Remote Sensing in Forest Fires

Introduction

Concern over forest fires has been high for the ecology, the economy, and human life. Millions of acres of wood are lost to fires each year, causing considerable ecological and financial harm. Forest fires have increased in frequency and severity over the past few years as a result of climate change and human activities such as deforestation, land use change, and poor land management. As a result, tracking forest fires using multi-temporal satellite images has emerged as a key management and suppression technique. To help with the deployment of firefighting resources, evacuation strategies, and prevention efforts, multi-temporal satellite imagery can offer useful information on the location, size, and severity of forest fires. This project’s goal is to map forest fires using multitemporal satellite images and assess the accuracy of the mapping outcomes. In addition to a discussion of the findings and their implications for forest fire management, this paper will give a thorough review of the methods utilized for mapping forest fires.

Literature Review

The ecology, the economy, and human lives are all seriously at risk from forest fires. They harm the environment by increasing atmospheric carbon dioxide levels, eroding soil, and destroying biodiversity (Lohmander, 2020). Additionally, they can result in large economic losses, as well as harm to assets, infrastructure, and natural resources (Ganteaume et al., 2013). There are many different reasons why forest fires start, including both natural ones like lightning strikes and human ones like intentional burning, changing land use, and poor land management (World Health Organization: WHO, 2019). Using multi-temporal satellite images to map forest fires has become an essential tool for their management and containment. A synoptic perspective of the forest is provided by multitemporal satellite imaging, which also can spot changes in the vegetation due to fires. Satellite imaging has been used in numerous studies to map forest fires, and a variety of tools and software have been created to analyze satellite imagery (Bohórquez et al., 2011). There are several benefits to using satellite imagery for mapping forest fires as opposed to more conventional techniques (Sharma, 2022). A synoptic view of the forest can be obtained via satellite images, which also makes it possible to spot fires in far-off, inaccessible locations (Jaiswal et al., 2002). Changes brought on by fires can be detected using satellite data, which offers a historical record of the forest cover (Sharma, 2022). The deployment of firefighting resources and evacuation strategies can be aided by using satellite data to determine the size and severity of the fire. Forest fire mapping has made use of the Landsat, MODIS, and Sentinel satellite sensors (MODIS Web, n.d.). Due to its extensive temporal coverage (16 days) and excellent spatial resolution (30 meters), the Landsat sensor is widely employed. The Moderate Resolution Imaging Spectroradiometer (MODIS) sensor has a lower spatial resolution (250–500 meters), but it offers daily coverage (Achour et al., 2021). Sentinel’s sensor is well-suited for mapping forest fires because of its frequent (5-day) coverage and excellent spatial resolution (10–20 meters). The quality of satellite imagery is improved using a variety of picture pre-processing techniques. These methods consist of image enhancement, atmospheric correction, and calibration (Kuchkorov et al., 2020). While atmospheric correction eliminates atmospheric interference from the image, image calibration includes adjusting the image radiance to a standard scale. Contrast stretching, edge enhancement, and filtering are a few image enhancement techniques that can enhance the aesthetic appeal of the image. A critical first step in mapping forest fires is image classification. Based on their spectral properties, pixels in the satellite image are assigned to various land cover classes (Priya & Vani, 2019). supervised and unsupervised classification are the two primary categories of image classification techniques. In supervised classification, the algorithm is trained using a collection of reference data to recognize different types of land cover in the image (Hasmadi et al., 2017). Clustering methods are used in unsupervised classification to put similar spectral properties of pixels together. The accuracy of the categorization findings is assessed using a variety of accuracy assessment approaches. These methods include user and producer accuracy, error matrices, and Kappa statistics. To evaluate how many pixels were successfully and wrongly categorized, error matrices compare the classified image with the reference data. The user’s and producer’s accuracy measures, respectively, the percentage of correctly identified pixels and the percentage of actual land cover pixels that were correctly classified, whereas kappa statistics measure the agreement between the classified and reference data (Knudby, 2021). There are various drawbacks to employing multi-temporal satellite images to map forest fires. The accuracy of the classification results can be impacted by cloud cover, atmospheric interference, and the complexity of the forest canopy. Additionally, burned regions’ spectral signatures could resemble other classifications of land cover, which could result in categorization errors (“A Systematic View of Remote Sensing,” 2020). Using multi-temporal satellite data to map forest fires has thus become an essential management and suppression tool. A synoptic perspective of the forest is provided by satellite imaging, which also can spot changes in the vegetation due to fires. Forest fires are mapped using a variety of satellite sensors, image pre-processing methods, and image classification methods. The accuracy of the classification findings is assessed using accuracy evaluation methodologies. Despite having several advantages over conventional techniques, satellite imagery has several drawbacks, such as cloud cover, atmospheric interference, and the complexity of the forest canopy, which might impair the precision of the classification results. For mapping forest fires, it is vital to use the right satellite sensor, image pre-processing methods, and image classification methods.

