Next Article in Journal
Preconditioning of Dust and Fluid in a 20 L Chamber During Ignition by a Chemical Ignitor
Previous Article in Journal
Flame-Retardant Polyurea Coatings: Mechanisms, Strategies, and Multifunctional Enhancements
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Landsat Time Series Analysis with BFAST for Detecting Degradation of Thyme Shrublands by Fire on Lemnos Island

by
Georgios K. Vasios
1,*,
Eleftheria Alexoudaki
1,
Aggeliki Kaloveloni
1 and
Andreas Y. Troumbis
2
1
Laboratory of Sustainable Agrifood and Smart Farming, Department of Food Science and Nutrition, School of the Environment, University of the Aegean, 81400 Myrina, Greece
2
Department of Environment, School of the Environment, University of the Aegean, 81100 Mytilene, Greece
*
Author to whom correspondence should be addressed.
Fire 2025, 8(8), 335; https://doi.org/10.3390/fire8080335
Submission received: 16 July 2025 / Revised: 18 August 2025 / Accepted: 19 August 2025 / Published: 21 August 2025

Abstract

Landsat time series data, which have become freely available in recent years, are commonly used to detect changes in land cover and monitor ecosystem disturbances. Thyme habitats are areas under protection due to their high ecological value. However, human activity leading to land use competition, mainly from overgrazing, poses an increased threat to these habitats. The impact of these disturbances is underreported, and their detection remains essential for thyme conservation. The island of Lemnos was chosen as the study area, because of the significant areas of thyme habitats, which are currently under pressure due to rural abandonment, desertification, overgrazing, and systematic fires in recent decades. A long-term Landsat time series was generated, and the Normalized Difference Vegetation Index (NDVI) was calculated. The change detection algorithm (BFAST) was used to detect and characterize significant changes (breakpoints) within the time series and compare them to local fire events. The analysis showed that Lemnos thyme habitats have been significantly reduced in size due to fires and their conversion to new grazing areas for livestock production. Measures should be taken to conserve thyme habitats with the participation of local stakeholders, including livestock farmers and beekeepers. Satellite monitoring techniques are important tools that could facilitate this conservation process.

1. Introduction

Detecting changes in land cover is important for monitoring ecosystem disturbances and understanding the key natural or anthropogenic processes and drivers that influence their dynamics through space and time [1,2]. In recent decades, remote sensing has become an important tool for detecting these changes in ecosystems and landscapes, facilitating the implementation of more sustainable conservation strategies [3]. In addition, in recent years, available data from satellite time series have provided the opportunity to reconstruct the spatio-temporal history of ecosystem disturbances and degradation [4]. The Landsat time series data, freely available in recent years, have become a useful tool for detecting land cover change and monitoring ecosystem disturbances [5,6]. New algorithms have been developed and introduced to the scientific community for analyzing these long-term datasets [7]. The most commonly used time series change detection models are (a) Breaks for Additive Season and Trend (BFAST) [8], (b) Continuous Change Detection and Classification (CCDC) [9], and (c) Landsat-based detection of Trends in Disturbance and Recovery (LandTrendR) [10].
The Breaks for Additive Season and Trend (BFAST) method analyzes time series using methods to detect and characterize change within the time series of a particular pixel. BFAST algorithm decomposes the original time series data into three components: trend, seasonal, and remainder [8]. Changes detected in the trend component indicate gradual and abrupt change, while changes detected in the seasonal component indicate phenological changes. Breaks identified in the trend component represent an ephemeral disturbance, while breaks identified in the seasonal component represent a change in land cover due to the difference in seasonal pattern (phenology) between different land cover types [11]. Furthermore, after identifying and modeling the fixed part of a time series, the algorithm can detect disturbances in near real-time within newly acquired data, facilitating faster response to manage their effects [12].
Regional ecosystem disturbances such as fires and floods in semi-arid areas were detected, and vegetation response patterns to these abrupt changes were studied using the BFAST method on MODIS EVI time series [13]. Focusing on fire occurrence in areas where grass and woody plants coexist, such as in African savannas [14], MODIS fire data were analyzed with the BFAST algorithm, providing information on fire trends that could support better fire management policies. For better spatial resolution, Landsat time series were used with BFAST trend break analysis to detect land cover changes in savannahs that occurred naturally or due to human activities, including livestock and tourism [15]. In another study, incorporating AVHRR, MODIS, and Landsat time series, BFAST was used to estimate detailed land use and land cover maps of China for the period 1980–2015 [16].
Fire detection and its long-term and/or short-term effects on vegetation have been estimated and analyzed using the indices applied to MODIS [17], Landsat [18], and various high-resolution [19] satellite time series data. The Normalized Difference Vegetation Index (NDVI) is the most commonly used. The BFAST method combined with NDVI was used to identify forest change [20] and study the effect of grazing on vegetation in the Mediterranean region [21]. Many semi-arid areas of the Mediterranean are under pressure from extensive grazing [22], and more research is needed to understand its long-term impact on ecosystems.
In this research, we focus on using Landsat satellite time series data to detect significant land cover change in semi-arid Mediterranean landscapes mainly due to localized fire events. The island of Lemnos, in the North Aegean of Greece, was selected as the case study. The semi-natural areas of the island, which were originally mainly thyme habitats, have been used for different agricultural activities, such as livestock production and beekeeping, during the last decades. Our aim is to apply time series analysis using the BFAST algorithm to detect significant land cover changes (breakpoints) in these areas that have occurred long-term from fire events and affected their ecosystems.

