the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Sep 2023
Representing socio-economic factors in the INFERNO global fire model using the Human Development Index
Abstract. Humans can act as fire starters or suppressors, changing fire regimes by increasing the number of ignitions, changing their timing, and altering fuel structure and abundance, which can be considered a human–environmental coupling. Considering the human influences on fire activity, representing socio-economic impacts on fires in global fire models is crucial to underpin the confidence in these modelling frameworks. In this work we implement a socio-economic factor in the fire ignition and suppression parametrisation in INFERNO based on a Human Development Index (HDI). HDI captures human development's income, health, and education dimensions leading to a representation where if there is more effort to improve human development, the population also invests in higher fire suppression. Including this representation of socio-economic factors in INFERNO reduces the annual mean burnt area (between 1997–2016) positive biases found in Temperate North America, Central America, Europe and Southern Hemisphere South America, by more than 100 % without statistically significant impact to other areas. In addition, it improves the representation of the burnt area trends, especially in Africa. Central Asia and Australia where observations show negative trends. Including socio-economic impacts on fire based on HDI in INFERNO provides a simple and linear representation of these effects on fire ignition and suppression, leading to an improvement of the model performance, especially in developed regions, These impacts are especially relevant to understand future climate regimes and inform policymakers on effects of fire policy in a changing climate.
- Preprint
(12466 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on bg-2023-136', Anonymous Referee #1, 27 Nov 2023
This manuscript aims to account for the impacts of human development on fire activity in the INFERNO model through an implementation of linear relationships between the number of human-caused ignitions to the Human Development Index (HDI), and the amount of fire suppression to the HDI. The manner in which these relationships are implemented into the model leave important questions unanswered (a general overview is included here, with a more detailed discussion attached):
First: Does the HDI reflect the human impact on the number of ignitions and on the number of fires that are suppressed?
- The authors provide no analysis and cite only a single previous paper in which the HDI was used as a predictor for fire activity. In this paper, Chuvieco et al. (2021), the HDI is used in a fundamentally different way, not in a relationship to the number of fires, as it is in this manuscript.
Second: If such relationships exist, what form do they take?
- The functional forms of the relationships to the HDI that the authors implement are not supported with sufficient evidence, apart from general conceptual arguments that more human development might reduce the number of fires. This neglects fire management practices in some developed countries where fire exclusion, i.e. the suppression of every observed fire as soon as possible, is no longer practiced, in favor of practices that include prescribed burns.
- Further, the relationships the authors include in their model impact the number of fires but do not reflect any relationship between fire management and reducing fire size, even though fighting fires that are actively spreading is a significant role of fire management.
- Finally, the relationships to the HDI that the authors impose involve simply multiplying existing equations in the model by 1-HDI. There is no evidence provided that the relationships between the number of fires and HDI should be linear or, if the relationships are linear, that these are the appropriate coefficients for this relationship.
The authors claim that their implementation results in an improved version of the INFERNO model, but the evidence for this is ambiguous at best, for the following reasons:
- The relationships between the HDI and the number of fires are not validated. Rather, the validation is only performed on burned area, to which the number of fires contributes, but in combination with fire size. Because of this it is not possible to gauge the accuracy of their parametrizations explicitly.
- In terms of bias in the burned area, in a comparison between model versions after the implementation of the HDI and before, the new model version, that includes the HDI implementation, has a greater bias on a global scale as well as in 8 of the 14 regions that the authors analyze (this is shown in Table 2 in this manuscript)
- In terms of the burned area trends, the implementation of the HDI does show an improvement on the global scale, and in 9 of the 14 regions. However, there are no R2 values or confidence intervals for the trends provided, or confidence levels for assertions that one trend is more accurate than another. Without these, it isn’t possible to say whether the trends are significant or with what certainty the errors in the trends have been reduced.
- In terms of standard deviation, the model version that includes the HDI shows an improvement on the global scale, but only in 5 of the 14 regional scales, with the previous model version performing better in the 9 others. This is a particular problem as the authors state that their aim is “to improve the regional representation of human–environmental coupling for applications at large spatial scales”
- Overall, a majority of the metrics that the authors provide show worse performance at the regional scale after implementing the HDI than before.
A specific section of the paper that is worth highlighting in these general comments is the abstract, which is highly misleading, and gives a false impression to a casual reader of this paper who would rely on it. It mentions only regions in which the model was improved by implementation of the HDI, without mentioning that the global bias in burned area was increased and that many regions experience worse model performance. Two statements in the abstract in particular are false.
