Fuel load analysis with Google Earth Engine (GEE) and Python Engine (PE) in forest fire management: Challenges and benefits

Muhammed Çetin

Makale detayları

Fuel load analysis with Google Earth Engine (GEE) and Python Engine (PE) in forest fire management: Challenges and benefits

Yazarlar:

Muhammed Çetin

Yayıncı:

Gumushane University

Ana Başlıklar:

Coğrafyada Ekoloji

Dergi:

Journal of Anatolian Geography Vol.2 Issue 1, pp.63-81

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001348425

pISSN:

3023-8978

Anahtar kelimeler:

Özet:

 Fuel load is one of the fundamental factors determining forest fire behavior. Dynamics such as fire intensity, flame length, and fuel consumption are directly related to this parameter. However, while traditional field measurements provide detailed data, repeating these processes over large areas leads to high costs and time loss. Remote sensing RS technologies overcome these limitations by enabling rapid and effective fuel load estimation over wide regions. In this study, the fuel load dynamics before and after the 2021 Manavgat megafire were analyzed using the cloud-based Google Earth Engine GEE platform and Sentinel-2 multispectral imagery. Maps of surface fuel load were generated in accordance with the Rothermel model by utilizing the Normalized Difference Vegetation Index NDVI and Normalized Burn Ratio NBR; median composites were created from time-series data with cloud coverage below 20. Descriptive statistics and meta-analysis techniques were applied using Python in the Google Colab environment to ensure data consistency. NDVI and NBR values at sampling points were classified into four groups to directly support fire behavior models. These values were grouped by percentile classification as Low  20, Moderate 2050, High 5080, and Very High  80, providing direct input to the Rothermel equation parameters. Additionally, to assess pre-fire drought dynamics, analyses of the Vegetation Condition Index VCI and the Vegetation Drought Severity Index SDVI were conducted. The results demonstrate that this integrated spatial approach offers strong predictive power for decision-makers in pre-fire intervention planning.

 Fuel load is one of the fundamental factors determining forest fire behavior. Dynamics such as fire intensity, flame length, and fuel consumption are directly related to this parameter. However, while traditional field measurements provide detailed data, repeating these processes over large areas leads to high costs and time loss. Remote sensing RS technologies overcome these limitations by enabling rapid and effective fuel load estimation over wide regions. In this study, the fuel load dynamics before and after the 2021 Manavgat megafire were analyzed using the cloud-based Google Earth Engine GEE platform and Sentinel-2 multispectral imagery. Maps of surface fuel load were generated in accordance with the Rothermel model by utilizing the Normalized Difference Vegetation Index NDVI and Normalized Burn Ratio NBR; median composites were created from time-series data with cloud coverage below 20. Descriptive statistics and meta-analysis techniques were applied using Python in the Google Colab environment to ensure data consistency. NDVI and NBR values at sampling points were classified into four groups to directly support fire behavior models. These values were grouped by percentile classification as Low  20, Moderate 2050, High 5080, and Very High  80, providing direct input to the Rothermel equation parameters. Additionally, to assess pre-fire drought dynamics, analyses of the Vegetation Condition Index VCI and the Vegetation Drought Severity Index SDVI were conducted. The results demonstrate that this integrated spatial approach offers strong predictive power for decision-makers in pre-fire intervention planning.

 Fuel load is one of the fundamental factors determining forest fire behavior. Dynamics such as fire intensity, flame length, and fuel consumption are directly related to this parameter. However, while traditional field measurements provide detailed data, repeating these processes over large areas leads to high costs and time loss. Remote sensing RS technologies overcome these limitations by enabling rapid and effective fuel load estimation over wide regions. In this study, the fuel load dynamics before and after the 2021 Manavgat megafire were analyzed using the cloud-based Google Earth Engine GEE platform and Sentinel-2 multispectral imagery. Maps of surface fuel load were generated in accordance with the Rothermel model by utilizing the Normalized Difference Vegetation Index NDVI and Normalized Burn Ratio NBR; median composites were created from time-series data with cloud coverage below 20. Descriptive statistics and meta-analysis techniques were applied using Python in the Google Colab environment to ensure data consistency. NDVI and NBR values at sampling points were classified into four groups to directly support fire behavior models. These values were grouped by percentile classification as Low  20, Moderate 2050, High 5080, and Very High  80, providing direct input to the Rothermel equation parameters. Additionally, to assess pre-fire drought dynamics, analyses of the Vegetation Condition Index VCI and the Vegetation Drought Severity Index SDVI were conducted. The results demonstrate that this integrated spatial approach offers strong predictive power for decision-makers in pre-fire intervention planning.

