Preprints
https://doi.org/10.5194/egusphere-2026-1085
© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/egusphere-2026-1085
© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
17 Mar 2026
| 17 Mar 2026
Fire across frontiers: Satellite-based investigation of climate-fire interactions in the Middle East (West Asia)
Abstract. Although the global burned area has declined in recent decades, fire activity worldwide is projected to become more frequent and intense due to both climate change and human activities, including fire ignition, suppression and land use changes. In the Middle East (West Asia), climate change, rapid urbanisation, cropland expansion and armed conflict impacts all play a part in the increased risk of fire. While this region has experienced several large-scale fires over the last decade, little research has focused on regional and cross-border fire dynamics and their links to climate and other biophysical factors. This study aims to fill this gap by investigating vegetation fire dynamics across Turkey, Syria, Iraq, Iran, Lebanon, Israel and Palestine between 2001 and 2022.
We assessed long-term spatio-temporal fire trends using satellite-derived active fire and burned area products. To explore the relationship between potential factors and fire activity, we used Spearman Rank Correlation to quantify the correlation between annual burned area, active fire, climate, topography and population density. Our results reveal a prominent arc-shaped transboundary fire pattern crossing international borders, with fire risk frequently concentrated along the boundaries between neighbouring countries (Turkey, Syria, Iraq and Iran). Crucially, the data show that 74 per cent of the total burned area occurred on croplands, underscoring the dominance of humans in the fire regime. Regional climate and population density show only weak or limited associations with annual active fires and burned area. While topographic factors show a stronger correlation, this relationship is largely indirect, reflecting the fact that intensive agricultural burning is concentrated in flatter, more accessible areas. This study advances understanding of fire dynamics in the Middle East and further supports more effective fire risk mitigation and preparedness in the future.
How to cite. Lin, W. N., Abdi, A. M., and Eklund, L.: Fire across frontiers: Satellite-based investigation of climate-fire interactions in the Middle East (West Asia), EGUsphere [preprint], https://doi.org/10.5194/egusphere-2026-1085, 2026.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Download & links
- Preprint
(1787 KB) - Metadata XML
-
Supplement
(1460 KB) - BibTeX
- EndNote
Status: final response (author comments only)
Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor
| : Report abuse
-
RC1: 'Comment on egusphere-2026-1085', Anonymous Referee #1, 03 Apr 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-1085/egusphere-2026-1085-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2026-1085-RC1 -
RC2: 'Comment on egusphere-2026-1085', Anonymous Referee #2, 12 Apr 2026
Lin and others have presented a compelling spatiotemporal analysis of fires in the Middle East and their correlations with several explanatory variables. This is an important manuscript given its cross-border nature and analysis in an otherwise sparsely studied area. The methods presented in this study can be further extended by other researchers to analyze the impacts of climate and socio-political changes in the region.
While this is a high impact manuscript, I believe that there are several gaps primarily in the presentation of the analysis, and description of methodology that if addressed, would significantly improve it for readers. I have listed my suggestions for improvements below:
- I am not convinced with the focus on “satellite-based investigation” in the title. For example, the word “satellite” does not appear even once in the Introduction. Since the authors have not done any direct analysis of satellite imagery, but used GIS products developed by other authors/agencies, I would suggest modifying the title to better reflect the content of the manuscript. For example, “GIS-based investigation” could be an alternative.
- Similar to the previous comment, the authors have not focussed exclusively or primarily on climate factors, but to a broader set of variables. I would suggest modifying the title to better reflect their analysis.
- Fig 1: Can latitudes and longitudes be superimposed on the map for reference, and for association with the text on Line 122.
- Line 128: Can the geographical references mentioned in the text be included in Fig 1 for readers unfamiliar with the region?
- Line 159: Please add citation for MODIS. Applies to multiple mentions of MODIS products in the manuscript.
- Line 159: Please include text that Table 1 includes spatial resolution of the products so readers can refer to it. I had this question until I reached Table 1.
- Line 159: Can the authors include temporal resolution of the MODIS products, either in Table 1 or in the main text?
- Line 169: The relevance of GEE and FIRMS in obtaining data is unclear. Are they the repositories that maintain the products?
- Line 171: The data selection criteria could be further expanded. Adding more information about the products would help understand the quality of data and which attributes are available for filtering.
- Line 183: Please describe the SPEI-X notation.
- Line 183: Please explain “nine”. I am guessing this is based on some sliding time window over a year and a definitive explanation will be helpful for readers.
- Line 189: Please expand NDVI, including its brief description, spatial and temporal resolution, and a description of processing the data.