2. Materials and Methods

2.1. Study Area

This study was conducted on the island of Lemnos, located in the North Aegean Sea of Greece. Lemnos (also spelled Limnos in contemporary sources) is the eighth largest island of Greece, covering an area of 479 km2 [23] (Figure 1). The landscape of the island is mainly flat, including low-altitude hills, with the highest hill (429 m) located in the northwestern part of the island. At the same time, the topography is created by volcanic, shale, and sandstone formations [24]. The island’s climate is almost semi-arid, with an average annual precipitation of 500 mm, a hot, dry summer, and a mild winter. Frequent winds blowing mainly from the north or northeast contribute to the dry climate of the island [23,24]. The island has a variety of habitats, including wetlands, sand dunes, agricultural crops, and extended phrygana and thyme habitats. A variety of human activities take place, such as tourism and extended agriculture, mainly livestock farming, including intense grazing and burning practices, which have affected the island’s ecosystems [23,24,25].

2.2. Data

A long-term Landsat time series of the island was created for the period 1984–2021 (37 years) by combining images from the L5/TM, L7/ETM, and L8/OLI satellites. The Landsat archive Collection-2 Level-2 Science products, which were used in our analysis, are high-quality time series observational data for detecting land cover changes [26,27,28] and were downloaded from the United States Geological Survey’s (USGS) online tool EarthExplorer https://earthexplorer.usgs.gov (accessed on 18 February 2022). From the 1103 available images, 420 high-quality images covering Lemnos Island were selected, after setting a cloud cover threshold of 10% or less in the study area (Figure 2). The Landsat time series with 420 images in total consists of 184 (43.8%) L5/TM, 151 (36.0%) L7/ETM, and 85 (20.2%) L8/OLI satellite images, which are distributed between 37 years for the period 1984–2021 (Figure 3), with yearly cover especially in the cloudless summer months (see Figure S1).
The CORINE Land Cover 2018 (CLC) dataset was selected for classifying the land cover of Lemnos surface. CLC provides a pan-European land cover inventory of 44 thematic classes divided into three levels of detail (accuracy ≥85%) [29,30]. Lemnos Island includes 5 main classes of Level 1, with two of them covering 94.4% of the island, the agricultural areas (54.0%), and the forest and semi-natural areas (40.4%). In more detail, CLC includes 13 classes of Level 2 and 18 classes of Level 3 (Figure 4). The agricultural areas include non-irrigated arable land (30.3%), land principally occupied by agriculture, with significant areas of natural vegetation (16.7%), complex cultivation patterns (5.2%), and pastures (1.8%). The forest and semi-natural areas include mainly natural grasslands (38.1%), with small areas of sclerophyllous vegetation (1.2%), sparsely vegetated areas (0.8%), and beaches, dunes, and sands (0.3%) (see Table S1).
All available data of burnt areas of Lemnos Island were collected from the Fire Service of Lemnos https://www.fireservice.gr (accessed on 25 May 2024) for the period 1999–2020 (22 years), and they were digitized, categorized, and organized in a database. A total area of 9.82 km2 was burned during this period, including forest and semi-natural areas (47.1%), agricultural areas (42.1%), wetlands (10.1%), and artificial surfaces (0.7%). In Table 1, the burnt area is analyzed into its main land cover types (CLC, Level 1) and how these are distributed per month for the whole period. The month with the highest burnt area is August, including 50.1% of the total burnt area (4.92 km2), with 29.7% (2.92 km2) estimated as agricultural areas and 20.0% (1.97 km2) as forest and semi-natural areas.
In a total of 635 fire events of the database, three large fire events were detected on forest and semi-natural areas: (1) from 3 August 2015 to 4 August 2015 in the Myrina area, west of the island, 1.50 km2 were burned; (2) from 19 November 2016 to 20 November 2016 in the Atsiki area, north of the island, 0.90 km2 were burned; and (3) from 24 December 2016 to 25 December 2016 in the Kaminia area, southeast of the island, 0.50 km2 were burned. The study of fire events was used as historical field data to compare with the results from the time series analysis.

