Fire Hazard Zoning in National Golestan Park Using Logistic Regression and GIS

جغرافیا و مخاطرات محیطی

Article 3, Volume 3, Issue 2 - Serial Number 10, January 2016, Pages 73-90   

Document Type: Research Article

DOI: 10.22067/geo.v3i2.26733

Authors

Hasan Faramarzi; Seyed Mohsen Hosseini; Mehdi Gholamalifard^*

Tarbiat Modares University

Abstract

1. Introduction
Extensive fires in the forests are examples of natural crises (Hosseinali, 2005) and forest fire events are increasing in the conditions of changing climate and global warming; (Azizi and Yousefi, 2005). This phenomenon is considered one of the most destructive factors for forest ecosystems, causing irreparable damages (Marozas et al, 2007). Some of these damages include land use changes, emissions of greenhouse gases, disruption of forest structure and nutrients losses in ecosystems resulting from burning of vegetation and forest floor layers (Vakalis et al,2007; Alexandridis et al, 2008; Bakhshandehsavad rudbari et al, 2011). According to FAO (1995) the annual average of fires in forested areas around the world is estimated to be two million hectares. In many parts of the world, fires in forests and ranges is one of the most important issues and concerns, not only from an environmental perspective, but also from economic, social and security viewpoints (Silvia Merino and Gonzale, 2010). One of the research fields for controlling forest fire is to identify critical points in terms of fire in forested areas, because the lack of knowledge about these areas will cause the occurrence and spreading of forest fire, delay in fire suppression, and damage to forest plant and animal life (Jaiswal et al, 2002). Therefore, preparing the potential map of fire (fire critical areas) plays an important role in reducing the number of fires and preventing the destruction of forests (Dong et al, 2005) and helps the forest managers to prevent fires with special care in critical areas before they occur. According to a study by Atrak Chali (2000) the average area in forest areas that is burned due to fires is 7,000 ha and return intervals of forest fire in Golestan province varies from 5 to 7 years; therefore preparing fire hazard maps with high-precision is necessary for planning and crisis management. When, preparing the maps of fire risk, GIS is used as a basic tool for spatial data management. In many studies, GIS have been used to identify critical points and potential techniques have been applied. Mohammadi et al (2009) attempted to map the fire susceptible areas in Paveh forests fires by conducting field operations and by considering the measures of vegetation, physiography, climate, human and distance from roads and streams as well as by applying the analytic hierarchy process (AHP) and they showed that the resulting map was highly corresponding with the real locations of fire. In a study in Spain, Nieto et al (2012) evaluated the occurrence of fire caused by lightning during a 3-year period from 2004 to 2002 using logistic regression. In this study some parameters such as lightning, topography, vegetation and climatic parameters were considered. The results showed that the lightning was most important factor causing fire in the area fires during mentioned period and a risk model with ROC of 0.7 obtained, indicating the good performance of the model.
Golestan National Park is First Park that was considered as the national park in Iran and one-eighth of plant species, one-third of bird species and more than 50% of the mammal species live in this park (Hassanzade Kiyabi et al, 1993). This park is susceptible to fire due to its characteristics vegetation as well as occurring at a region with winds formed by the clash between wet and dry weather fronts (Shokri et al, 2002).
This study models and identifies areas susceptible to fire in Golestan National Park using logistic regression, taking into account parameters influencing on the fire in the region.
2. Study Area
Golestan National Park is located at the Northeastern part of Iran, East of Golestan province, Northwest of Khorasan province, and North of Semnan province between 37º 17´ 43´´ to 37º 31´ 35´´ N, and 55º 43´ 25´´ to 56º 17´ 48´´ E. The park area is 91,895 ha with a perimeter of 198 km. A transit road, known as Asian road that connect the northern and central Iran to the north-eastern Iran passes through the park. There are several villages around the park including Tangeh rah, Terjenli, Ghoch Cheshmeh, Zav, Tomak, Kondosku and Dasht Shad in West of the park, Dasht, CheshmehKhan and Armadlu in South of the park, Robat Ghahrahbil in East of the park and Lohondor, Yelcheshmeh and Behkade in North of the park.