- Line 208: I expect variability in population density of a dynamic region like the Middle East over the 20-year period of this study. Can the authors provide further justification of their choice to use the population density from a single more recent year, and its influence on their results?
- Line 209: Please expand on the definition of constrained vs unconstrained products.
- Line 214: Please include resolution of the product.
- Table 1: Please expand on “input variable” in the the caption for the table.
- Section 2.3.1: The description is unclear about the data cleanup processes undertaken here. Could the authors clarify further, and if possible, with an example schematic? For example, why would only the specified classes reflect spatial misalignments, what is the effect of area differences, what are comparable fractions?
- Line 234: Does the grid size reflect the horizontal and vertical axes, or the diagonal?
- Line 235: Was a single non-overlapping grid created over the entire study region, or was it created separately for each country? In case of the former, was there a starting x,y coordinate of the grid? Was the grid created based on geodesic distance starting at some x, y?
- Line 236: Can the authors add a sentence describing queen contiguity, and include a reference?
- Line 244: Is wi,j parametric or non-parametric depending on the distance between cells i and j? Can the authors include its formulation?
- Line 245: What is the value of n for the study region?
- Line 252: What is active data?
- Line 258: Does xj have a time component or is it the total count over the entire time period?
- Eq 2: Does the statistic assume weights to be spatial only or is that a choice made by the authors to assume uniform temporal weights?
- Eq 2: What is the time component in the equation?
- Line 268: It will be helpful to include a table defining each term in the main manuscript as it is essential for the understanding of the methodology results.
- Line 278: What is zero value redundancy?
- Line 278: Are the MODIS products temporally consistent across the spatial domain, e.g., the products are available at the same time frequency for the domain so that summing the burned pixels is equivalent?
- Line 280: Description of the MK test can be moved to the previous section since it was already introduced there.
- Eq 3: Are xi at pixel level or grid level?
- Line 289: Eq 3 is a summation with (I believe) max value = n(n-1)/2. Is there a missing normalization to convert S to [-1, +1]?
- Line 291: Since only the BA is computed here, why would >0 values indicate more frequent fires?
- Line 291: I suggest clarifying that the trend is toward recent years, i.e., “more frequent or severe fires in recent years”.
- Line 293: How does the p-value threshold, mentioned on Line 284, relate with the chosen values of -0.5 and 0.5?
- Line 297: Table 1 does not seem to describe data homogenization. Can the authors please describe this further?
- Line 298: What is meant by “non-directional measure of association”?
- Line 303: What is the difference between cells and observations, to help understand their different counts?
- Line 304: Is there a clean separation of cropland and non-cropland BA? For example, can BA fires expand into non-BA fires?
- Line 304: Could the authors add information about how each variable was assigned to a grid cell? For example, it is possible that the same grid cell has multiple land cover classifications since its resolution is 500m. How was that grid cell classified for the analysis, and for cropland?
- Line 312: Is the limited detection in Gaza because of the country’s size combined with the grid layout?
- Figure 2: Can the authors elaborate on the y-axis? Is the normalization of y-axis done over the entire country’s area, over the grid cells of that type, or only over grid cells that have active fires? Same comment for Figure 3.
- Table 2: How is 10km2 area determined for counts?
- Line 326: Can the authors please elaborate on how the conclusion about more diverse land cover types is drawn from Figures 2 and 3? The land cover distribution looks similar for AF and BA, especially in the countries with majority of the fires.
- Table 3: How do pixels relate to the grid cells?
- Table 3: Can the table values be further described? Does a value of 2 mean - 2 pixels (?) burned over the entire time period in the entire country for that land cover type?
- Line 336: Can the authors quantify highest recurrence, e.g., “(20 pixels)”?
- Figure 4: How is 10km2 area determined for counts?
- Figure 4: Given the similar counts/area of the top and bottom panel figures, the figures will be easier to interpret and compare if they used the same axes scales in each panel. If possible and if the figures remain clear, I would suggest improving the axes scales.
- Line 356: Can the authors quantify significant p-value?
- Line 358: How is the classification done based on the EHA results?
- Section 3.4: Would it be possible to associate the significant peaks in Fig 5 with associated major fire events? This would bolster the authors’ findings, help the readers understand the underlying reasons for the peaks, and provide assurance that these are not data anomalies. After reading the Discussion section, I realized that this analysis was performed there. It would be helpful for future readers if the manuscript was reorganized to include that analysis along with the statistical observations, or at least include a reference to the explanations presented in the Discussion section.
- Figure 5: It will be helpful to see an overlay of BA on the time series for these regions, and whether the high frequencies correlate with higher BA.