2.3. Methods

Our analysis consists of two main analytical steps: (a) calculating the NDVI values of each Landsat image of the study area and (b) detecting land cover changes in the time series using the BFAST algorithm. Focusing on forest and semi-natural areas categorized by Corine 2018 [29], large fire events were detected based on the fire events database (Figure 5).
The Normalized Difference Vegetation Index (NDVI) is calculated by combining the near-infrared band and red band, based on the following equation [2]:
NDVI = (NIR − RED)/(NIR + RED)
For the study area, the NDVI was calculated per pixel for each satellite image, using the NIR (band 4) and the RED (band 3) from the L5 and L7 satellites and the NIR (band 5) and the RED (band 4) from the L8 satellite, respectively [2].
The change detection algorithm Breaks for Additive Season and Trend (BFAST) was used to detect and characterize significant changes (breakpoints) within the NDVI time series [8,12]. The BFAST technique estimates the time and number of abrupt changes within time series and characterizes the change by its magnitude and direction. The BFAST01 algorithm was used from the BFAST package (version 1.7.0) in the R statistical environment (version 4.4.3), which produces a model containing either one or zero breakpoints per pixel of the study area.

3. Results

The NDVI was applied to the Landsat time series (1984–2021), which consists of 420 images of the Lemnos Islands. With the BFAST01 algorithm, we calculated the decomposition of the pixel-wise NDVI time series into different components: linear trend, seasonal fitted curve, and residuals. Figure 6 presents these components and the original NDVI values (Response curve) of a sample pixel of thyme habitat.
Various change patterns combining the linear trends and breakpoints calculated by the BFAST01 algorithm for each pixel were detected. In Figure 7, four examples of different change types of thyme habitats are presented (a–d): (a) restoration period following a significant disturbance at the breakpoint; (b) continued significant disturbance after a breakpoint; (c) from a mixed land-cover type change to a stable type after a breakpoint; and (d) mixed land-cover type under disturbance before a breakpoint and restoration afterwards. In Figure 7e, maps of NDVI values of Lemnos Island for four sample years are presented (23 March 1990, 2 March 2000, 1 May 2010, and 10 April 2020).
The breakpoints estimated using the BFAST01 algorithm cover 88.7% of the total surface of Lemnos Island (see Figure S2, Table S2). Figure 8 shows that most of the forest and semi-natural areas of the island are detected with significant changes (96.1%). Most of these breakpoints occurred in the last two decades of the study period, especially after 2010 (see Table S2).
In Figure 9, the largest fire event in the forest and semi-natural areas of Lemnos that burned 1.50 km2 from 3 August 2015 to 4 August 2015 in the Myrina area, west of the island, is mapped. The comparison between pre-fire (3 August 2015) and post-fire (19 August 2015) images shows an estimation of the burned area. The dNDVI values between pre/post-fire event enhance the burned area and are calculated as the difference between the two NDVI images [dNDVI = NDVI (pre-fire) − NDVI (post-fire) = NDVI (3 August 2015) − NDVI (19 August 2015)]. The dNDVI is used as a background layer for mapping the fire event’s breakpoints in Figure 9c. The significant change detection due to the fire event is estimated using the BFAST01 algorithm by calculating the breakpoints of the area (751 pixels) based on the NDVI values of the time series. The area of change (0.68 km2) that was detected covers only 45.1% of the burned area and shows the long-term disturbance of the ecosystems by the fire event that occurred in August 2015.

4. Discussion and Conclusions

The types of changes (patterns) classified with BFAST-based change-point analysis allow the detection of a major break event in the time series. These patterns are statistically a unique combination of linear trends: two trends before and after a major breakpoint, and only one trend if no breakpoint is detected. Their classification provides an important tool to detect land cover changes over a long-term period. On Lemnos Island, a total of 472,151 breakpoints were estimated by our analysis, which shows a significant disturbance detection in all land cover classes and covering 88.7% of the total surface. These land cover changes occurred mostly after 2010 and are relevant to literature findings on significant degradation of rural and wild areas in the Mediterranean region and Lemnos Island [31,32,33]. Focusing on the forest and semi-natural areas, we detected significant disturbances (96.1%) in most of these areas, which occurred mainly after 2010 (see Figure 8 and Figure S2, Table S2). These results show the significant pressure and impact on the agricultural cultivation and important wild vegetation of the island, such as thyme habitats, even though a large part of these areas are under the Natura 2000 network conservation https://natura2000.eea.europa.eu (accessed on 15 June 2024) [24,34].
Conflicts over the use of wildlands are frequently documented and analyzed in the literature [35]. Although the concept is somewhat ambiguous, specific cases occur in situations where land ownership is unclear, such as due to a lack of cadastral plans, or where there is a high stake in real estate, such as in the Wildland–Urban Interface (WUI) [36]. Conflicts also arise when different primary production needs on the land contribute to landscape degradation [37]. It is plausible to hypothesize that, when available land decreases in size and production practices are fundamentally different, conflicting parties may resort to illegal activities. A prime example of such a scenario is on the Lemnos Island, Greece, where beekeepers and livestock producers clash over thyme shrublands. The former are concerned because thyme serves as the primary habitat for bees, contributing to the production of high-quality Lemnos honey, while the latter face low feed value for their sheep and goats due to the dominance of thyme. Arson has unfortunately become an established practice used to resolve this conflict on the island https://www.fireservice.gr (accessed on 25 May 2024). Our analysis showed that the thyme habitats have been affected by the fire events that have occurred on the island over the past decades, which have suffered habitat degradation and loss. These significant changes provided new grazing areas for livestock, which supported the increase in local dairy production. An interesting note is the belief of old-in-age livestock breeders that the milk produced in the past had a stronger aroma than nowadays, due to the herbs, especially thyme, with which the animals were grazed a few decades ago [38].
Long-term monitoring of forest and semi-natural areas using satellite archives is an important methodology for detecting land cover changes, such as those that have occurred from fire events and have a long-term impact on ecosystems [39,40]. Our findings show the long-term impact of fire on the ecosystems of the island. Due to the spatial resolution of Landsat satellite images, which have pixel sizes of 30 m × 30 m, Landsat archives can only reveal important events that took place on the surface of Lemnos Island [41]. Supplementary to Landsat, other satellite sensors could provide additional time series datasets for analysis on the study area [42,43], combining the BFAST algorithm and the Normalized Difference Vegetation Index (NDVI), with other vegetation and fire indices, such as the Normalized Burned Ratio (NBR) [44,45,46].
During the study period (1984–2021) of the last 37 years, the archive of the Landsat time series (Collection-2 Level-2 Science products) is an important spatial documentation of the history of Lemnos land cover change and BFAST techniques that could support spatio-temporal data mining analysis. Especially for the conservation of thyme habitats, measures should be taken with the participation of local stakeholders, including livestock breeders and beekeepers [47]. Satellite archive analysis techniques are important state-of-the-art tools that could facilitate a more efficient conservation process and support monitoring ecosystem changes and management.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/fire8080335/s1, Figure S1: Distribution per month of the 420 Landsat images from the satellites L5, L7 and L8; Figure S2: Map of the breakpoints on the Lemnos Island estimated using the BFAST01 algorithm. The significant changes are detected in different years for the study period 1984–2021. Year distribution is shown with color palette; Table S1: CORINE Land Cover 2018 (CLC) classes for the study area of Lemnos island. CLC of Lemnos in an area of 478.8 km2 includes 5 classes of Level-1, 13 classes of Level-2 and 18 classes of Level-3; Table S2: Distribution of the breakpoints per year on the Lemnos Island for (a) the total area and (b) the forest and semi natural areas. The number of breakpoints (in pixels) is estimated using the BFAST01 algorithm.