- The first is that the bias in burned area is reduced in 4 regions in the model “without statistically significant impact to 10 other areas.” The statistical significance of these biases is not discussed in the manuscript at any point, the global bias is increased by implementing the HDI, and some regions show quite large increases in their bias. Australia, in particular, shows an increase in its relative bias from -21% to -82%. It is unclear what statistical test would define this as insignificant.
- The second statement is that the new model version “improves the representation of the burnt area trends, especially in Africa, Central Asia and Australia.” One part of Africa shows an improvement, while the other shows a worse performance. Combined, the trend over Africa in total is worse in the updated model version (details are provided in the specific comments). This statement is therefore highly misleading at best.
For this manuscript to be publishable, the authors must fundamentally reframe it in a manner that transparently reflects their results. This includes stating clearly that the functions implemented in their model are initial attempts at parametrization that are not based on previous analysis, clearly stating that the results of implementing these parametrizations are mixed, and that further work is required to derive and validate equations that reflect the impact of human development on fire activity at a global scale. This should be reflected in sections where results are summarized and discussed. The current framing overstates the results and is not supported by sufficient evidence.
-
AC2: 'Reply on RC1', João Teixeira, 29 Feb 2024
The authors thank the reviewer for constructive comments on our manuscript. We appreciate your feedback and will address the questions you have raised in the attached detailed discussion. We commit to revise our manuscript to provide more clarity on the implementation of Representing socio-economic factors in the INFERNO global fire model using HDI.
A reply to the main reviewer comments follows below, with original reviewer comment presented in bold. A file with detailed comments and respective replies is attached.
This manuscript aims to account for the impacts of human development on fire activity in the INFERNO model through an implementation of linear relationships between the number of human-caused ignitions to the Human Development Index (HDI), and the amount of fire suppression to the HDI. The manner in which these relationships are implemented into the model leave important questions unanswered (a general overview is included here, with a more detailed discussion attached):
First: Does the HDI reflect the human impact on the number of ignitions and on the number of fires that are suppressed?
- The authors provide no analysis and cite only a single previous paper in which the HDI was used as a predictor for fire activity. In this paper, Chuvieco et al. (2021), the HDI is used in a fundamentally different way, not in a relationship to the number of fires, as it is in this manuscript.
Regarding the points the reviewer has made regarding HDI reflect the human impact on both the number of ignitions and on the number of fires that are suppressed. To represent the socio-economic factors impacting fire ignition and suppression, we include a Human Development Index (HDI) term (1-HDI) in our human ignition and suppression Eq. 2 and 3. This is highlighted in lines 103 to 105 of the manuscript.
The authors would like to note that the focus of this work is not on a specific analysis of how HDI is used as a predictor for fire activity. Socio-economic impacts on fire are complex and dependent on may factor that are difficult to represent in Earth System Models (ESM). These factors depend on policies implemented at government level, as well as cultural behaviour which varies widely across the world. In addition, it needs to be highlighted the formulation of Climate and ESM does not allow for representing these details.
The aim of this study is not to represent that complexity achieve that but rather to explore the use of the HDI to represent socio-economic impacts on fires, aiming to improve the regional representation of human–environmental coupling for applications at large spatial scales within an Earth System Model (ESM) context.
To the knowledge of the authors this work presents novel research. At the time this work was developed, this was the first attempt to represent socio-economic impacts on fire using HDI, there are no other works in the literature that could be referenced to support this implementation.
Although the HDI is used in a different way by the work of Chuvieco et al. (2021), in their study the authors show that HDI is one of the main drivers of burnt area interannual variability supporting the use of HDI in the way intended in this study. Paraphrasing the work of Chuvieco et al. (2021)
“Our results indicate that an improved representation of how humans impact fire occurrence, and how this is modulated by socio-economic conditions, as for example indicated by the HDI in our results, might improve the representation of year-to-year variation in fires within fire-vegetation models.”
“Since HDI can also be negatively related to people's dedication to agrarian activities (commonly also to more mechanization), higher HDI implies that population tends to rely less on fire activity for livestock grazing and agriculture than those in less developed areas. In other words, fires are more influenced by human agrarian activity (and therefore less fluctuating) when population is more dependent on agricultural resources (more crop and livestock density) and it has a lower level of development or income (HDI and GDP are highly correlated).”
