Journal of Spatial Analysis

---

Forest fire is one of the important problems in Iran which is caused by different factors such as human and natural factors. One of these factors is climate conditions that can be created by heat wave and special circulation of atmospheric phenomena. Occurrence of forest fire in north of Iran have different impacts on environment such as destruction of natural. According to the position of Iran in the dry climate zone provides required conditions for this hazard. Unfortunately,every year thousands of hectares of precious green cover is burned. Forest fires have harmful effects on human life directly,or in directly and lead to environmental destruction and pollution, global warming, loss of vegetation, and dry soil erosion. As a result, research on forest fires will become necessary. The study region is Mazandaran province forests located  in north of Iran with area of  23756.4 square Kilometers.The main object of this study is to detect the forest fires using satellite data and associated analysis with synoptic approach based on weather maps.

To detect fire in the study area different satellite data such as  synchronized and geostationary satellite data were used. In this study, MODIS satellite imagery and global algorithm detection of fire to detect fire in the forest and meadows of Mazandaran province were used. The climate data including weather station data and weather map were used. Other data include data of LST and vegetation products of MODIS. In order to downscale the global data an appropriate threshold was defined. In the proposed method,  After geometric correction and radiometric the cloud mask was removed, And then fire potential was identified with different thresholds and tests. Three fire episodes of  Savadkooh 2006, Noor , 2009 , and Behshahr, 2010 were selected for study.

Results showed  a threshold value of 310 ° K for MODIS sensor band 22 is good for a global scale. Cold and small fires are not detected, Therefore Local threshold was used. In addition, surface temperature and vegetation mapping , chlorophyll amount of vegetation were used before and after the fire episode.It became apparent that the amount of chlorophyll was reduced and the temperature was increased after the fire.

   The synoptic maps of the fire day showed a low pressure over the region and mid level systems indicated the advection of warm air over the area. Surface stations showed the increase of temperature and reduction of moisture during the fire days over the long period mean values.

According to the results of the study the ground level data accompanied the upper level images and pressure patterns.

Universal high performance of fire detection algorithm was used  to identify areas of forest fires Using MODIS satellite images and global algorithm modified to suit the characteristics of the study area fire detection. Then three of the fires were identified with this method. The algorithms with MODIS images and weather data together indicated the validity of the study and performance of this algorithm to identify the location of fire in the study region. Therefore the method of this study can be used in other areas to detect forest fires.

---

Determining Effective Factors on Forest Fire Using the Compound of Multivariate Adaptive Regression Spline and Genetic Algorithm, a Case Study: Golestan, Iran   
Pahlavani, P., Assistant professor at School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran
Raei, A., PhD Candidate of GIS at School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran
Bigdeli, B., Assistant professor at School of Civil Engineering, Shahrood University of Technology
 
Keywords: Forest Fire, Multivariate adaptive regression spline, Multiple linear regression, Logistic regression, Genetic Algorithm.
 

1. Introduction

Nowadays, Determining the effective factors on fire is so important, because the plenty areas of forests around the world are destroyed annually by fire and recurrence of that in the long term can irreparably damage to the earth and its inhabitants. It helps us to identify most dangerous locations and times in forest fire. Hence, we can prevent many of driving factors of forest fire by law enforcement, efficient forest management policies and more supervision. In the current study, we identified the effective factors on the fire in Golestan forest through integration of three different methods including multiple linear regression, logistic regression and multivariate adaptive regression spline with Genetic Algorithm.

2. Study Area

Golestan Province is in the North of Iran and 18% of it is covered by forests. Golestan Province is a touristic province and several roads pass through its forests and according to statistical records, most of the occurred fires were in proximity of these roads. Our study area is located in 36°53′-37°25′N and 55°5′- 55°50′E and its area is about 3719.5 km2. We selected this area, because includes the most of fires have been occurred in Golestan Province in recent years.

