Could be replaced with “of”.
line 380: “digraphs” I assume that you are referring to directed graphs, but a definition would help.
The manuscript would greatly benefit from editing to tighten up the writing.
I don't think that what the authors have proposed is wrong, but I'm not sure that it is necessarily very new. I’m also concerned that it’s not really tested.
The authors state (correctly, in my opinion): "In summary, landscape classification is a way of dividing a landscape into components where the characteristics within the components are more similar than the characteristics between the components." How have they established this for their classification system which appears to be ad-hoc? They state that their method differs from those that “apply statistical dimensionality reduction and classifications such as proximity analysis”. So how can we assess their methodology?
There is virtually no discussion of scale in the paper, which is concerning, given the importance to hydrological processes. I appreciate that the data sets that the authors have used have many differing scales, but it was not clear from the writing what the authors’ scale objectives were. They refer to the “landscape level” and a “regional level landscape” (line 98), without explaining what these mean.
What are the scales of the landscape groups plotted in Fig 3, and the landscape classes plotted in Figs 6a and 6b? How will the scales of their groups and classes affect the hydrological models to be developed?
In a related issue, there is very little discussion of the hydrological processes that will be modelled, other than their association with landscape units, which is also concerning. It would be interesting to understand the effects of the classification system on the development of the quantitative models. For example, it’s interesting not to see vegetation used as a classifier for the stream uplands in Figure 2. I suppose that the authors are using a single vegetation type for these four classes. I would also assume that the resulting hydrological model would use the same parameters for the topography and vegetation for qunatitative hydrological models of all landscape units in these classes, is that correct? If so, would it be hydrologically valid?
The authors state (line 257):
“The purpose of the landscape groups was to combine non-water dependent landscape classes and relate water dependent landscape classes to region specific aspects of their water dependency, which enabled conceptualisation of the landscape for modelling purposes.”
Again, this is vague. What type of modelling are they referring to? In Figure 2, the “Non-floodplain or upland riverine” group is comprised of 8 different classes, which have very different vegetation types. Are the authors proposing to use their groups as a basis for their quantitative model, despite their having such great variation in the hydrological process parameters within each group? Wouldn't the uee of these groups in any form of modelling violate the requirement that "the characteristics within the components are more similar than the characteristics between the components"?
Most importantly, there does not appear to be any attempt to validate the general approach. The authors provide examples of the use of their classification system and state that it "works" (line 471), but how do we know this? How would the approach work in a region with very different topography and/or hydrological processes, such as an alpine region, where local slope, aspect and elevation will likely dominate the hydrology, and where the hydrological processes (snow accumulation and melt, glaciers) will be very different?
Without an attempt to test the methodology, the manuscript appears to be more of a technical note than a scientific paper.
Citation: https://doi.org/10.5194/hess-2022-408-RC2 -
AC2: 'Reply on RC2', Alexander Herr, 05 May 2023
Publisher’s note: a supplement was added to this comment on 8 May 2023.
Dear Patricia,
I am responding here to the main assertions from Reviewer 2 (R2). These include
(1) Scale: “There is virtually no discussion of scale in the paper, which is concerning, given the importance to hydrological processes”
(2) Validation: “there does not appear to be any attempt to validate the general approach”
(3) Novelty: “I don't think that what the authors have proposed is wrong, but I'm not sure that it is necessarily very new”
1. Scale
The concept of scale has meanings and measurements that are different for different disciplines. Here we are working at an ecologically relevant landscape scale because the focus is on water mitigated impacts of coal resource developments (CRDs) affecting ecological entities. Ecological systems are complex and work at a range of scales within regions/landscapes, and they exhibit interactions and feedbacks that work across scales. For our purpose, the relevant scales are associated with eco-hydrological linkages (and associated causality) between ecological components of interest and the hydrological changes from CRD (and for that matter any other development that changes ground and surface water availability). Here, we define our ecological landscape components in a way that combines homogenous components (our landscape classes) within a heterogenous landscape. Ecologically relevant scales are varied, and it would therefore be difficult when attempting to assign a specific scale (e.g. 1:1,000,000 or 10000 m pixel size) because ecological complexity implies multi-scale processes and working at single scales only would create exposure to the modifiable area unit problem and could lead to ecological fallacies (Newman et al 2019, Wu and Loucks 1995, Openshaw 1982).