 Yakıt yükü, orman yangını davranışını belirleyen temel unsurlardan biridir. Yangının şiddeti, alev uzunluğu ve yakıt tüketimi gibi dinamikler doğrudan bu parametreye bağlıdır. Ancak geleneksel saha ölçümleri büyük alanlarda tekrarlandığında yüksek maliyet ve zaman kaybı yaratmaktadır. Uzaktan algılama UA teknolojileri bu sınırlamaları aşarak geniş alanlarda hızlı ve etkin yakıt yükü tahmini imkânı sunmaktadır. Bu çalışmada, 2021de Manavgatta meydana gelen mega yangın örneği üzerinden Google Earth Engine GEE platformu ve Sentinel-2 multispektral görüntüleri kullanılarak yangın öncesi ve sonrası yakıt yükü davranışı analiz edilmiştir. Normalize Fark Bitki Örtüsü NDVI ve Normalize Yanma Oranı NBR indeksleri kullanılarak, Rothermel modeline uygun yüzey yakıt yükü haritaları oluşturulmuş; bulut örtüsü 20nin altında olan görüntülerden zaman serisi bazlı medyan kompozitler elde edilmiştir. Google Colab ortamında Python diliyle yürütülen analizlerde betimleyici istatistikler ve meta-analiz yöntemleriyle veri tutarlılığı sağlanmıştır. Örnekleme yöntemiyle belirlenen noktalardaki NDVI ve NBR değerleri dört sınıfa ayrılmış ve yangın davranışı modellerine doğrudan veri sağlamıştır. NDVI ve NBR değerleri, persentile sınıflandırmasıyla Düşük  20, Orta 2050, Yüksek 5080 ve Çok Yüksek  80 olarak gruplandırılmış; bu ayrım Rothermel denkleminin parametrelerine doğrudan girdi sağlamıştır. Ayrıca yangın öncesi kuraklık dinamiklerini incelemek amacıyla Bitki Örtüsü Durumu Endeksi VCI ve Bitki Örtüsü Kuraklık Şiddeti İndeksi SDVI analizleri yapılmıştır. Sonuçlar, bu mekânsal yaklaşımın yangın öncesi müdahale planlamasında karar vericilere güçlü bir öngörü sunduğunu göstermektedir.

Makale detayları

Fuel load analysis with Google Earth Engine (GEE) and Python Engine (PE) in forest fire management: Challenges and benefits

Yazarlar:

Muhammed Çetin

Yayıncı:

Gumushane University

Ana Başlıklar:

Coğrafyada Ekoloji

Dergi:

Journal of Anatolian Geography Vol.2 Issue 1, pp.63-81

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001348425

pISSN:

3023-8978

Forest

Ecology

GIS

GEE

Fuel Load MapEkoloji

Orman

Coğrafi Bilgi Sistemleri

Google Earth Engine

Yakın Yükü Haritası

Sayı

Makale detayları

Yazarlar: Muhammed Çetin

Yayıncı: Gumushane University

Ana Başlıklar: Coğrafyada Ekoloji

Dergi: Journal of Anatolian Geography Vol.2 Issue 1, pp.63-81

Belge Tipi: Makale

Yayın Tarihi: 6/2025

Belge Dili: Türkçe

Document ID (Org No.): 2001348425

pISSN: 3023-8978

Sayı

Yazarlar:

Muhammed Çetin

Yayıncı:

Gumushane University

Ana Başlıklar:

Coğrafyada Ekoloji

Dergi:

Journal of Anatolian Geography Vol.2 Issue 1, pp.63-81

Belge Tipi:

Makale

Yayın Tarihi:

6/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001348425

pISSN:

3023-8978