3. Material and Methods
In this study we first determined parameters and their influences on fire event, then information related to these parameters was collected from the related organizations as well as by inventory in the study area. Affecting parameters were prepared in the ArcGIS and IDRISI software packages. Based on topographic map, Elevation ranges from 467 to 2342 m above sea level in the study area. slope from 0 to 72 degrees and slope aspect maps with 9 classes were extracted from a Digital Elevation Model (DEM). Normalized Difference Vegetation Index (NDVI) map was extracted from MODIS satellite images of 2011. Humidity and temperature isopleths maps were obtained from interpolation of meteorological data 2001-2011 years of climatology and synoptic stations around the park based on Inverse Distance Weighting (IDW). Considering the effect of the presence of hunters and ranchers on fire events in the park, the distance from arresting positions of offenders in 2007-2012 as was mapped the higher probable presences in the park. Also a buffer zone was established from camps as tourist sites, villages around the park, transit road passing through the park. In the study area, 9 vegetation classes were identified including dense-canopy forest, low-canopy forest, average-canopy forest, very low-canopy forest, good pasture, medium-vegetated pasture, woodlands, bushy vegetation, and agricultural land. The occurred fires during the last 30 years was determined based on records of Central Office of Park, information provided by the guards as well as signs of the fire using GPS, then the Boolean fire maps was drown in Arc GIS software.
After selection the parameters, fire hazard map prepared using logistic regression method in the IDRISI software. Logistic regression is run as computational binomial regression in which the dependent variables of nature are entered binary.
The uses model accessory was assist with Relative operating Characteristic (ROC) and Pseudo-R2 statistics. Finally, sensitivity of variables was examined by removing them using ROC and Pseudo-R2.
4. Results and Discussion
The regression equation resulting from the model was calculated as follows:
Logit (fire) = 38.792 + 0.05475*Aspect + 0.79038*Tem - 0.000666*Camping + 0.0014*Elevation - 0.00037*Shepherds + 0.00036*Hunter + 0.0081*Landuse -0.7874*Moisture + 0.0011*NDVI + 0.0004*Road - 0.051747*Slope - 0.00006*Village
In this model, variables such as aspect, elevation, mean temperature (tem), distance from road, and presence of hunter, land use and NDVI are positively correlated with fire risk while parameters of slope, camping, presence of ranchers, moisture and distance from village are negatively correlated with fire risk.
The ROC, Pseudo-R2 and Chi-square were calculated 0.9267, 0.3133 and 102227.3 in the model, respectively. The fire risk map was prepared using logistic regression.
The results of this study both from the obtained equation and from measuring the sensitivity of variables indicate that temperature and humidity are affecting on the occurrence of regional fire. Since the ROC statistic of 0.7 indicates low accuracy, 0.7-0.9 value indicate the practicability and a value more than 0.9 represents the high accuracy of model and Pesudo-R2 values greater than 0.2 can be considered as a relatively good fitness, in this study ROC Pesudo-R2 values were found 0.9264 and 0.3133, respectively, indicating the high accuracy and good fitness for the fire risk model. Therefore logistic regression can be introduced as an appropriate method to classify the fire risk. Understanding the behavior of forest fire and factors causing the appropriate environment for fire occurrence and influencing on fire behavior are of high importance for fire forest management, thus this model can be useful for designing of fire management and taking preventive measures in high fire risk areas.
5. Conclusion
Considering the location of area, high temperatures and low humidity that provides favorable conditions for fires, and also given the high traffic of travelers, tourists, ranchers and hunters allowed in the park, the results showed that fire risk is in this park. Due to the high accuracy obtained from the logistic regression model in this study, one can claim that this method is appropriate for the evaluation and modeling of fire risk thus the resulting model can be used for preventive measures.

Keywords

Risk Fires, Regression Logistic; Modeling; Crises Management; Golestan National Park

References

اترک چالی، عبدالناصر. 1379. بررسی اثرات آتش سوزی بر روی تغییرات پوشش گیاهی. پایان نامه کارشناسی ارشد، دانشگاه مازندران، 85 ص.

