Spatial Assesment of Forest Fire Distribution, Occurrence and Dynamics in Province-2, Nepal

This study was objectively conducted to assess trends of forest fire occurrence, burnt area and its causes and management measures in Province-2, Nepal. Altogether 48 Questionnaire Survey and 32 Key Informant Interviews were organized to collect primary data while secondary data were collected from Fire Information for Resource Management System from year, 2002 to 2019. Total 36 maps were produced showing fire occurrences (18) and burnt area pattern (18) of targeted area. Result showed that total 5289 forest fire incidents and total 499,538.9 ha were burnt from 2002 to 2019. The highest number of forest fire incidents was observed in March with 2975 incidents covering 56.24%. The highest incidence was recorded in Lower Tropical Sal and Mixed Broadleaf Forest with 3237 observations. One-Way ANOVA showed that fire occurrence and burnt area among Lower Tropical Sal and Mixed Broadleaf Forest (LTSMF), Hill Sal Forest (HSF) and Outside Forest Region (OFR) were significantly different at 95% confidence level. Mann-Kendal correlation showed that there was positive correlation (R=0.393) between year and forest fire occurrence in LTSMF as well as between year and burnt area of HSF (R=0.09). Principal Component Analysis in Parsa district showed, unextinguished cigarette butts and litter fall was positively correlated. Abbreviations: Rainfall; LDP: Litter fall; NF: Natural Fire; Cigarette Smoke; Unnecessary Weeds and Litter; Strict Rules Fine Against Forest Occurred by Carelessness; CPMFF: Community Involvement in Management of Forest Fire.


Introduction
Forest resources are the vital requirement of every living being to survive on the earth as without this life cannot be imagined.
Globally, there are many types of limiting factors which affect the growth of the forest [1]. Some of the important limiting factors are forest fire, extreme rainfall, erosion, illegal logging, invasive in Amazon [5]. A number of incidents of forest fire were recorded and over 1.2 million trees were lost in Pakistan's northwest Khyber Pakhtunkhwa province during the time period of July 2018 to June 2019. 6 In Sri Lanka massive forest fire was recorded which was found to broke out in the Maragala Mountain area in Moneragala, in the Uva Province and it was found with extreme temperature nearby over 400oC and damaged about 2.023 square km forest area [6]. Over 2,318.88 hectares of forest was lost in July 2019 in Thailand. The MODIS data showed over 30,000 forest fires took place in India in 2019. And it was also found that around 95 percent of the forest fires in India were on account of human activity [7].
Forest fire occurs every year in Nepal too losing large junk of forest. We have no database of how much damage is caused and we have no record of number of lives is lost every year due to forest fire. Thus, such researches are important particular in Province 2, Nepal which is the most vulnerable and prone area to forest fire [8,9]. Of the total forest fire incidences it was found that about 58% forest fire are caused by deliberate burning by grazers, [10] poachers and non-timber forest product collectors, 22% are caused by negligence and 20% occur by accident [11][12][13]. Along with GIS, the MODIS data is very useful to analyze the forest fire and damage caused [8,14,15]. Nepal lacks proper institutional as well as financial and technological capability to combat forest fires and also the country doesn't have the full record of forest fire occurrence and its impacts [8,16]. So, to bridge this gap in the assessment of trend and dynamics of forest fire, this study is an attempt to map the forest fire dynamics and to assess the trend of forest fire occurrence and burnt area pattern in Province number-2 of Nepal using Geographic Information System (GIS) and explore causes of forest fire and its management options in the targeted area.

Study Area
The study area is located in Province-2 of Nepal. Province-2, lies in the southeastern region of Nepal that was formed after

Primary Data Collection
The primary data and information was collected using PRA tools like Questionnaire survey and Key Informant Interview.
Questionnaire survey was conducted among 48 respondents (6 from each 8 districts) and total 32 Key Informant Interview was organized (4 from each 8 districts). Respondents were from Parsa (10), Bara (10), Rautahat (10), Sarlahi (10), Mahottari (10), Dhanusha (10), Siraha (10) and Saptari (10). For the Key Informant Interview, key informants having deeper understanding about the subject matter were selected. 32 key informants selected were forest officials, local users and CFUG members from each 8 districts. Perception level on damage, causes, consequences and different adaptation measures required for control of forest fire was gathered using the applied PRA tools. Close-ended questions were prepared about the fire, its causes, preventive methods, and management strategies.