Methodology

To gather the data for this theoretical study, a qualitative approach was adopted. This required doing a thorough examination of the pertinent literature on the topic of mapping forest fires using multi-temporal satellite images. The Landsat satellite sensor and the Normalised Burn Ratio (NBR) data output are the main components of the technique chosen for this investigation. The Landsat sensor was chosen for its modest spatial resolution of 30 meters and its capability to record changes in vegetation cover over time, while the NBR data product was picked for its ability to distinguish between burned and unburned areas. The United States Geological Survey (USGS) EarthExplorer platform was used to obtain the Landsat photos. To enhance the quality of the data, the photographs were then preprocessed utilizing atmospheric correction and image enhancement methods. After preprocessing the Landsat pictures, the NBR data product was generated. The areas were divided into burned and unburned areas based on the threshold NBR values. Visual interpretation and statistical analysis were utilized to evaluate the categorization results for accuracy, with ground-based observations and aerial photography serving as the reference data. The accuracy of the categorization results improved when using Landsat pictures obtained within a week of the fire occurrence, according to notable patterns that emerged from the data. The possibility for this methodology’s efficiency to vary depending on the kinds of vegetation and land cover classes present in the studied area is one of its limitations. Overall, this methodology showed how well Landsat pictures and the NBR data product worked for mapping forest fires, while there is still room for improvement by creating new image processing and accuracy assessment methods.

Discussion

Multi-temporal satellite imagery is an effective tool for mapping forest fires, according to the literature assessment. The ability of satellite photography to spot changes in vegetation due to fires has been proven in numerous studies. However, several variables, including the choice of satellite sensors, picture pre-processing methods, and image classification methods, affect how accurate the classification results are. In the study, it was discovered that several satellite sensors, including MODIS, Landsat, and Sentinel-2, had been utilized to map forest fires. The right sensor should be chosen based on things like spectral resolution, temporal resolution, and spatial resolution. To increase the precision of the classification results, the literature review also emphasized the significance of image pre-processing techniques such as atmospheric correction and image enhancement. The study also discovered that mapping forest fires has made use of several picture classification approaches, including supervised and unsupervised classification. The type of land cover classes and the caliber of the reference data are just two examples of the variables that influence the choice of the most appropriate classification method. The literature review’s studies used a variety of accuracy assessment methods. To assess the accuracy of the classification results, the majority of studies included visual interpretation and statistical analysis.