Using a narrowly defined scale for our purpose would also violate the stationary distribution assumption (i.e. the generating function for a stochastic process is constant) of many statistically based models. Hence, it is important to provide the flexibility for experts to develop their own conceptual understanding when developing their qualitative (and subsequent quantitative) impact models. Additionally, the use of scale in hydrology is not as clearly articulated as one might perceive as there are different views on the usefulness to define scale in hydrology. The reasons for this include the lack of agreement on scale terminology, which is however, not a large impediment for communication between hydrologists. In addition, our work identifies the regional scale as the size of our study regions, ranging from “… 36,000 km2 to 600,000 km2 in size…” (line 125), which is what most hydrologists would consider the “regional” scale range (Gleeson and Paszkowski 2014).
The outcome of the impact assessment, however, is directly related to the hydrological assessment scale. Hence, figure 6a & b in the manuscript provide a pixelated view of the potentially impacted area, which is a direct result of the hydrological assessment units (1 km2). In this context, the reviewer asks: “How will the scales of their groups and classes affect the hydrological models to be developed”. I think there is a misunderstanding here. The hydrological models are not related to the development of the landscape classification. In line 348 and 349 we state that our approach “… provides an essential framework for experts to understand and conceptualise how modelled future hydrological changes from coal resource developments link to potential ecological changes at the landscape level…”, which implies existing hydrological models.
One of the main reasons for writing this paper was show a way of integrating ecology, hydrology and applied Bayesian statistics for the purpose of environmental impact assessments. This demands the extension of narrowly defined disciplinary terminology (in this case scale) to a broader concept and so enabling disciplines to move beyond their narrow applications to an interdisciplinary space. Inter- and transdisciplinary analysis are required in modern natural resource management, where complexity and wicked problems become the norm.
While it would be too extensive in this paper to discuss scale and its use in different disciplines, I take from this comment that there is need to clarify and define that the purpose of this paper is to work at the scale that is relevant for ecological impacts of water changes from CRD using an expert assessment approach.
2. Validation
I do believe that the paper shows a “validation” of the method. Our intent was to develop a fit-for-purpose landscape classification that incorporates causal linkage between hydrological and ecological landscape components, and that experts can use it to develop qualitative and quantitative impact models. We show the outcome of these models in three different regions and with it, we have a validation of the approach by implication that the experts used the landscape classification for their purpose.
If I understand correctly, the reviewer asks for inductive null hypothesis testing and hypothesis falsification in the minimalistic, reductionist sense when relying on Popper’s hypothesis falsifying approach. Popper saw science as an evolutionary process, not a simple matter of falsification (Mulkay and Gilbert 1981, Popper 1972, 1981). It is beyond the scope of this review process to discuss this further here, but in summary, limiting modelling to falsification science is a misappropriation of Popper’s philosophy. (For a more detailed treatment of the philosophy in hypothesis testing, modelling and modern statistical analysis, I recommend Chapter 1 of McElrath (2016).)
The reviewer also asks “How do we know this”. In this paper we have shown that the classification works for three physiognomic and ecologically very different regions. While we did not apply this to “an alpine region, where local slope, aspect and elevation will likely dominate the hydrology”, we are confident that our formal approach will work. Firstly, because the landscape classification only relies on hydrological analysis indirectly. It uses spatial data related to hydrological landscape features to develop landscape classes, so the datasets have inherent hydrological elements. Secondly, our formalised approach uses a prioritisation schematic that allows the user to “build” a landscape classification from existing data with a focus on eco-hydrological connectivity and causality. These are the essential elements of the classification that experts require, so they can develop impact models that relate water changes to ecological entities in the landscape.
3. Novelty
I agree that what we develop is not necessarily very new. What is new is that we make use of existing information contained in datasets instead of starting from scratch, and we interrogate these datasets for essential elements that are relevant for our purpose. We already outlined in the introduction that there is a myriad of landscape classifications and even some ecohydrological classifications, but these are either for a different purpose and/or too narrow in scope for environmental impact assessment at a regional scale.