Second: If such relationships exist, what form do they take?
- The functional forms of the relationships to the HDI that the authors implement are not supported with sufficient evidence, apart from general conceptual arguments that more human development might reduce the number of fires. This neglects fire management practices in some developed countries where fire exclusion, i.e. the suppression of every observed fire as soon as possible, is no longer practiced, in favor of practices that include prescribed burns.
- Further, the relationships the authors include in their model impact the number of fires but do not reflect any relationship between fire management and reducing fire size, even though fighting fires that are actively spreading is a significant role of fire management.
With reference to the functional forms of the relationships between the socio-economic effects on fire ignitions and suppression and HDI, it should be noted that socio-economic impacts on fire are complex and dependent on may factor that are difficult to model, depend on government policies, as well as cultural behaviour. The work by Pandey et al. (2023) is a good example of a study that highlights this complexity, as well as the different fire management policies around the world, showing that despite their differences they all result in a gradual reduction in fire occurrences and burned areas over time. In addition, the formulation of climate/ESM does not allow for representing these details.
The aim of this study is not to represent that complexity but rather to explore the use of the HDI to represent socio-economic impacts on fires, aiming to improve the regional representation of human–environmental coupling for applications at large spatial scales within an Earth System Model (ESM) context. As stated in the work of Mangeon et al. (2016), INFERNO is a simple physically based representation of fire activity aimed at representing fires in the ESM context. The current implementation of the HDI aims to follow that same philosophy.
- Finally, the relationships to the HDI that the authors impose involve simply multiplying existing equations in the model by 1-HDI. There is no evidence provided that the relationships between the number of fires and HDI should be linear or, if the relationships are linear, that these are the appropriate coefficients for this relationship.
Despite being a simple representation while trying to encompass, this approach does align with the few studies found in literature that looked at the impact governmental policies have on prevention of wildfires. For example, the work by Curt and Frejaville (2017) shows that, the wildfire policies implemented in in mediterranean France, resulted in the number of fires has decreased almost linearly since 1975, whereas the burned area changed more abruptly.
The authors claim that their implementation results in an improved version of the INFERNO model, but the evidence for this is ambiguous at best, for the following reasons:
- The relationships between the HDI and the number of fires are not validated. Rather, the validation is only performed on burned area, to which the number of fires contributes, but in combination with fire size. Because of this it is not possible to gauge the accuracy of their parametrizations explicitly.
The discussion and analysis of results of this study is focus on burnt area. INFERNO does not model number of fires. The spatial scales this model is applied to does not allow to model individual fires. In this work, we explore the use of the HDI to represent socio-economic impacts on fires in the Pechony and Shindell (2009) anthropogenic fire ignitions representation, building on the work of Mangeon et al. (2016), to represent socio-economic impacts on fires in INFERNO, and the analysis was based on the results diagnosed by INFERNO – burnt area
- In terms of bias in the burned area, in a comparison between model versions after the implementation of the HDI and before, the new model version, that includes the HDI implementation, has a greater bias on a global scale as well as in 8 of the 14 regions that the authors analyze (this is shown in Table 2 in this manuscript)
The approach presented in this work does improve the results from INFERNO by reducing the large bias produced by the model. JULES-INFERNO has large positive bias at a regional level, for example the bias off regions such as TENA (17.21), CEAM (4.4), SHSA (49.24), EURO (2.23), and MIDE (3.88) total to 76.96 Mha. All these biases are reduced in JULES-INFERNO+HDI. This alone shows that JULES-INFERNO performs well at the global scale as reginal bias compensate each other. Although this is highlighted in section 4 – Conclusions, in lines 330 and 331, the authors agree that improving this sentence would strengthen the manuscript.
“Furthermore, it should be highlighted that although JULES-INFERNO performs well at the global scale, as can be seen when comparing the annual mean burnt area against GFED4s in Table 2, this is due to compensating errors at the regional level”.
Although there is as a negative impact in some of the regions, this is either small (e.g., the difference in the metric is in the order of the decimal place), or the negative impact is understood and discussed in Section 4.2 Model limitations and known issues. It should be highlighted that in the discussion model limitations and known issues we have identified that mechanisms that dominate the fire behaviour of some regions are not represented in INFERNO, the fact that JULES-INFERNO perform better in regions dominated by peat land fires and high interannual variability.