3. Materials and Methods

A big fire was occurred on 12 December, 2010 in our study area and we used it as the dependent variable. The actual burnt area and some other data, such as Digital Elevation Model (DEM), the roads network, the rivers, the land uses, and soil types in the area were provided from Golestan Province Department of Natural Resources. Also, geographic coordination of the synoptic weather stations near the area and their data, including maximum, minimum, and mean temperature; total rainfall, as well as maximum wind speed and azimuth in December 2010 were obtained from National Meteorological Organization of Iran.
The land use and soil layers were in scale of 1:100000 and the roads and the rivers layers were in 1:5000 and all of them were provided in 2006. The region DEM is generated from topographic maps of Iran National Cartographic Center in scale of 1:25000 with positional resolution of 30m and we produced the slope and the aspect layers from it in ArcGIS software with the same resolution. The roads and the rivers were in vector format, hence, we used the Euclidean Distance analysis to generate rasters that each cell of them shows the distance from the nearest road or river.
At first we had 5 weather stations, which is very few for GWR. In this regard, we generated 1000 random points in the area and interpolated data to these points using Ordinary Kriging method with exponential semivariogram model in 30m resolution in ArcGIS software.
The multiple linear regression (MLR) model is the generalization of simple linear regression that is modeling the linear relation between one dependent variable and some independent variables. The general formula of MLR is seen below:
                                                                                                                                    (1)
The unknown coefficients are obtained using least squares adjustment as follows:
                                                                                                                                                      (2)
The logistic regression (LR) model is a nonlinear model for determination of the relation between a binary dependent variable and some independent variables. If we use the values of 0 and 1 for non-fire and fire points respectively, then the probability that a point be a fire point is obtained by Eq. (3):
                                                                                            (3)
If the number of parameters is insignificant compared to the observations, then we use the unconditional maximum likelihood estimation shown by Eq. (4) to compute the unknown coefficients of this model.
                                                                                                                                (4)
The multivariate adaptive regression spline (MARS) model is a flexible non-parametric model that requires no assumption about the relation between the dependent and independent variables. Hence it has a high ability in determination of complex nonlinear relations among the variables. The general formula of MARS is seen below:
                                                                                                             (5)
 is the m’th basic function that is obtained by Eq. (6):
                                                                                                  (6)
These basic functions are chosen in such a way that leads to minimum RMSE of model.
We use the genetic algorithem (GA) with the fitness function of the normalized RMSE to select the optimum combination of effective factors on forest fire.
 

4. Results and Discussion

In this paper we study the dependence of the forest fire to 14 factors shown in table 1, in the study area. Our results are shown in figures 1 to 3.
 
Table 1. The studied factors in the present research

Factor	Num.	Factor	Num.	Factor	Num.
Aspect	11	Maximum Wind Speed (m/s)	6	Maximum Temperature (℃)	1
Slope	12	Soil Type	7	Minimum Temperature (℃)	2
Elevation (m)	13	Land Use	8	Mean Temperature (℃)	3
Distance from The Residential Zones (m)	14	Distance from The Roads (m)	9	Total Rainfall (mm)	4
		Distance from The Rivers (m)	10	Maximum Wind Azimuth	5

 
 
  
Figure 1. (a) The best and the mean values of fitness, (b) The last best individuals, (c) The average distance between individuals, (d) The fitness of each individual in the last generation using MLR
 
Figure 2. (a) The best and the mean values of fitness, (b) The last best individuals, (c) The average distance between individuals, (d) The fitness of each individual in the last generation using LR
 

Figure 3. (a) The best and the mean values of fitness, (b) The last best individuals, (c) The average distance between individuals, (d) The fitness of each individual in the last generation using MARS

5. Conclusion

This research shows that both of the biophysical and anthropogenic factors have significant effects on forest fire in our study area. Just two factors were identified as impressive factors in all three cases including the minimum temperature and the maximum speed of wind. This study concluded to the NRMSE=0.4291 and R2=0.9862 for the multiple linear regression, NRMSE=0.9416 and R2=0.9912 for the logistic regression and NRMSE=0.1757 and R2=0.9886 for the multivariate adaptive regression spline and totally the multivariate adaptive regression spline method showed a better performance in comparison to the other two methods.
 