The novelty in our approach is that:
- We develop a novel prioritisation scheme for existing spatial data, which means we do not need to “re-invent the wheel” by developing new spatial data and applying spatial modelling to, for example, remote sensed data.
- Our scheme is fit-for-purpose for expert impact assessment, providing causal linkage of hydrology and ecology at landscape scale. We provide evidence for this in the discussion.
- This is the first time, that an ecological impact assessment has shown regional coverage and direct causality between cumulative changes in water associated with (including future) CRDs. Our landscape classification is the cornerstone that enabled this assessment.
References
Gleeson, T. and Paszkowski, D. (2014) Perceptions of scale in hydrology: what do you mean by regional scale?, Hydrological Sciences Journal, 59, 99-107, 10.1080/02626667.2013.797581, 2014.
McElreath, R. (2016) Statistical Rethinking: A Bayesian Course with Examples in R and Stan (1st ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9781315372495
Mulkay, M. and Gilbert, G. N. (1981) Putting philosophy to work: Karl Popper’s influence on scientific practice. Philosophy of the Social Sciences, 11:389–407.
Newman, E. A., Kennedy, M. C., Falk, D. A., and McKenzie, D. (2019) Scaling and Complexity in Landscape Ecology, Frontiers in Ecology and Evolution, 7, https://doi.org/10.3389/fevo.2019.00293
Openshaw, S. (1984) Ecological Fallacies and the Analysis of Areal Census Data. Environment and Planning A: Economy and Space, 16(1), 17–31. https://doi.org/10.1068/a160017
Popper, K R. (1972) Objective Knowledge: An Evolutionary Approach. Oxford, England: Oxford, Clarendon Press.
Popper, K.R. (1995) The Myth of the Framework. In Defence of Science and Rationality. London and New York: Routledge. ISBN 0-415-13555-9
Wu, J. and Loucks, O. (1995). From Balance of Nature to Hierarchical Patch Dynamics: A Paradigm Shift in Ecology. The Quarterly Review of Biology 1995 70:4, 439-466. https://www.journals.uchicago.edu/doi/10.1086/419172/.
-
AC2: 'Reply on RC2', Alexander Herr, 05 May 2023
Status: closed
-
RC1: 'Comment on hess-2022-408', Willem Vervoort, 09 Apr 2023
This manuscript describes a classification framework to classify landscapes into different categories which can subsequently be used for conceptual modelling.
I am afraid that after reading the manuscript I am still unsure how the framework actually works. In my opinion the methodology is not well described and it is unclear how decisions were made for different classes. At some point it seemed that the main framework that was being described was a GIS overlay, but how the different features were weighted is unclear, or I must have missed this. The manuscript regularly refers to other papers in the methods and the result without explaining how exactly this was integrated in the current paper and was used in the framework. As a result, it is not clear how the results were actually derived. From a reproducibility perspective, I think it would be hard to replicate the results.
I am left wondering how decisions were made about different classes, where these simply in the original data, or were those classes decided on in this study. It almost seems like this study was the culmination of a series of other studies, but these studies (while referenced) are not discussed in the paper. I think there needs to be a much clearer methodology and workflow to be able to reproduce the results and to make the paper easier to read and understand.
Here is an example of some of the unclear discussion (l408 and further):
“The modelling of risk to ecosystems at regional scale focuses on recognising which parts of the region are potentially impacted and which parts are unlikely to experience harm. Using our landscape classification as a crucial input, the modelling delineated impacted areas within each region, based on a zone of potential hydrological change.”
From this, I fail to understand how the classification was aa “crucial input” and how this assisted in delineating the impacted areas. There is an earlier reference to Hosack et al., is this the paper that describes the modelling? It would still be useful to help the reader understand what the modelling was (Summarising the earlier study) and highlighting how it was shown that the classification was a “crucial input”.