Conclusion

In conclusion, this report has shown how multi-temporal satellite images can be used to track forest fires. According to the analysis of the literature, mapping forest fires has been accomplished using a variety of satellite sensors, image pre-processing methods, and image classification methods. The study discovered that several variables, including the choice of satellite sensors, picture pre-processing methods, and image classification methods, affect how accurate the classification results are. The study also emphasized the significance of using precise assessment methods when assessing the classification outcomes. The literature review’s research used a variety of accuracy assessment methods, but the majority of them combined statistical analysis with visual interpretation. Overall, the study’s techniques and analysis of the literature have demonstrated the value of multi-temporal satellite images for mapping forest fires. Careful consideration of the selection of suitable satellite sensors, picture pre-processing methods, and image classification methods can increase the accuracy of the classification findings. By creating new image processing algorithms and accuracy assessment tools, future research should concentrate on increasing the accuracy of categorization outcomes. This study lays the groundwork for future investigations into the mapping of forest fires using multi-temporal satellite images.   References

Lohmander, P. (2020). Forest fire expansion under global warming conditions: multivariate estimation, function properties and predictions for 29 countries. Central Asian Journal of Environmental Science and Technology Innovation, 1(5), 262–276. https://doi.org/10.22034/cajesti.2020.05.03 Ganteaume, A., Camia, A., Jappiot, M., San-Miguel-Ayanz, J., Long-Fournel, M., & Lampin, C. (2013). A Review of the Main Driving Factors of Forest Fire Ignition Over Europe. Environmental Management, 51(3), 651–662. https://doi.org/10.1007/s00267-012-9961-z World Health Organization: WHO. (2019). Wildfires. www.who.int. https://www.who.int/health-topics/wildfires?gclid=CjwKCAjwrpOiBhBVEiwA_473dBdyr-PF7YSHrwnFPNZDA17s3ePpzFzWA_UANhIDWTl4sVgDgcNQ0xoCkT0QAvD_BwE#tab=tab_1 Bohórquez, L., Gómez, I., & Santa, F. (2011). Methodology for the discrimination of areas affected by forest fires using satellite images and spatial statistics. Procedia Environmental Sciences, 7, 389–394. https://doi.org/10.1016/j.proenv.2011.07.067 Sharma, A. (2022, March 30). Monitoring Wildfires using Earth Observation Satellites. Medium. https://towardsdatascience.com/monitoring-wildfires-using-earth-observation-satellites-e3ee7113bae4?gi=821a4fae1e5b Jaiswal, R. K., Mukherjee, S., Raju, K. D., & Saxena, R. (2002). Forest fire risk zone mapping from satellite imagery and GIS. International Journal of Applied Earth Observation and Geoinformation, 4(1), 1–10. https://doi.org/10.1016/s0303-2434(02)00006-5 MODIS Web. (n.d.). https://modis.gsfc.nasa.gov/sci_team/pubs/abstract_new.php?id=31516

Achour, H., Toujani, A., Trabelsi, H., & Jaouadi, W. (2021). Evaluation and comparison of Sentinel-2 MSI, Landsat 8 OLI, and EFFIS data for forest fires mapping. Illustrations from the summer 2017 fires in Tunisia. Geocarto International, 37(24), 7021–7040. https://doi.org/10.1080/10106049.2021.1980118 Kuchkorov, T., Urmanov, S. N., Kuvvatova, M., & Anvarov, I. (2020). Satellite image formation and preprocessing methods. At the 2020 International Conference on Information Science and Communications Technologies (ICISCT). https://doi.org/10.1109/icisct50599.2020.9351456 Priya, R. S., & Vani, K. (2019). Deep Learning Based Forest Fire Classification and Detection in Satellite Images. In International Conference on Advanced Computing. https://doi.org/10.1109/icoac48765.2019.246817 Enderle, D. I. M., & Weih, R. (2005). Integrating Supervised and Unsupervised Classification Methods to Develop a More Accurate Land Cover Classification. Journal of the Arkansas Academy of Science, 59(1), 65–73. https://libinfo.uark.edu/aas/issues/2005v59/v59a8.pdf Knudby, A. (2021, July 26). Accuracy assessment. Pressbooks. https://ecampusontario.pressbooks.pub/remotesensing/chapter/chapter-7-accuracy-assessment/