- In terms of the burned area trends, the implementation of the HDI does show an improvement on the global scale, and in 9 of the 14 regions. However, there are no R2 values or confidence intervals for the trends provided, or confidence levels for assertions that one trend is more accurate than another. Without these, it isn’t possible to say whether the trends are significant or with what certainty the errors in the trends have been reduced.
The observed dataset (GFED 4s) shows that out of 14 regions, 4 have positive burnt area trend. From these JULES-INFERNO only presents a positive trend for TENA and SEAS. It
While JULES-INFERNO+HDI tends to enforce decreasing trends, this only happens in 4 regions out of 14, TENA, SHAF, MIDE, and SEAS. For the remaining 10 regions, JULES-INFERNO+HDI presents a similar trend to JULES-INFERNO or even an improved trend when compared to GFED 4s.
- In terms of standard deviation, the model version that includes the HDI shows an improvement on the global scale, but only in 5 of the 14 regional scales, with the previous model version performing better in the 9 others. This is a particular problem as the authors state that their aim is “to improve the regional representation of human–environmental coupling for applications at large spatial scales”
Regarding the impacts this approach has in burnt area variability, the authors have focussed the analysis on the standard deviation. Standard deviation is the average amount of variability in your dataset. It tells you, on average, how far each value lies from the mean. The impacted of the use of socio-economic factors in INFERNO is discussed in section 4.2 - Model limitations and known issues.in this section it is discussed that the use of HDI reduces the ability of the model to represent the burnt area regions that are characterized by high interannual variability, namely, BONA, BOAS, AUST, CEAS, SHSA and NHSA. Although this is seen as a negative impact, it must be noted that the control model - JULES-INFERNO - despite having a larger inter-annual variability, also has a poor performance in this aspect compared to observations. Lines 392 to 396. Furthermore, in lines 410 to 417 discuss the fact that INFERNO does represent the mechanisms that are characteristic off regions with high interannual variability in burnt area. The authors agree that improving these sentence to ensure this is clear would strengthen the manuscript.
- Overall, a majority of the metrics that the authors provide show worse performance at the regional scale after implementing the HDI than before.
While the authors recognize that the analysis presented in this work would be more complete by highlighting the regions where the implementation of socioeconomic factors in fire suppression and ignitions in INFERNO have a degradation of performance. The results presented in this study show that including socio-economic factors in the fire ignition and suppression parametrisation within INFERNO leads to improved performance in regions that were affected by large biases in the JULES-INFERNO configuration (line 325). Although this as a negative impact in some of the regions, this is either small (e.g., the difference in the metric is in the order of the decimal place), or the negative impact is understood and discussed in Section 4.2 Model limitations and known issues. It should be highlighted that in the discussion model limitations and known issues we have identified that mechanisms that dominate the fire behaviour of some regions are not represented in INFERNO, the fact that JULES-INFERNO perform better in regions dominated by peat land fires and high interannual variability.
A specific section of the paper that is worth highlighting in these general comments is the abstract, which is highly misleading, and gives a false impression to a casual reader of this paper who would rely on it. It mentions only regions in which the model was improved by implementation of the HDI, without mentioning that the global bias in burned area was increased and that many regions experience worse model performance. Two statements in the abstract in particular are false.
- The first is that the bias in burned area is reduced in 4 regions in the model “without statistically significant impact to 10 other areas.” The statistical significance of these biases is not discussed in the manuscript at any point, the global bias is increased by implementing the HDI, and some regions show quite large increases in their bias. Australia, in particular, shows an increase in its relative bias from -21% to -82%. It is unclear what statistical test would define this as insignificant.
- The second statement is that the new model version “improves the representation of the burnt area trends, especially in Africa, Central Asia and Australia.” One part of Africa shows an improvement, while the other shows a worse performance. Combined, the trend over Africa in total is worse in the updated model version (details are provided in the specific comments). This statement is therefore highly misleading at best.
For this manuscript to be publishable, the authors must fundamentally reframe it in a manner that transparently reflects their results. This includes stating clearly that the functions implemented in their model are initial attempts at parametrization that are not based on previous analysis, clearly stating that the results of implementing these parametrizations are mixed, and that further work is required to derive and validate equations that reflect the impact of human development on fire activity at a global scale. This should be reflected in sections where results are summarized and discussed. The current framing overstates the results and is not supported by sufficient evidence.