Another example from the start of the methodology, where essentially the overall approach is summarised (l184..):
“The purpose of this ecohydrological landscape classification is to characterise the landscape based on patterns in land use, ecology, geomorphology and hydrology, and from these, develop landscape classes of water-dependent, remnant and human-modified features. Existing spatial data for each region forms the basis for categorising the landscape features using a rule-set based on attribute features within the spatial datasets.”
The first problem I have is that why landscape classes of “water-dependent, remnant and human-modified features” are chosen doesn’t seem to be explained. I can see that this is a useful classification, but at least some rational for the choice (and why no other classes) should be presented. The second problem is the references to a “rule-set”. I presumed this was going to be discussed later in the paper, but either I have totally missed it, or it is never discussed. There is further reference to the “rule-set” in l200 with no further explanation, simply a listing of the features (and again no explanation why these features were chosen). There is subsequently mention of a “hierarchical approach, where hydrological features have priority…” (l215) but again no explanation how this priority is incorporated.
At the moment, I can only reject this paper, it really needs a much clearer explanation of what was actually done.
Many more comments on the attached pdf.
-
AC1: 'Reply on RC1', Alexander Herr, 21 Apr 2023
Publisher’s note: a supplement was added to this comment on 8 May 2023.
Dear Willem and Patricia,
Thank you Willem for taking the time to review this paper. In the following I provide a brief response to some of Willem’s main assertions and for Patricia’s considerations.
Willems first assertion is that “the main framework that was being described was a GIS overlay, but how the different features were weighted is unclear, or I must have missed this”.
I take the point that the document gives no description of how the different features are weighted. This is because we did not apply any weighting. The approach we took is one of prioritisation and this we describe in in lines 186 to 187: "Existing spatial data for each region forms the basis for categorising the landscape features using a rule-set based on attribute features within the spatial datasets" and lines 195-196 state: "Our approach uses a defined rule-set and priorities, which we apply to regionally available data sets to achieve a landscape classification for each of our regions. " We then detail the rule sets in lines 200 onwards in the remainder of the methodology section.
The resulting rule set we present in the Results as figures 2,4,5 for each of the regions and we outline the hierarchical approach in lines 214 – 221. This paragraph also provides the reason for the prioritisation, that is identifying hydrological connectivity (and causality) between water and landscape elements.
A second contention Willem lists is that the publication “refers to other papers in the methods and the result without explaining how exactly this was integrated in the current paper and was used in the framework. As a result, it is not clear how the results were actually derived. From a reproducibility perspective, I think it would be hard to replicate the results“.
I suggest that the latter statement is based on a false premise because (1) the citations in the Methods are supporting information, for example, to point to further details regarding the study areas and in lines 198-199 to contrast our method to what others have done. Tables 1-3 provide example datasets for the spatial data. These citations are not needed to understand the methods, but they are important for replicating the results, as they contain the attributes we use for prioritisation. (2) Citations in the Results are datasets only, which are also required for reproducing the work. Indeed, transparency and repeatability are one of the major features of the Bioregional Assessments, which developed an extensive datastore for just these reasons (see https://www.bioregionalassessments.gov.au/bioregional-assessment-program, and e.g. https://www.bioregionalassessments.gov.au/data/3-4-impact-and-risk-analysis-namoi-subregion)
Willem also asserts in a third contention that another “problem is the references to a ’rule-set‘. I presumed this was going to be discussed later in the paper, but either I have totally missed it, or it is never discussed”.
I suggest that Willem missed this, as it is clearly outlined in Figures 2,4,5, which summarise the resulting rule sets. Details for the rule sets are in the methods and specifically in lines 200-226; and the reasoning for prioritisations, which we state in lines 214-215: “For our work, where hydrological connectivity is the main reason for developing a new classification, the most important characteristics are the hydrological features”.
While I do not agree with Willem’s statement “In my opinion the methodology is not well described and it is unclear how decisions were made for different classes”, I take from this that there is need to better clarify in-text that the work is about a GIS overlay using prioritisation to obtain a hydrological landscape classification, and this classification then forms the crucial part for a subsequent risk assessment of hazards from coal resource development that lead to water related impacts on ecological entities.