Author Contributions

Conceptualization, G.K.V., E.A. and A.Y.T.; methodology, G.K.V. and E.A.; software, G.K.V., E.A. and A.K.; validation, G.K.V., E.A. and A.K.; formal analysis, E.A. and A.K.; investigation, G.K.V. and E.A.; resources, E.A.; data curation, E.A. and A.K.; writing—original draft preparation, G.K.V., E.A. and A.Y.T.; writing—review and editing, G.K.V., E.A., A.K. and A.Y.T.; visualization, G.K.V., E.A. and A.K.; supervision, G.K.V.; project administration, G.K.V. and A.Y.T.; funding acquisition, E.A. and A.K. All authors have read and agreed to the published version of the manuscript.

Funding

This research is co-financed by Greece and the European Union (European Social Fund-ESF) through the Operational Programme “Human Resources Development, Education and Lifelong Learning 2014–2020” in the context of the project “Methodology of spatio-temporal analysis for disturbances on thyme areas, for their conservation and restoration at the level of local communities, combining multi-spectral systems across multiple spatial scales” (MIS 5048198).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Acknowledgments

The authors thank the three anonymous reviewers, who helped improve the manuscript at earlier stages.

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

References

  1. Fang, X.; Zhu, Q.; Ren, L.; Chen, H.; Wang, K.; Peng, C. Large-scale detection of vegetation dynamics and their potential drivers using MODIS images and BFAST: A case study in Quebec, Canada. Remote Sens. Environ. 2018, 206, 391–402. [Google Scholar] [CrossRef]
  2. Higginbottom, T.P.; Symeonakis, E. Identifying ecosystem function shifts in Africa using breakpoint analysis of long-term NDVI and RUE data. Remote Sens. 2020, 12, 1894. [Google Scholar] [CrossRef]
  3. Rose, R.A.; Byler, D.; Eastman, J.R.; Fleishman, E.; Geller, G.; Goetz, S.; Guild, L.; Hamilton, H.; Hansen, M.; Headley, R.; et al. Ten ways remote sensing can contribute to conservation. Conserv. Biol. 2015, 29, 350–359. [Google Scholar] [CrossRef]
  4. Cabezas, J.; Fassnacht, F.E. Reconstructing the vegetation disturbance history of a biodiversity hotspot in central Chile using Landsat, BFAST and LandTrendr. Int. Geosci. Remote Sens. Symp. (IGARSS) 2018, 8518863, 7636–7639. [Google Scholar] [CrossRef]
  5. Zhu, Z.; Wulder, M.A.; Roy, D.P.; Woodcock, C.E.; Hansen, M.C.; Radeloff, V.C.; Healey, S.P.; Schaaf, C.; Hostert, P.; Strobl, P.; et al. Benefits of the free and open Landsat data policy. Remote Sens. Environ. 2019, 224, 382–385. [Google Scholar] [CrossRef]
  6. Wulder, M.A.; Roy, D.P.; Radeloff, V.C.; Loveland, T.R.; Anderson, M.C.; Johnson, D.M.; Healey, S.; Zhu, Z.; Scambos, T.A.; Pahlevan, N.; et al. Fifty years of Landsat science and impacts. Remote Sens. Environ. 2022, 280, 113195. [Google Scholar] [CrossRef]
  7. Yan, J.; He, H.; Wang, L.; Zhang, H.; Liang, D.; Zhang, J. Inter-Comparison of Four Models for Detecting Forest Fire Disturbance from MOD13A2 Time Series. Remote Sens. 2022, 14, 1446. [Google Scholar] [CrossRef]
  8. Verbesselt, J.; Hyndman, R.; Newnham, G.; Culvenor, D. Detecting trend and seasonal changes in satellite image time series. Remote Sens. Environ. 2010, 114, 106–115. [Google Scholar] [CrossRef]
  9. Zhu, Z.; Woodcock, C. Continuous change detection and classification of land cover using all available Landsat data. Remote Sens. Environ. 2014, 144, 152–171. [Google Scholar] [CrossRef]
  10. Kennedy, R.; Yang, Z.; Cohen, W. Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr—Temporal segmentation algorithms. Remote Sens. Environ. 2010, 114, 2897–2910. [Google Scholar] [CrossRef]
  11. Wu, L.; Li, Z.; Liu, X.; Zhu, L.; Tang, Y.; Zhang, B.; Xu, B.; Liu, M.; Meng, Y.; Liu, B. Multi-type forest change detection using BFAST and monthly landsat time series for monitoring spatiotemporal dynamics of forests in subtropical wetland. Remote Sens. 2020, 12, 341. [Google Scholar] [CrossRef]