The authors agree with the reviewer that the manuscript would benefit from increasing the emphasis on the regions that are negatively impacted by the approached presented, as well as the benefit that including a statistical test to objectively quantify significance of changes and will aim to improve this in a revised version of the manuscript.
-
AC3: 'Reply on RC1', João Teixeira, 29 Feb 2024
Publisher’s note: this comment is a copy of AC2 and its content was therefore removed.
Citation: https://doi.org/10.5194/bg-2023-136-AC3
-
RC2: 'Comment on bg-2023-136', Anonymous Referee #2, 28 Nov 2023
The authors aim to describe the representation of socio-economic factors in a global fire model using HDI. They describe applying a linear term to the human ignition parameterisations in INFERNO and argue that it improves the model performance in general as well as producing more accurate burnt area patterns.
Apart from a decrease in bias in some regions, performance decreases in other and especially the global values of burnt area are significantly worse than in the non-HDI version of INFERNO. Fig A1 and table 2 show that the HDI implementation doesn’t seem to work at all in areas with low to very low population density and still high HDI like AUST and BOAS. This is not a model improvement, and it doesn’t show the potential of including HDI in a global fire model. INFERNO is considered a global fire model and, therefore, an effort to add extra value to the model should aim at a general increase in performance.
The authors have updated the model by implementing revised per-PFT-BurntArea values that are independent of the implementation of HDI and added a (1-HDI) term to the ignition equations in the fire-model (equations 2&3).
The results presented in this study suggest that the straight-forward application of this dampening term (1-HDI) is not sufficient to improve the global performance. A look at Fig A2 and the regional burnt area for AUST and BOAS might suggest an application of a correction term that might weigh HDI itself by e.g. human population density, as it seems unlikely that a generally high HDI should still have its maximal effect in remote regions.
Further, one could imagine that a retuning of the whole set of empirical parameters in equations 2-4 might help.
Finally, a linear application of (1-HDI) seems arbitrary. A derivation of a factor depending on HDI for equations 2 and 3 is needed to justify the approach of choice.
A file with detailed comments is attached.-
AC1: 'Reply on RC2', João Teixeira, 29 Feb 2024
The authors are grateful for the insightful feedback provided by the reviewer on our manuscript. Your comments are valuable to us, and we will address the concerns you have raised in the attached detailed discussion. We pledge to revise our manuscript to enhance the clarity on the implementation of socio-economic factors in the INFERNO global fire model using HDI.
A reply to the main reviewer comments follows below, with original reviewer comment presented in bold. A file with detailed comments and respective replies is attached.
The authors aim to describe the representation of socio-economic factors in a global fire model using HDI. They describe applying a linear term to the human ignition parameterisations in INFERNO and argue that it improves the model performance in general as well as producing more accurate burnt area patterns.
Apart from a decrease in bias in some regions, performance decreases in other and especially the global values of burnt area are significantly worse than in the non-HDI version of INFERNO.
The approach presented in this work does improve the results from INFERNO by reducing the large bias produced by the model. JULES-INFERNO has large positive bias at a regional level, for example the bias off regions such as TENA (17.21), CEAM (4.4), SHSA (49.24), EURO (2.23), and MIDE (3.88) total to 76.96 Mha. All these biases are reduced in JULES-INFERNO+HDI. This alone shows that JULES-INFERNO performs well at the global scale as regional biases compensate each other.
Fig A1 and table 2 show that the HDI implementation doesn’t seem to work at all in areas with low to very low population density and still high HDI like AUST and BOAS. This is not a model improvement, and it doesn’t show the potential of including HDI in a global fire model. INFERNO is considered a global fire model and, therefore, an effort to add extra value to the model should aim at a general increase in performance.
Although there is as a negative impact in some of the regions, this is either small (e.g., the difference in the metric is in the order of the decimal place), or the negative impact is understood and discussed in Section 4.2 Model limitations and known issues. It should be highlighted that in the discussion model limitations and known issues we have identified that mechanisms that dominate the fire behaviour of some regions are not represented in INFERNO, the fact that JULES-INFERNO perform better in regions dominated by peat land fires and high interannual variability.
The improvements that JULES-INFERNO+HDI brings to some of the regions such as TENA, NHAF, and SHAF have a greater impact in the global standard deviation than the degradation of the standard deviation seen for regions such as CEAM, NHSA, SHSA, EURO, and MIDE. For regions such as BOAS, CEADS, SEAS, EQAS, and AUST, both model configurations perform poorly in terms of standard deviation and any differences between the STD/STDGFED4s are small when compared to the observed standard deviation (e.g., difference between the JULES-INFERNO and JULES-INFERNO+HDI STD/STDGFED4s smaller than 15%).