I would assert that the paper already provides a clear outline of methods, results and application of the approach. While this is just my opinion, I would like to add that two previous reviewers (from a strictly hydrological focussed journal that suggested to find a journal with a broader water-landscape focus) stated: (R1) "In general, it is well written and clear structured, the reasons why it was developed were given and three aims were defined: characterize the system at regional level, develop the system and ensure that the new developed system is able to fulfil its purpose (aiding in formulating conceptual models and patterns of water dependency across the landscape)", and (R2) "The paper is clear about what has been done and why, and the outcomes".
What I take from Willems’ comment is that for understanding the paper, readers may require immersive attention and focus on the details within the methods. Only then is it possible to make sense of the Results and to understand the main thrust of the Discussion, which asserts that for determining the impact of coal resource developments on ecological elements spatially, our landscape classification was crucial. It formed one of the main elements (besides hydrology) that experts used to determine risk to a region’s ecology. I also take from this that, a high-level overview guide in the introduction, maybe with a visual summary, may aid in the process of gaining a detailed understanding.
-
AC1: 'Reply on RC1', Alexander Herr, 21 Apr 2023
-
RC2: 'Comment on hess-2022-408', Anonymous Referee #2, 24 Apr 2023
I found the paper to be confusing and difficult to read. This was probably due in part to the nature of the work which spans hydrological and ecological modelling, but it was also because of the style of writing, which is wordy and vague. These are a few examples:
line 17, and 109: "at the landscape level" What does this mean?
line 105: "It places the landscape classification within a common framework" I don't understand this. Common with what?
line 108: "conceptually describe" How is this different from "describe"?
line 217 “spatially complete”. I don’t understand this.
line 347: "conceptual understanding". How is different from "understanding"?
line 371: “to conceptualise and prioritise” Could be replaced with “of”.
line 380: “digraphs” I assume that you are referring to directed graphs, but a definition would help.
The manuscript would greatly benefit from editing to tighten up the writing.
I don't think that what the authors have proposed is wrong, but I'm not sure that it is necessarily very new. I’m also concerned that it’s not really tested.
The authors state (correctly, in my opinion): "In summary, landscape classification is a way of dividing a landscape into components where the characteristics within the components are more similar than the characteristics between the components." How have they established this for their classification system which appears to be ad-hoc? They state that their method differs from those that “apply statistical dimensionality reduction and classifications such as proximity analysis”. So how can we assess their methodology?
There is virtually no discussion of scale in the paper, which is concerning, given the importance to hydrological processes. I appreciate that the data sets that the authors have used have many differing scales, but it was not clear from the writing what the authors’ scale objectives were. They refer to the “landscape level” and a “regional level landscape” (line 98), without explaining what these mean.
What are the scales of the landscape groups plotted in Fig 3, and the landscape classes plotted in Figs 6a and 6b? How will the scales of their groups and classes affect the hydrological models to be developed?
In a related issue, there is very little discussion of the hydrological processes that will be modelled, other than their association with landscape units, which is also concerning. It would be interesting to understand the effects of the classification system on the development of the quantitative models. For example, it’s interesting not to see vegetation used as a classifier for the stream uplands in Figure 2. I suppose that the authors are using a single vegetation type for these four classes. I would also assume that the resulting hydrological model would use the same parameters for the topography and vegetation for qunatitative hydrological models of all landscape units in these classes, is that correct? If so, would it be hydrologically valid?
The authors state (line 257):
“The purpose of the landscape groups was to combine non-water dependent landscape classes and relate water dependent landscape classes to region specific aspects of their water dependency, which enabled conceptualisation of the landscape for modelling purposes.”
Again, this is vague. What type of modelling are they referring to? In Figure 2, the “Non-floodplain or upland riverine” group is comprised of 8 different classes, which have very different vegetation types. Are the authors proposing to use their groups as a basis for their quantitative model, despite their having such great variation in the hydrological process parameters within each group? Wouldn't the uee of these groups in any form of modelling violate the requirement that "the characteristics within the components are more similar than the characteristics between the components"?
Most importantly, there does not appear to be any attempt to validate the general approach. The authors provide examples of the use of their classification system and state that it "works" (line 471), but how do we know this? How would the approach work in a region with very different topography and/or hydrological processes, such as an alpine region, where local slope, aspect and elevation will likely dominate the hydrology, and where the hydrological processes (snow accumulation and melt, glaciers) will be very different?