اکبری، داوود.، امینی، جلال.، سعادت سرشت، محمد.، 1386. ارائه‌ی مدلی ساده و سریع برای تهیه‌ی نقشه‌ی ریسک آتش‏سوزی در نواحی جنگلی، مجموعه‌ی مقالات دومین همایش مقابله با سوانح طبیعی، دانشگاه تهران، تهران:1151- 1157.

بخشنده سوادرودباری، مریم، مروی مهاجر، محمدرضا،. 1390. بررسی تغییرات پوشش علفی جنگل بعد از آتش‏سوزی، مجموعه مقالات نخستین همایش بین المللی در عرصه های منابع طبیعی در دانشگاه علوم کشاورزی و منابع طبیعی گرگان، 4 و 5 آبان 1390: 272 ص.

حسن زاده کیابی، بهرام.، زهزاد، بهرام.، فرهنگ دره شوری، بیژن.، مجنونیان، هنریک.، گشتاسب میگو ئی، حمید.،1372. پارک ملی گلستان.، انتشارات سازمان محیط زیست، چاپ اول: 203 صفحه.

حسینعلی، فرهاد.، رجبی، محمد علی.، 1384. شبیه سازی آتش سوزی جنگل با استفاده از سامانه اطلاعات مکانی، همایش ژئومانیک 84، 18 اردبیهشت 84: 11 ص.

جزیره ای، محمد حسین،. 1390. نگهداشت جنگل، انتشار دانشگاه تهران:231 ص.

رهبر، داود،. 1379. اثرات زیست محیطی صنعت گردشگری، توسعه مدیریت شماره 19: 4 ص.

شکری، مریم،. صفائیان، نصرت، الله،. اترک چالی، عبدالناصر ،. 1381. بررسی پیامد آتش بر پوشش گیاهی تختی ییلاق پارک ملی گلستان، مجله منابع طبیعی ایران، جلد 55 شماره 2: 281-273.

عزیزی، قاسم.، یوسفی، یدالله.،1384. گرمباد (بادفون) و آتش‏سوزی جنگل در استان‌های مازندران و گیلان، فصلنامه‌ی تحقیقات جغرافیایی 92(749): 15434- 15407.

محمدی، فریده.، شعبانیان، نقی.، پور هاشمی، مهدی.، فاتحی، پرویز.، 1388. تهیه‌ی نقشه‌ی خطر آتش‏سوزی با استفاده از GIS و AHP در بخشی از جنگل‌های پاوه، فصلنامه‌ی علمی ـ پژوهشی تحقیقات جنگل و صنوبر ایران 18 (4):569-586.

مجنونیان هنریک،. زهزاد، بهرام،. کیابی، بهرام،. فرهنگ دره شوری، بهزاد،. گشتاسب میگوئی، حمید،. 1378. شناسنامه پارک ملی گلستان ذخیره گاه زیست کره، انتشارات سازمان محیط زیست، 129 ص.

Akbari, D., Amini J,. SadatSeresht, M., 2007. Providing a simple model and quick for fire risk mapping in forested areas, Proceedings of the Second Conference on Natural Disaster Confronting, Tehran University, Tehran, 1151-1157.

Aldrich, J., and Forrest, D., 1984. Linear Probability, Logit and Probit Models. Series: Quantitative Applications in the Social Sciences 45. London: Sage.

Alexandridis, A., Vakalis, D., Siettos, C.I. and Bafas, G. V., 2008. A cellular automata model for forest fire spread prediction: The case of the wildfire that swept through Spetses Island in 1990. Applied Mathematics and Computation 204,191–201.

Amatulli Perez-Cabello, D., 2007. Mapping lightning/human-caused wildfires occurrence under ignition point location uncertainty. Ecological Modelling 200, 321–333.

AtrakChaly, A., 2000. Investigate the effects of fire on vegetation change. MA thesis, University of Mazandaran, 85 pp.

Azizi, Gh., Yusefi, Y., 2005. Sirocco (wind-Von) and forest fires in the Mazandaran and Gilan province, Journal of Geographical Research 92 (749): 15407-15434.