  12. Verbesselt, J.; Zeileis, A.; Herold, M. Near real-time disturbance detection using satellite image time series. Remote Sens. Environ. 2012, 123, 98–108. [Google Scholar] [CrossRef]
  13. Watts, L.M.; Laffan, S.W. Effectiveness of the BFAST algorithm for detecting vegetation response patterns in a semi-arid region. Remote Sens. Environ. 2014, 154, 234–245. [Google Scholar] [CrossRef]
  14. Marden, A.W.; Meyer, T.; Crews Meyer, K.A. Regional fire occurrence in Southern Africa using BFAST iterative break detection in seasonal and trend components of a MODIS time series. S. Afr. Geogr. J. 2023, 105, 200–221. [Google Scholar] [CrossRef]
  15. Borges, J.; Higginbottom, T.P.; Cain, B.; Jones, M.; Symeonakis, E. Landsat time series reveal forest loss and woody encroachment in the Ngorongoro Conservation Area, Tanzania. Remote Sens. Ecol. Conserv. 2022, 8, 808–826. [Google Scholar] [CrossRef]
  16. Xu, Y.; Yu, L.; Peng, D.; Zhao, J.; Cheng, Y.; Liu, X.; Li, W.; Meng, R.; Xu, X.; Gong, P. Annual 30-m land use/land cover maps of China for 1980–2015 from the integration of AVHRR, MODIS and Landsat data using the BFAST algorithm. Sci. China Earth Sci. 2020, 63, 1390–1407. [Google Scholar] [CrossRef]
  17. João, T.; João, G.; Bruno, M.; João, H. Indicator-based assessment of post-fire recovery dynamics using satellite NDVI time-series. Ecol. Indic. 2018, 89, 199–212. [Google Scholar] [CrossRef]
  18. Tyukavina, A.; Potapov, P.; Hansen, M.C.; Pickens, A.H.; Stehman, S.V.; Turubanova, S.; Parker, D.; Zalles, V.; Lima, A.; Kommareddy, I.; et al. Global Trends of Forest Loss Due to Fire From 2001 to 2019. Front. Remote Sens. 2022, 3, 825190. [Google Scholar] [CrossRef]
  19. Vanderhoof, M.K.; Burt, C.; Hawbaker, T.J. Time series of high-resolution images enhances efforts to monitor post-fire condition and recovery, Waldo Canyon fire, Colorado, USA. Int. J. Wildland Fire 2018, 27, 699–713. [Google Scholar] [CrossRef]
  20. Costa, H.; Giraldo, A.; Caetano, M. Exploring BFAST to detect forest changes in Portugal. In Proceedings of the SPIE—The International Society for Optical Engineering, Online, 20 September 2020; Volume 11533, pp. 38–45. [Google Scholar] [CrossRef]
  21. Von Keyserlingk, J.; de Hoop, M.; Mayor, A.G.; Dekker, S.C.; Rietkerk, M.; Foerster, S. Resilience of vegetation to drought: Studying the effect of grazing in a Mediterranean rangeland using satellite time series. Remote Sens. Environ. 2021, 255, 112270. [Google Scholar] [CrossRef]
  22. Psyllos, G.; Hadjigeorgiou, I.; Dimitrakopoulos, P.G.; Kizos, T. Grazing Land Productivity, Floral Diversity, and Management in a Semi-Arid Mediterranean Landscape. Sustainability 2022, 14, 4623. [Google Scholar] [CrossRef]
  23. Panitsa, M.; Snogerup, B.; Snogerup, S.; Tzanoudakis, D. Floristic investigation of Lemnos Island (NE Aegean area, Greece). Willdenowia 2003, 33, 79–105. [Google Scholar] [CrossRef]
  24. Thomas, K.; Thanopoulos, R.; Knüpffer, H.; Bebeli, P.J. Plant genetic resources of Lemnos (Greece), an isolated island in the Northern Aegean Sea, with emphasis on landraces. Genet. Resour. Crop Evol. 2012, 59, 1417–1440. [Google Scholar] [CrossRef]
  25. Papageorgiou, D.; Bebeli, P.J.; Panitsa, M.; Schunko, C. Local knowledge about sustainable harvesting and availability of wild medicinal plant species in Lemnos Island, Greece. J. Ethnobiol. Ethnomed. 2020, 16, 36. [Google Scholar] [CrossRef]
  26. Earth Resources Observation and Science (EROS) Center. Landsat 4–5 Thematic Mapper Level-2, Collection 2 [Dataset]. U.S. Geological Survey. 2020. Available online: https://www.usgs.gov/centers/eros/science/usgs-eros-archive-landsat-archives-landsat-4-5-tm-collection-2-level-2-science (accessed on 18 February 2022).
  27. Earth Resources Observation and Science (EROS) Center. Landsat 7 Enhanced Thematic Mapper Plus Level-2, Collection 2 [Dataset]. U.S. Geological Survey. 2020. Available online: https://www.usgs.gov/centers/eros/science/usgs-eros-archive-landsat-archives-landsat-7-etm-plus-collection-2-level-2 (accessed on 18 February 2022).