Without an attempt to test the methodology, the manuscript appears to be more of a technical note than a scientific paper.
Citation: https://doi.org/10.5194/hess-2022-408-RC2 -
AC2: 'Reply on RC2', Alexander Herr, 05 May 2023
Publisher’s note: a supplement was added to this comment on 8 May 2023.
Dear Patricia,
I am responding here to the main assertions from Reviewer 2 (R2). These include
(1) Scale: “There is virtually no discussion of scale in the paper, which is concerning, given the importance to hydrological processes”
(2) Validation: “there does not appear to be any attempt to validate the general approach”
(3) Novelty: “I don't think that what the authors have proposed is wrong, but I'm not sure that it is necessarily very new”
1. Scale
The concept of scale has meanings and measurements that are different for different disciplines. Here we are working at an ecologically relevant landscape scale because the focus is on water mitigated impacts of coal resource developments (CRDs) affecting ecological entities. Ecological systems are complex and work at a range of scales within regions/landscapes, and they exhibit interactions and feedbacks that work across scales. For our purpose, the relevant scales are associated with eco-hydrological linkages (and associated causality) between ecological components of interest and the hydrological changes from CRD (and for that matter any other development that changes ground and surface water availability). Here, we define our ecological landscape components in a way that combines homogenous components (our landscape classes) within a heterogenous landscape. Ecologically relevant scales are varied, and it would therefore be difficult when attempting to assign a specific scale (e.g. 1:1,000,000 or 10000 m pixel size) because ecological complexity implies multi-scale processes and working at single scales only would create exposure to the modifiable area unit problem and could lead to ecological fallacies (Newman et al 2019, Wu and Loucks 1995, Openshaw 1982).
Using a narrowly defined scale for our purpose would also violate the stationary distribution assumption (i.e. the generating function for a stochastic process is constant) of many statistically based models. Hence, it is important to provide the flexibility for experts to develop their own conceptual understanding when developing their qualitative (and subsequent quantitative) impact models. Additionally, the use of scale in hydrology is not as clearly articulated as one might perceive as there are different views on the usefulness to define scale in hydrology. The reasons for this include the lack of agreement on scale terminology, which is however, not a large impediment for communication between hydrologists. In addition, our work identifies the regional scale as the size of our study regions, ranging from “… 36,000 km2 to 600,000 km2 in size…” (line 125), which is what most hydrologists would consider the “regional” scale range (Gleeson and Paszkowski 2014).
The outcome of the impact assessment, however, is directly related to the hydrological assessment scale. Hence, figure 6a & b in the manuscript provide a pixelated view of the potentially impacted area, which is a direct result of the hydrological assessment units (1 km2). In this context, the reviewer asks: “How will the scales of their groups and classes affect the hydrological models to be developed”. I think there is a misunderstanding here. The hydrological models are not related to the development of the landscape classification. In line 348 and 349 we state that our approach “… provides an essential framework for experts to understand and conceptualise how modelled future hydrological changes from coal resource developments link to potential ecological changes at the landscape level…”, which implies existing hydrological models.
One of the main reasons for writing this paper was show a way of integrating ecology, hydrology and applied Bayesian statistics for the purpose of environmental impact assessments. This demands the extension of narrowly defined disciplinary terminology (in this case scale) to a broader concept and so enabling disciplines to move beyond their narrow applications to an interdisciplinary space. Inter- and transdisciplinary analysis are required in modern natural resource management, where complexity and wicked problems become the norm.
While it would be too extensive in this paper to discuss scale and its use in different disciplines, I take from this comment that there is need to clarify and define that the purpose of this paper is to work at the scale that is relevant for ecological impacts of water changes from CRD using an expert assessment approach.
2. Validation
I do believe that the paper shows a “validation” of the method. Our intent was to develop a fit-for-purpose landscape classification that incorporates causal linkage between hydrological and ecological landscape components, and that experts can use it to develop qualitative and quantitative impact models. We show the outcome of these models in three different regions and with it, we have a validation of the approach by implication that the experts used the landscape classification for their purpose.