Furthermore, for some of these regions INFERNO is not expect to perform, especially in terms of variability. As discussed in Section 4, the fire behaviour of some of these regions is characterised by mechanisms that are not represented in INFERNO, therefore INFERNO is not expected to perform well in these regions.
The authors have updated the model by implementing revised per-PFT-BurntArea values that are independent of the implementation of HDI and added a (1-HDI) term to the ignition equations in the fire-model (equations 2&3).
Previous per-PFT-BurntArea values defined by Mangeon et al (2016) were heuristically determined, as referred in their work. These values account for the tunning towards reproducing the observed average burnt areas and may compensate for processes not represented in this implementation of INFERNO.
In this work we revised these parameters and used values that are supported by Andela et al. (2019).
The results presented in this study suggest that the straight-forward application of this dampening term (1-HDI) is not sufficient to improve the global performance. A look at Fig A2 and the regional burnt area for AUST and BOAS might suggest an application of a correction term that might weigh HDI itself by e.g. human population density, as it seems unlikely that a generally high HDI should still have its maximal effect in remote regions.
With regards to the approach taken to represent the relationships between the socio-economic effects on fire ignitions and suppression and HDI, it should be noted that socio-economic impacts on fire are complex and dependent on may factor that are difficult to model, depend on government policies, as well as cultural behaviour. The work by Pandey et al. (2023) is a good example of a study that highlights this complexity, as well as the different fire management policies around the world, showing that despite their differences they all result in a gradual reduction in fire occurrences and burned areas over time. In addition, the formulation of climate/ESM does not allow for representing these details.
The aim of this study is not to represent that complexity but rather to explore the use of the HDI to represent socio-economic impacts on fires, aiming to improve the regional representation of human–environmental coupling for applications at large spatial scales within an Earth System Model (ESM) context. As stated in the work of Mangeon et al. (2016), INFERNO is a simple physically based representation of fire activity aimed at representing fires in the ESM context. The current implementation of the HDI aims to follow that same philosophy.
Further, one could imagine that a retuning of the whole set of empirical parameters in equations 2-4 might help.
As the reviewer mention, the equations used could be tuned to provide the best results. However, that could be masking compensating bias that are existent in the model. For example, the formulation of INFERNO described in Mangeon et al. (2016) overestimated the ignitions and suppression of fires. At the time this formulation was developed average burnt area values where heuristically determined, while posterior work by Andela et al. (2019) shows that these values can be ten times larger than the ones used by Mangeon et al. (2016). Including this HDI based parametrisation it is needed, when average burnt area values in INFERNO are used to match the ones reported by Andela et al. (2017). The authors commit to better explain this in a revise manuscript and increase the clarity of the reader.
Finally, a linear application of (1-HDI) seems arbitrary. A derivation of a factor depending on HDI for equations 2 and 3 is needed to justify the approach of choice.
Despite being a simple representation while trying to encompass, this approach does align with the few studies found in literature that looked at the impact governmental policies have on prevention of wildfires. For example, the work by Curt and Frejaville (2017) shows that, the wildfire policies implemented in in Mediterranean France, resulted in the number of fires has decreased almost linearly since 1975, whereas the burned area changed more abruptly.
The authors thank the reviewer for promoting this constructive discussion and agree that there should be more detail analysis and explanation of this at an early stage in the manuscript, and commit to improve this, including this discussion in a revised version of the manuscript.
-
AC1: 'Reply on RC2', João Teixeira, 29 Feb 2024
-
RC3: 'Comment on bg-2023-136', Anonymous Referee #3, 29 Nov 2023
Teixeira et al. have integrated the Human Development Index (HDI) into the fire ignition and suppression parametrization within the INFERNO global fire model. Their findings indicate that incorporating socio-economic factors in INFERNO has led to a reduction in positive biases of burned area in specific regions and an enhanced representation of burned area trends in Central Africa. The methods proposed in this study are valuable for the scientific community, shedding light on the improved performance of the model and the potential interplay between socio-economic factors, climate, and vegetation.