  28. Earth Resources Observation and Science (EROS) Center. Landsat 8–9 Operational Land Imager/Thermal Infrared Sensor Level-2, Collection 2 [Dataset]. U.S. Geological Survey. 2020. Available online: https://www.usgs.gov/centers/eros/science/usgs-eros-archive-landsat-archives-landsat-8-9-olitirs-collection-2-level-2 (accessed on 18 February 2022).
  29. European Environment Agency. CORINE Land Cover 2018 (Raster 100 m), Europe, 6-Yearly, Version 2020_20u1. Release Date: 24 February 2020. 2018. Available online: https://land.copernicus.eu/pan-european/corine-land-cover/clc2018 (accessed on 25 May 2024).
  30. Gallardo, M.; Cocero, D. Using the European CORINE Land Cover Database: A 2011–2021 Specific Review. In Sustainable Development Goals in Europe. Key Challenges in Geography; De Lázaro Torres, M.L., De Miguel González, R., Eds.; Springer: Cham, Switzerland, 2023; pp. 303–325. [Google Scholar] [CrossRef]
  31. Jucker Riva, M.; Daliakopoulos, I.N.; Eckert, S.; Hodel, E.; Liniger, H. Assessment of land degradation in Mediterranean forests and grazing lands using a landscape unit approach and the normalized difference vegetation index. Appl. Geogr. 2017, 86, 8–21. [Google Scholar] [CrossRef]
  32. Silva, V.; Catry, F.X.; Fernandes, P.M.; Rego, F.C.; Paes, P.; Nunes, L.; Caperta, A.D.; Sérgio, C.; Bugalho, M.N. Effects of grazing on plant composition, conservation status and ecosystem services of Natura 2000 shrub-grassland habitat types. Biodivers. Conserv. 2019, 28, 1205–1224. [Google Scholar] [CrossRef]
  33. Georgiadis, N.M.; Dimitropoulos, G.; Avanidou, K.; Bebeli, P.; Bergmeier, E.; Dervisoglou, S.; Dimopoulos, T.; Grigoropoulou, D.; Hadjigeorgiou, I.; Kairis, O.; et al. Farming practices and biodiversity: Evidence from a Mediterranean semi-extensive system on the island of Lemnos (North Aegean, Greece). J. Environ. Manag. 2022, 303, 114131. [Google Scholar] [CrossRef] [PubMed]
  34. Bergmeier, E.; Ristow, M.; Krause, J.; Meyer, S.; Panitsa, M. Phytodiversity of Limnos (North Aegean, Greece)—An update and evaluation. Flora Mediterr. 2021, 31, 233–246. [Google Scholar] [CrossRef]
  35. Fienitz, M. Taking Stock of Land Use Conflict Research: A Systematic Map with Special Focus on Conceptual Approaches. Soc. Nat. Resour. 2023, 36, 715–732. [Google Scholar] [CrossRef]
  36. Troumbis, A.Y.; Gaganis, C.M.; Sideropoulos, H. Probabilistic Wildfire Risk Assessment and Modernization Transitions: The Case of Greece. Fire 2023, 6, 158. [Google Scholar] [CrossRef]
  37. Burkinshaw, A.M.; Bork, E.W. Shrub encroachment impacts the potential for multiple use conflicts on public land. Environ. Manag. 2009, 44, 493–504. [Google Scholar] [CrossRef]
  38. Alday, J.G.; O’Reilly, J.; Rose, R.J.; Marrs, R.H. Long-term effects of sheep-grazing and its removal on vegetation dynamics of British upland grasslands and moorlands; local management cannot overcome large-scale trends. Ecol. Indic. 2022, 139, 108878. [Google Scholar] [CrossRef]
  39. Applestein, C.; Germino, M.J. Detecting shrub recovery in sagebrush steppe: Comparing Landsat-derived maps with field data on historical wildfires. Fire Ecol. 2021, 17, 5. [Google Scholar] [CrossRef]
  40. Bright, B.C.; Hudak, A.T.; Kennedy, R.E.; Braaten, J.D.; Henareh Khalyani, A. Examining post-fire vegetation recovery with Landsat time series analysis in three western North American forest types. Fire Ecol. 2019, 15, 8. [Google Scholar] [CrossRef]
  41. Katagis, T.; Gitas, I.Z.; Toukiloglou, P.; Veraverbeke, S.; Goossens, R. Trend analysis of medium- and coarse-resolution time series image data for burned area mapping in a Mediterranean ecosystem. Int. J. Wildland Fire 2014, 23, 668–677. [Google Scholar] [CrossRef]
  42. Xu, L.; Herold, M.; Tsendbazar, N.-E.; Masiliūnas, D.; Li, L.; Lesiv, M.; Fritz, S.; Verbesselt, J. Time series analysis for global land cover change monitoring: A comparison across sensors. Remote Sens. Environ. 2022, 271, 112905. [Google Scholar] [CrossRef]