---

Real-time detection of forest fire using NOAA/AVHRR data
Study area :(Kayamaki Wildlife Refuge)
 
Extended Abstract
Introduction
Land and forest fires are one of the most common problems in the world that cause various disturbances in forest and land efficiency. Real-time fire detection is crucial to prevent large-scale casualties. In order to identify early fire in areas where there is a high risk of fire, it is necessary to monitor these areas regularly. Forest monitoring is a technique used to detect fires in the past using traditional techniques such as surveillance, helicopter and aircraft. Today, satellite imagery is one of the most imperative and effective tools for detecting active fires in the world.
Materials and Methods
In this study, NOAA/AVHRR images were used for fire detection and MODIS products were applied for evaluation and validation.
Fire Detection Algorithms
There are several algorithms for detecting fires using satellite imagery. In this study, 3 algorithms of Giglio, extended and IGPP were used. The selection of these algorithms was due to the extensive background research in most of the previous studies that used them and the results of these algorithms, especially the IGPP, were far more than other algorithms.
Giglio Algorithm
Giglio et al., (1999) criticized Arino and Melinott (1993) threshold as too high for certain regions of the world such as tropical rain forests, temperate climates and marshes where the air temperature for small fires (100 m³) is usually between 308 and 314 degrees Kelvin. They believed that the smaller fires were not fully recognized by Arino and Melinott (1993) thresholds. They concluded that in suburban forests 60% of fires had temperatures below 320K of which 70% were in rainforests and 85% happened in the Savanna. Thus, the threshold cannot be applied on a large scale and it is only applicable for a regional scale.
IGBP Algorithm
The IGBP fire detection algorithm is implemented in two steps. The first step is the threshold test in which a pixel in micrometers (11.03 μm) minus the band 4 is greater than 8 degrees Kelvin, the desired pixel being considered as a potential fire pixel. Band 3 (3.9 μm) exceeds 311 K, and band 3 illumination temperature is 3.9.
Developed Algorithm
This algorithm is used to detect small and large fires (both at night and day).
 
Interpretation of the Results
After selecting fire detection algorithms, pre-processing (geometric, radiometric and atmospheric corrections), processing (applying fire relationships and fire formulas for fire detection) and post-processing (evaluating and validating the results), the fires were identified by the fire algorithms (images). Final results of fires identified for 2016 and 2017 (for 4 days) by fire algorithms indicate that fires identified by Giglio algorithm were 22 cases, those by IGPP algorithm were 27 cases and the ones by the developed algorithm were 15 cases. For this reason, the IGPP algorithm can be taken as the most appropriate algorithm in this study for fire detection using satellite imagery.
Evaluation of fires identified through MODIS products
To evaluate identified fires, after recognizing them with relevant algorithms, we used MODIS products for their evaluation (due to the lack of ground data on the days studied for evaluation). MODIS products were obtained from sites where the location of each fire was reported. For the evaluation of identified fires based on fire detection algorithms with MODIS products, 10 fire occurrences were used. The evaluation results express that out of 10 fires only 7 fires were recognized by the algorithms of MODIS products. 5 fire events were identified by Giglio algorithm (from 7 fires), 6 fires from IGBP (out of 7 fires), and 3 fire events from 7 extended algorithm were selected as fire pixels.
Comparison of the implications of the fire algorithms
The implications of fire occurrence algorithms indicate that the IGBP algorithm with 6 fires (out of 7 tested fires with error rate of 14% and with the number of fires detected (86%)), Giglio algorithm with 5 fires (out of 7 tested fires, with error rate of 28% and with the number of fires (72%)) and the developed algorithm with 3 fires (out of 7 fires tested with an error rate of 57% and with fire rate of 43%) have been identified. Therefore, it is concluded that the IGBP is the most appropriate algorithm for real-time fire detection, followed by Giglio and the developed algorithm in second and third orders, respectively.
Keywords:Real Time Fire Detection, Fire Algorithms, NOAA/AVHRR, Kiamaki Wildlife Refuge.
 

---

 
Prioritizing the counties of Sistan and Baluchestan province in order to optimize the management of drought budget
 