BakhshandeSavadRudbari, M,. MarviMohajer, MR., 2011. Changes in herbaceous cover after forest fires, Proceedings of the First International Conference on the areas of natural resources at the University of Agricultural Sciences and Natural Resources, Gorgan, 26-27 October, 2011: 272 p.

Bakirci, M,. 2010. Negative Impacts of Forest Fires on Ecological Balance and Environmental. Revija za geografijo - Journal for Geography, 5-1:15-32.

Balzter, H., Gerard, F., George, Ch., Rowland, C., Jupp, T., McCallum, I., Shivdenko, A., Nilsson, S., Sukhinin, A., Ounchin, A., Schmullius, Ch., 2005. Impact of the Arctic Oscillation pattern on interannual forest fire variability in Central Siberia, Geophysica Research Letters, vol. 32:4p.

Brooks, M.L., Matchett, J.R., 2006. Spatial and temporal patterns of wildfires in the Mojave Desert, 1980–2004. Journal of Arid Environments 67, 148–164.

Cantarello, E., Newton A., Hill R., Tejedor –Garavito, N.,Williams-linera, G., Lopez-Barrera, F,. H.manson R., J.Golicher, D., 2011. Simulating the potential for ecological restoration of dryland forests in Mexico under different disturbance regimes, Ecological Modelling 222, 1112–1128.

Chuvieco, E., Cocero D., Riano, D., Martin, P., Martinez-Vega, J., Riva, j., Perez, F., 2004. Combining NDVI and surface temperature for the estimation of live, fuel moisture content in forest fire danger rating, Remote Sensing of Environment 92:322–331.

Clark, W.A., Hosking, P.L., 1986. Statistical Methods for Geographers, John Wiley and Sons, New York, 528pp.

Dong, X.U., Li-min, D., Guo-fan, Sh., Lei, T., Hui, W., 2005. Forest fire risk zone mapping from satellite images and GIS for Baihe Forestry Bureau, Jilin, China. Journal of Forestry Research 16(3), 169-174.

Erten, E., Kurgun, V. ve Musaoglu, N., 2004. Forest Fire Risk Zone Mapping from Satellite Imagery and GIS: A Case Study, XX.th Congress of the International Society for Photogrammetry and Remote Sensing (ISPRS), July, Istanbul, Turkey,Proceedings,12-25.

FAO. 1995. International forest fire news. No 16.

Franklin, J., McCullough, P., Gray, C., 2000. Terrain variables used for predictive mapping of vegetation communities in Southern California. In ‘Terrain Analysis: Principles and Applications: pp. 331–353.

Fu, Z.Q., Wang, Y., Wang, C., 2001. Research on the spring fire-risk forecasting model in dry grasslands in Inner Mongolia. Quarterly Journal of Applied Meteorology 12, 202–209.

Gil Pontius, R., Schneider, L., 2001. Land-cover change model validation by an ROC method for the Ipswich watershed, Massachusetts, USA, Agriculture, Ecosystems and Environment 85:239–248.

HassanZadehKiyabi, B., Zahzad, B., FarhangDareShuri, B., Majnuniyan, H., GashtasbMigui, H., 1993. Golestan National Park, environmental organizations Publications, First Edition, 203p.

HosseinALi, F., Rajabi, M., 2005. Forest fire simulation using GIS, Geomatic Conference 84, 8 may 2005: 11 p.

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.

Jazirei, MH., 2011. Forest Conservation, Tehran University Press: 231 p.

Jobson, J.D,. 1992. Applied multivariate data analysis, Volumn II: Categorical and multivariate methods. New York: Springer-Verlag.

Keeley, JE., Fotheringham, CJ., 2003. Impact of past, present, and future fire regimes on North American Mediterranean shrublands. In ‘Fire and Climatic Change in Temperate Ecosystems of the Western Americas, Springer-Verlag: NewYork: pp. 218–262.

Lee, S., Pradhan, B,. 2006. Probabilistic landslide hazards and risk mapping on Penang Island, Malaysia. Journal of Earth System Science 115 (6), 661-672.

Lozano, F. J., Suarez-seoane, S., Luis, E. D., 2007. Assessment of several spectral indices derived from multi-temporal Landsat data for fire occurrence probability modelling. Remote Sensing of Environment 107, 533 - 544.