  43. Lee, S.-H.; Lee, M.-H.; Kang, T.-H.; Cho, H.-R.; Yun, H.-S.; Lee, S.-J. Comparative Analysis of dNBR, dNDVI, SVM Kernels, and ISODATA for Wildfire-Burned Area Mapping Using Sentinel-2 Imagery. Remote Sens. 2025, 17, 2196. [Google Scholar] [CrossRef]
  44. Kurbanov, E.; Vorobyev, O.; Leznin, S.; Polevshikova, Y.; Demisheva, E. Assessment of burn severity in Middle Povozhje with Landsat multitemporal data. Int. J. Wildland Fire 2017, 26, 772–782. [Google Scholar] [CrossRef]
  45. McKenna, P.; Phinn, S.; Erskine, P.D. Fire severity and vegetation recovery on mine site rehabilitation using worldview-3 imagery. Fire 2018, 1, 22. [Google Scholar] [CrossRef]
  46. Liu, S.; Zheng, Y.; Dalponte, M.; Tong, X. A novel fire index-based burned area change detection approach using Landsat-8 OLI data. Eur. J. Remote Sens. 2020, 53, 104–112. [Google Scholar] [CrossRef]
  47. Henkin, Z. Cattle grazing and vegetation management for multiple use of Mediterranean shrubland in Israel. Isr. J. Ecol. Evol. 2011, 57, 43–51. [Google Scholar] [CrossRef]
Figure 1. Landsat 5 natural color (RGB:3,2,1) satellite image of Lemnos Island dated 10 June 1984. This is the oldest image in the dataset.
Figure 1. Landsat 5 natural color (RGB:3,2,1) satellite image of Lemnos Island dated 10 June 1984. This is the oldest image in the dataset.
Fire 08 00335 g001
Figure 2. The study area of Lemnos Island (blue pointer) and the overview area in the North Aegean Sea of Greece, North-East Mediterranean. The study area is included in the footprint (red square) of the Landsat time series with WRS2 182/032 Path/Row. All images from satellites L5, L7, and L8 were freely available and downloaded from USGS https://earthexplorer.usgs.gov (accessed on 18 February 2022).
Figure 2. The study area of Lemnos Island (blue pointer) and the overview area in the North Aegean Sea of Greece, North-East Mediterranean. The study area is included in the footprint (red square) of the Landsat time series with WRS2 182/032 Path/Row. All images from satellites L5, L7, and L8 were freely available and downloaded from USGS https://earthexplorer.usgs.gov (accessed on 18 February 2022).
Fire 08 00335 g002
Figure 3. Distribution per year of the 420 Landsat images from the satellites L5, L7, and L8. From 1999 and earlier, images are available from two satellites, and in recent years, more images are available, improving the statistical analysis.
Figure 3. Distribution per year of the 420 Landsat images from the satellites L5, L7, and L8. From 1999 and earlier, images are available from two satellites, and in recent years, more images are available, improving the statistical analysis.
Fire 08 00335 g003
Figure 4. CORINE Land Cover 2018 (CLC) for the study area of Lemnos Island. CLC of Lemnos, in an area of 478.8 km2, includes 18 classes of Level 3 that show the complex landscape that covers the surface of the island and the richness of its ecosystems [29].
Figure 4. CORINE Land Cover 2018 (CLC) for the study area of Lemnos Island. CLC of Lemnos, in an area of 478.8 km2, includes 18 classes of Level 3 that show the complex landscape that covers the surface of the island and the richness of its ecosystems [29].
Fire 08 00335 g004
Figure 5. Framework for detecting significant land cover changes and fire events, by monitoring long-term breakpoints using Landsat images, NDVI index, and BFAST algorithm.
Figure 5. Framework for detecting significant land cover changes and fire events, by monitoring long-term breakpoints using Landsat images, NDVI index, and BFAST algorithm.
Fire 08 00335 g005
Figure 6. BFAST analysis of NDVI values (period 1984–2021) from a sample pixel of thyme habitat of Lemnos Island.
Figure 6. BFAST analysis of NDVI values (period 1984–2021) from a sample pixel of thyme habitat of Lemnos Island.
Fire 08 00335 g006