Abstract
Drought is one of the most significant natural hazards which affects various aspects of human life with its gradual, slow and creeping occurrence. As a disastrous climate phenomenon, it directly impacts on human communities through making changes in access to water resources, and imposes enormous economic, social and environmental costs on various communities. One of the strategies governments always use to compensate the damages that droughts impose on various economic, social and environmental sectors is to include special financial loans in the form of grants, special development credits, special structural projects, and so on. Therefore, in this regard, a lot of financial aids have been allocated by the government for it. But, sometimes, it is observed that the allocation and distribution of drought credits among different cities of the province have been more influenced by political, ethnic, and religious interests rather than by the severity of the drought and its detrimental effects on various economic, agricultural, and social sectors. Therefore, in this study, it has been tried to present a method based on an optimization model in order to take a step in optimizing the management of drought budget based on realities in Sistan and Baluchestan province in south-east of Iran.
In this study, the indicators proposed by the Cabinet of Ministers of the Islamic Republic of Iran for reducing the effects of drought were applied to manage the optimal distribution of drought budgets among the cities of Sistan and Baluchestan province in southeast of  Iran. These indicators were population, having urban and rural drinking water, vulnerability of water resources, surface area under cultivation, number of users of agricultural sector, annual consumption of agricultural water, and rainfall. Using the figures given in the 2011 statistical yearbook, numerical values ​​for each of these indices were obtained. But given that these indices are not of equal importance in each city and some may be more important in one city and less important in another, the Analytical Hierarchy Process (AHP) technique using paired comparison method was applied to determine the relative importance of these indices in Sistan and Baluchestan province. Using this technique, a relative weight was determined for each of these indicators and for each city. In addition to determining the weight of these indices, it was needed to convert droughts in the province to numerical values. In this study, to analyze droughts in Sistan and Baluchestan province, Standardized Precipitation Index (SPI), which is one of the proposed indicators of World Meteorological Organization (WMO) has been used for a period of 30 years (1984-2014) and for 6 synoptic stations.
The results of drought analysis in Sistan and Baluchestan province in an annual scale showed that the frequency distribution of different drought levels in Sistan and Baluchestan province has not been uniform, with the most droughts occurring in this 30 year period in northern part of the province (Zabol and Zahedan) including 14 and 13 repeated cases; respectively, with the lowest in central part of Baluchistan (Khash and Saravan) including 11 repeated cases. But, what should be noted after the frequency of droughts is the persistence and durability of droughts. As a result, weak but long-lasting droughts can be much more damaging than severe but short-lived droughts. Sistan and Baluchestan province has been under drought with varying degrees of severity from the water year of 1998-1999 to the water year of 2005-2006. But in the meantime, the north of Sistan and Baluchestan province (Zabol station) has a condition which is very different from other parts of the province. The north of Sistan and Baluchistan province has been experiencing severe droughts for 14 years from the year 1999-2000 to 2012-2013 with the exception of the year 2004-2005. It has been the long-term drought persistence in the province that has caused many environmental, economic and social problems, such as the drying up of Lake Hamoun and the water reduction in the Hirmand River in the north of the province, and severe reduction of groundwater resources in central Baluchistan. Therefore, given that Sistan and Baluchestan province has experienced 7 years of continuous drought between the two water years of 1998-1999 to 2009-2010, these seven years were chosen as examples, and the ant colony algorithm was implemented on them.
Based on the selected criteria and the proposed method, it seems that the combination of Analytic hierarchy process (AHP) technique and Ant colony algorithm can formulate an intelligent system to determine the priority of the cities in Sistan and Baluchestan province during droughts for allocating the optimal budget and reducing the harmful effects of droughts. Based on the prioritization maps of optimal drought budget allocation among the cities in Sistan and Baluchestan province, an almost good agreement can be observed between the results and the realities existing in the province.
Keywords: Sistan and Baluchestan, ant colony algorithm, Analytical Hierarchy Process, budget, Management
 
 
 

---

---

Landslides, as one of the most dangerous natural hazards in mountainous regions, continuously threaten human infrastructure, especially roads and transportation routes. Their occurrence often results in significant loss of life and property, making it crucial to study and assess landslide hazards for effective zoning. The purpose of this research is to zone the landslide hazard along the Masal to Gilvan road using a neural network algorithm. The neural network algorithm is recognized as one of the most effective machine learning models, capable of solving complex problems in prediction and classification despite its simplicity. For this zoning analysis, nine influencing factors were considered: (1) geology, (2) vegetation cover, (3) slope, (4) land use, (5) distance from the road, (6) slope aspect, (7) elevation, (8) distance from fault lines, and (9) distance from rivers. The data were prepared, preprocessed, and then entered into MATLAB 2018. A neural network model was designed and implemented with 9 input neurons, 8 hidden neurons, and 1 output neuron. The results indicated that the four most influential factors, ranked by weight, were: slope (0.24), vegetation cover (0.17), distance from fault lines (0.14), and geology (0.11). Final validation using the ROC curve showed that the AUC values were 0.854 for the training phase and 0.971 for the testing phase, both of which reflect highly favorable results. The error rate was found to be very low.