If I understand correctly, the reviewer asks for inductive null hypothesis testing and hypothesis falsification in the minimalistic, reductionist sense when relying on Popper’s hypothesis falsifying approach. Popper saw science as an evolutionary process, not a simple matter of falsification (Mulkay and Gilbert 1981, Popper 1972, 1981). It is beyond the scope of this review process to discuss this further here, but in summary, limiting modelling to falsification science is a misappropriation of Popper’s philosophy. (For a more detailed treatment of the philosophy in hypothesis testing, modelling and modern statistical analysis, I recommend Chapter 1 of McElrath (2016).)
The reviewer also asks “How do we know this”. In this paper we have shown that the classification works for three physiognomic and ecologically very different regions. While we did not apply this to “an alpine region, where local slope, aspect and elevation will likely dominate the hydrology”, we are confident that our formal approach will work. Firstly, because the landscape classification only relies on hydrological analysis indirectly. It uses spatial data related to hydrological landscape features to develop landscape classes, so the datasets have inherent hydrological elements. Secondly, our formalised approach uses a prioritisation schematic that allows the user to “build” a landscape classification from existing data with a focus on eco-hydrological connectivity and causality. These are the essential elements of the classification that experts require, so they can develop impact models that relate water changes to ecological entities in the landscape.
3. Novelty
I agree that what we develop is not necessarily very new. What is new is that we make use of existing information contained in datasets instead of starting from scratch, and we interrogate these datasets for essential elements that are relevant for our purpose. We already outlined in the introduction that there is a myriad of landscape classifications and even some ecohydrological classifications, but these are either for a different purpose and/or too narrow in scope for environmental impact assessment at a regional scale.
The novelty in our approach is that:
- We develop a novel prioritisation scheme for existing spatial data, which means we do not need to “re-invent the wheel” by developing new spatial data and applying spatial modelling to, for example, remote sensed data.
- Our scheme is fit-for-purpose for expert impact assessment, providing causal linkage of hydrology and ecology at landscape scale. We provide evidence for this in the discussion.
- This is the first time, that an ecological impact assessment has shown regional coverage and direct causality between cumulative changes in water associated with (including future) CRDs. Our landscape classification is the cornerstone that enabled this assessment.
References
Gleeson, T. and Paszkowski, D. (2014) Perceptions of scale in hydrology: what do you mean by regional scale?, Hydrological Sciences Journal, 59, 99-107, 10.1080/02626667.2013.797581, 2014.
McElreath, R. (2016) Statistical Rethinking: A Bayesian Course with Examples in R and Stan (1st ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9781315372495
Mulkay, M. and Gilbert, G. N. (1981) Putting philosophy to work: Karl Popper’s influence on scientific practice. Philosophy of the Social Sciences, 11:389–407.
Newman, E. A., Kennedy, M. C., Falk, D. A., and McKenzie, D. (2019) Scaling and Complexity in Landscape Ecology, Frontiers in Ecology and Evolution, 7, https://doi.org/10.3389/fevo.2019.00293
Openshaw, S. (1984) Ecological Fallacies and the Analysis of Areal Census Data. Environment and Planning A: Economy and Space, 16(1), 17–31. https://doi.org/10.1068/a160017
Popper, K R. (1972) Objective Knowledge: An Evolutionary Approach. Oxford, England: Oxford, Clarendon Press.
Popper, K.R. (1995) The Myth of the Framework. In Defence of Science and Rationality. London and New York: Routledge. ISBN 0-415-13555-9
Wu, J. and Loucks, O. (1995). From Balance of Nature to Hierarchical Patch Dynamics: A Paradigm Shift in Ecology. The Quarterly Review of Biology 1995 70:4, 439-466. https://www.journals.uchicago.edu/doi/10.1086/419172/.
-
AC2: 'Reply on RC2', Alexander Herr, 05 May 2023
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 522 | 182 | 18 | 722 | 10 | 12 |
- HTML: 522
- PDF: 182
- XML: 18
- Total: 722
- BibTeX: 10
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1