Figure 7. (ad) BFAST analysis (response, fitted, and trend) of four sample pixel NDVI time series from Lemnos Island for the period 1984–2021 (left). (e) Map of NDVI values of Lemnos Island for four sample years (1990, 2000, 2010, and 2020) (right).
Figure 7. (ad) BFAST analysis (response, fitted, and trend) of four sample pixel NDVI time series from Lemnos Island for the period 1984–2021 (left). (e) Map of NDVI values of Lemnos Island for four sample years (1990, 2000, 2010, and 2020) (right).
Fire 08 00335 g007
Figure 8. Map of the breakpoints on the forest and semi-natural areas of Lemnos Island estimated using the BFAST01 algorithm. The significant changes are detected in different years for the study period 1984–2021. Year distribution is shown with color palette.
Figure 8. Map of the breakpoints on the forest and semi-natural areas of Lemnos Island estimated using the BFAST01 algorithm. The significant changes are detected in different years for the study period 1984–2021. Year distribution is shown with color palette.
Fire 08 00335 g008
Figure 9. Detection of the largest fire event in the forest and semi-natural areas of Lemnos that burned 1.50 km2 from 3 August 2015 to 4 August 2015 in the Myrina area, southwest of the island: (a) Landsat 8 satellite images (false color, RGB:7,5,4) of SW of Lemnos before (3 August 2015) and after (19 August 2015) the fire event; (b) NDVI images zooming in on the burned area, respectively; (c) area of significant change detection estimated by the BFAST algorithm (red color) overlaying the dNDVI area (grayscale color) used as background layer.
Figure 9. Detection of the largest fire event in the forest and semi-natural areas of Lemnos that burned 1.50 km2 from 3 August 2015 to 4 August 2015 in the Myrina area, southwest of the island: (a) Landsat 8 satellite images (false color, RGB:7,5,4) of SW of Lemnos before (3 August 2015) and after (19 August 2015) the fire event; (b) NDVI images zooming in on the burned area, respectively; (c) area of significant change detection estimated by the BFAST algorithm (red color) overlaying the dNDVI area (grayscale color) used as background layer.
Fire 08 00335 g009
Table 1. Burnt area of Lemnos Island monthly distributed per land cover type for the period 1999–2021 (measured in km2).
Table 1. Burnt area of Lemnos Island monthly distributed per land cover type for the period 1999–2021 (measured in km2).
Land Cover TypeMonthTotal
JanFebMarAprMayJunJulAugSepOctNovDec
Forest and semi-natural areas0.2430.0760.0270.0680.0670.0440.1201.9660.0440.1331.1990.6434.628
Agricultural areas0.0010.0040.0170.0010.0050.0800.2992.9220.1390.2950.3670.0024.131
Wetlands0.0120.0710.0150.0470.0090.0150.0000.0270.0220.1780.4810.1180.993
Artificial surfaces0.0030.0020.0020.0040.0010.0070.0020.0070.0010.0250.0170.0000.069
Total0.2580.1530.0590.1190.0820.1460.4214.9220.2060.6302.0620.7639.821
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Vasios, G.K.; Alexoudaki, E.; Kaloveloni, A.; Troumbis, A.Y. Landsat Time Series Analysis with BFAST for Detecting Degradation of Thyme Shrublands by Fire on Lemnos Island. Fire 2025, 8, 335. https://doi.org/10.3390/fire8080335

AMA Style

Vasios GK, Alexoudaki E, Kaloveloni A, Troumbis AY. Landsat Time Series Analysis with BFAST for Detecting Degradation of Thyme Shrublands by Fire on Lemnos Island. Fire. 2025; 8(8):335. https://doi.org/10.3390/fire8080335

Chicago/Turabian Style

Vasios, Georgios K., Eleftheria Alexoudaki, Aggeliki Kaloveloni, and Andreas Y. Troumbis. 2025. "Landsat Time Series Analysis with BFAST for Detecting Degradation of Thyme Shrublands by Fire on Lemnos Island" Fire 8, no. 8: 335. https://doi.org/10.3390/fire8080335

APA Style

Vasios, G. K., Alexoudaki, E., Kaloveloni, A., & Troumbis, A. Y. (2025). Landsat Time Series Analysis with BFAST for Detecting Degradation of Thyme Shrublands by Fire on Lemnos Island. Fire, 8(8), 335. https://doi.org/10.3390/fire8080335

Article Metrics

Back to TopTop