Majnuniyan, H., Zahzad, B., Kiyabi, B., Farhangdareshuri, B., GashtasbMiguiei, H., 1999. Certificate National Park Biosphere Reserve, Environment Publications, 129 p.

Marozas, v., Racinskas, J,. Bartkevicius, E,. 2007. Dynamics of ground vegetation after surface fires in hemiboreal Pinus sylvestris forests, Forest Ecology and Management, Volume 250, Issues 1–2, 15 October 2007: 47-55.

Mohamadi, FM., Shabaniyan, N., Purhashemi, M., Fatehi, P., 2009. Risk fire Mapping using GIS and AHP in part of the Pave forest Journal of Forest and Poplar Iran 18 (4),569-586.

Nieto, H., Aguado, I., Garcia, M., & Chuvieco, E., 2012. Lightning-caused fires in Central Spain: Development of a probability model of occurrence for two Spanish regions, Agricultural and Forest Meteorology 162-163, 35-43.

Pewb, K.L., Larsen, C.P.S., 2001. GIS analysis of spatial and temporal patterns of human-caused wildfires in the temperate rain forest of Vancouver Island, Canada. Forest Ecology and Management 140, 1–18.

Rhbar, D., 2000. Environmental impacts tourism industry, management development 19: 4 p.

Rossiter, D.G., Loza, A., 2010. Analyzing land cover change with logistic regression in R, Technical Report ITC, Enschede, 71 pp.

Rundel PW, King JA,. 2001. Ecosystem processes and dynamics in the urban/ wildland interface of Southern California. Journal of Mediterranean Ecology 2, 209–219.

Shokri, M., Safaeian, N., Atrak Chaly, AA., 2002. Evaluate the impact of fire on vegetation countryside in the Takhti yelaghe Golestan National Park, Iranian Journal of Natural Resources 55(2), 273-281.

Silvia Merino-de-Miguela Huescab, M., Gonzalez-Alonsob, F., 2010. Modis reflectance and active fire data for burn mapping and assessment at regional level. Ecological Modelling 221, 67-74.

Stephens, S., Finney, M,. 2002. Prescribed fire mortality of Sierra Nevada mixed conifer tree species: effects of crown damage and forest floor combustion, Forest Ecology and Management 162, 261–271.

Stephens, SL., 2005. Forest fire causes and extent on United States Forest Service lands. International Journal of Wildland Fire 14, 213–222.

Swets, J. A. 1988. Measuring the accuracy of diagnostic systems. Science, 240, 1285–1293.

Syphard, A.D., Radeloff, V.C., Keuler ,N.S., Taylor R.S., Hawbaker, T.J., Stewart S.I., Clayton, M.K., 2008. Predicting spatial patterns of fire on a southern California landscape. International Journal of Wildland Fire 17 (5), 602–613.

Tanskanen, H., Venäläinen, A., Puttonen, P., Granström A., 2005. Impact of stand structure on surface fire ignition potential in Picea abies and Pinus sylvestris forests in southern Finland. Canadian Journal of Forest Research 35, 410–420.

Thenkabail, P. S., Gamage, M. S. D. N., Smakhtin, V. U., 2004. The use of remote sensing data for drought assessment and monitoring in Southwest Asia, Research Report 85, International water management Institute.35p.

USDA, USDI., 2001. Urban–wildand interface communities within vicinity of Federal lands that are at high risk from wildfire. Federal Register 66, 751–777.

Vakalis, D., Sarimveis, H., Kiranoudis, C., Alexandridis, A., Bafas, G., 2007. AGIS based operational system for wildland fire crisis management, mathematical modeling and simulation Applied Mathematical Modeling 28(4),389 – 410.

Whelan Robert, J., 1995. The ecology of fire. Cambridge University Press, New York, NY, 346 pp.

Zhang, Z. X., Zhang, H. Y., & Zhou, D. W,. 2010. Using GIS spatial analysis and logistic regression to predict the probabilities of human-caused grassland fires. Journal of Arid Environments 74(3), 386-393.