While the paper is interesting and provides crucial insights into the human impacts on fire activity in the global fire model, there are notable areas that could benefit from further attention. The data and methodological approach employed seem robust, yet additional analyses are warranted to strengthen the support for the proposed approach. To enhance the manuscript, I suggest addressing the following major suggestions:
Specific comments:
1. Page 2, line 49: When introducing HDI for the first time, it would be beneficial to provide more detailed information about HDI, including its calculation method, data sources, the range (0 to 1?), and spatial and temporal resolution.
2. Page 2, line 53: Clarify how Zou et al. (2019)'s approach (based on Li et al. (2013), which includes region- and PFT-specific functions of population density and a global unified function of GDP, differs from the HDI approach in this study.
3. Page 3, Section 2: Introduce HDI in a dedicated section, covering details such as its calculation method, data sources, the range (0 to 1?), and spatial and temporal resolution.
4. Page 4, line 95: Emphasize that fNS represents the fraction "not" suppressed by humans to avoid confusion regarding the decrease in suppression as HDI increases in equation 3.
5. Page 4, line 97: While the rationale behind the parameterization with 1-HDI is explained, additional analyses are crucial to support this parameterization, as the addition of 1-HDI may appear somewhat arbitrary.
6. Page 5, line 115: Include a map of HDI to help readers visualize the global distribution of HDI.
7. Page 8, line 189: Consider moving the evaluation metrics to the methods section for better organization.
8. Page 10, Table 2: Consider marking the numbers with improvement in bold to enhance readability.
9. Page 11, Figure 5: (1) Add colors to differentiate JULES-INFERNO and JULES-INFERNO+HDI. (2) Provide explanations for -Clim simulations in the main text to avoid confusion.
10. Page 11, line 221: It’s very interesting that including HDI in the parameterization would reduce the interannual variability of modeled burned area. Is it because the HDI reduces the contributions from climate drivers?
11. Page 12, line 240: Is it because the trends over these regions are driven by climate rather than human activities? Like in southern Africa the increasing trend is mainly driven by ENSO (Andela & van der Werf, 2014).
12. Page 13, lines 278-281: (1) Specify the criteria used to determine dominant drivers in the experiments, e.g., based on differences in trend_1990 control minus trend_clim larger than a certain value or statistical significance. (2) Consider presenting the information in a figure rather than a table to enhance clarity. For example, the x-axis can be the regions and y-axis are the differences of trend between 1990 control and climate, with two bars representing JULES- INFERNO and JULES- INFERNO+HDF. A horizontal line representing the criteria mentioned above.
13. Page 17, line 352: Page 17, line 352: Please clarify that, despite consistent results, it might be confusing to state this since the HDI index does not encompass the impacts of fire management policies.
14. Page 18, lines 397-398: I think that is a very important information which should be brought up at the beginning when introducing HDI. The fact that HDI based at a national level can explain several biases when HDI is implemented in the algorithms, e.g., negative biases in northern Australia.
Technical comments:
- Page 9, line 196: RMSEUB should be RMSEUE?
- Page 16, line 321: “Discussion & Conclusion” should be placed in the section title.
Reference
Andela, N., & van der Werf, G. R. (2014). Recent trends in African fires driven by cropland expansion and El Niño to La Niña transition. Nature Climate Change, 4(9), 791–795. https://doi.org/10.1038/nclimate2313
Zou, Y., Wang, Y., Ke, Z., Tian, H., Yang, J., & Liu, Y. (2019). Development of a REgion-Specific Ecosystem Feedback Fire (RESFire) Model in the Community Earth System Model. Journal of Advances in Modeling Earth Systems, 11(2), 417–445. https://doi.org/10.1029/2018MS001368
Citation: https://doi.org/10.5194/bg-2023-136-RC3 -
AC4: 'Reply on RC3', João Teixeira, 29 Feb 2024
The authors thank the reviewer for the insightful comments on our paper. We appreciate your feedback and agree that there are areas that could benefit from further attention. We are glad to hear that you find our data and methodological approach to be robust, and we will certainly consider conducting additional analyses to strengthen the support for our proposed approach.
In response to your major suggestions, we will carefully review and address each point to enhance the manuscript. We value your input and will work to incorporate your suggestions to the best of our ability.
A file with detailed comments and respective replies is attached.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 514 | 137 | 42 | 693 | 30 | 29 |
- HTML: 514
- PDF: 137
- XML: 42
- Total: 693
- BibTeX: 30
- EndNote: 29
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