- Line 375: Perhaps “southern Turkey” could be changed to south-western Turkey (area 2) to contrast with the southern Turkey (area 1).
- Figure 5: Similar to Fig 5 comment, it will be helpful to see an overlay of AF on the time series for these regions, and whether the high BA correlates with higher AF.
- Line 385: Suggest adding rho at the beginning of the paragraph to clarify that it refers to the correlation, e.g., Spearman rank correlation, rho, …
- Line 385: From Table 4, it appears that significance is determined by p-value. Suggest clarifying this in the text. Is the p-value also calculated as part of the Spearman rank correlation?
- Line 393: Are the correlations computed between spei_wet of the previous year to the BA the following year? Otherwise, the following statement cannot be established.
- Line 398: The table footnote is in line with the text. Suggest moving it to the table caption for clarity.
- Line 401: Does spatial heterogeneity refer to no consistent correlation across areas with explanatory variables? The term can be further clarified.
- Section 3.6.1: Along with the statistical observations, are the authors able to comment on the reasons behind the observations? Without them, it is not possible to ascertain whether the observations are supported by physical evidence. Also applies to section 3.6.2-3.
- Line 430: The arc shaped fire pattern has not been highlighted in previous manuscript sections but listed here as an observation.
- Line 433: The cropland and non-cropland analysis is shown for Spearman correlation but other distributions are not included in the main manuscript. Since 74% contribution is highlighted as an observation throughout the manuscript, I suggest including further analysis of cropland vs non cropland distribution of fires within the main manuscript.
- Line 433: The lagged event is not sufficiently demonstrated as it is unclear from the analysis whether the correlations were developed with lagged parameters.
- Line 435: Since the correlation is developed between the dependent variable with a single explanatory variable, wouldn’t this constitute univariate attribution?
- Line 441: It will be helpful if the authors included a graphic map marking the Mediterranean zone, for example in Fig 1, since they reference it multiple times, and it would make it easier for the readers to follow their observations.
- Line 442: The authors have performed correlation analysis, but not demonstrated predictive capability of different variables (which would require time lagged modeling), to support this statement.
- Line 443: Similar to a previous comment, the term “spatial heterogeneity” is not sufficiently defined.
- Line 444: Since the statement is included as one of the key conclusions in the main manuscript, I would suggest moving this material and any associated text from supplementary to main manuscript.
- Line 487: This observation has not been highlighted in previous sections and to some extent, contradict the statement on line 349: “Most regions exhibit bimodal fire seasonality”.
- Line 505: Similar to some previous comments, it will be helpful for the readers if text and figures that are relevant for the primary conclusions in the main manuscript were included within the main manuscript. Also applies to Line 507.
Citation: https://doi.org/10.5194/egusphere-2026-1085-RC2
Viewed
Total article views: 283 (including HTML, PDF, and XML)
Cumulative views and downloads
(calculated since 17 Mar 2026)
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 193 | 78 | 12 | 283 | 29 | 10 | 13 |
- HTML: 193
- PDF: 78
- XML: 12
- Total: 283
- Supplement: 29
- BibTeX: 10
- EndNote: 13
Viewed (geographical distribution)
Total article views: 252 (including HTML, PDF, and XML)
Thereof 252 with geography defined
and 0 with unknown origin.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
1
Latest update: 29 Apr 2026
Department of Earth and Environmental Sciences, Lund University, Sölvegatan 12, 223 62, Lund, Sweden
Centre for Advanced Middle Eastern Studies (CMES), Lund University, 22362, Lund, Sweden
Abdulhakim M. Abdi
Department of Earth and Environmental Sciences, Lund University, Sölvegatan 12, 223 62, Lund, Sweden
Centre for Advanced Middle Eastern Studies (CMES), Lund University, 22362, Lund, Sweden
Lina Eklund
Department of Earth and Environmental Sciences, Lund University, Sölvegatan 12, 223 62, Lund, Sweden
Centre for Advanced Middle Eastern Studies (CMES), Lund University, 22362, Lund, Sweden
United Nations University, Institute for Water, Environment and Health (UNU-INWEH), 225 East Beaver Creek Road, Richmond Hill, ON, L4B 3P4, Canada
Short summary
This study explores 22 years of fires in the Middle East using satellite data. We found that 74 % of fires occur on croplands, indicating that human activities are central to fire regimes in this region. A striking arc-shaped pattern along borders shows that fire is a transboundary challenge. Our results suggest looking at human decisions and politics to provide a roadmap for better fire management and cross-border cooperation in the region.
This study explores 22 years of fires in the Middle East using satellite data. We found that 74...