The email communication, telephone communication and field
visits were applied to collect the required data. The closed ended questions were ranked with series of statements using Likert scale and the respondents were to rate their views in the scale ranging from 5 "Highest score" to 1 "Lowest score". These scores were then analyzed using Principal Component Analysis (PCA) method. This study analyzed the trend of occurrence of forest fire and mapped the forest fire incidence and burnt area dynamics of the forest of Province-2 of Nepal using active fire records from multitemporal data of Moderate Resolution Imaging Spectroradiometer satellites, provides the processed data to universities and research institutes as part of the academic frontier project. The MODIS active fire product detects fires in 1x1km pixels that are burning at the time of overpass under relatively cloud-free conditions [18,19]. In concern to Nepal, it does not have own reliable information, tools and technology for the detection and monitoring due to which MODIS vector data were downloaded from NASA's official website.

MODIS Data
The forest fire point and burnt area related shape file was archived from the MODIS data as data sources for this study. The MODIS burned area product (MCD-64) was extracted from NASA's official website (FIRMS) which gives whole burned area of South Asia (Table 1).

Ground Verification
Data given by NASA (MODIS fire occurrence and burnt area data) was verified by the information and data provided by the Department of Forest and Soil Conservation and forest officials of each district of Province-2. The record of GPS coordinates of fire occurrence are available in Department of Forest and Soil Conservation which were provided by officials and were used to validate the fire occurrence. The location of the burnt area was verified with the Key Informant Interview during data collection.
Field visit was also done to collect the location of frequently fire occurring areas.

Spatial Analysis
MODIS fire data was used to analyze patterns and trend of forest fire incidence and burnt area by forest fire in Province-2 from 2002 to 2019. Software, Arc GIS 10.5 was used for analyzing and interpreting the satellite images, GIS layers data, burnt area pattern and forest fire trend analysis based on density of fire accumulation.
Statistical data was entered and analyzed using SPSS. Charts, bar line and graphs was prepared using Microsoft Excel 2010 and Microsoft Word 2010 (Table 2).

Spatio-Temporal Distribution of Forest Fire Occurrence
The Out of total detection, about 4991 fires were detected with greater than 50% confidence. Forest fire distribution pattern in Province-2 in last 18 years are presented in (Figure 2).

Month Wise Spatio-Temporal Distribution of Forest Fire Occurrence
Forest fire occurrence on a monthly basis was studied for each year. The result showed that the highest fire incidents occurred in

Spatio-Temporal Distribution of Burnt Area based on Forest Types
The burnt area pattern was differed according to the forest      ANOVA showed that there was a significant correlation between the burnt area and the fire incidences in HSF at 95% confidence level as P-value was 0.001 (i.e. P<0.05). The variables were also examined using the t-test, which showed significant relation between the two variables viz. burnt area and forest fire incidences, as the P-value was 0.001.

Correlation between Fire Incidence and Burnt Area in Outside Forest Region
The Curve estimation showed that there was positive correlation (R=0.562) between the burnt area and forest fire incidences in the OFR. The equation showed that, y = 28.340*x + 764.408, where y is the burnt area coverage and x is fire incidences. The equation depicts that the burnt area coverage in Outside Forest Region was increased by of 28.340 ha every year (Figure 17). ANOVA and t-test were applied to test the correlation between the burnt area and forest fire incidences in Outside Forest Region (OFR). ANOVA showed that there was a significant correlation between the burnt area and the fire incidences in OFR at 95% confidence level as P-value was 0.015 (i.e. P<0.05). The variables were also examined using the t-test, which showed significant relation between the two variables i.e. burnt area and forest fire incidences, as the P-value was 0.015.  (Table 3).

Causes and Management Practices of Forest Fire: The result
showed that there were 6 major causes of forest fire which were               The key informants also provided information on causes of forest fire. About 75% of them responded that most of the forest fire is caused mainly due to human carelessness like unextinguished cigarettes, debris burning, and carelessness while handling fire etc. and remaining 25% responded that causes as delay in rainfall, accumulation of litter (fuel) etc. are causes of forest fire. When asked about difficulties in fire control and management, 65% informants said that the community participation in control of forest fire has improved in recent years as awareness programs have been facilitated by the Division Forest Office. But the local viewpoint was found to be quite different than the forest officials, who mentioned that the forest fire rate increases due to lack of forest control tools, training and patrolling provided by the responsible departments under Division Forest Office.

Spatial and Temporal Trend of Forest Fire
The spatial and temporal distribution patterns of forest fire Among the total forest fire event, 2498 forest fire incidence was observed in Parsa district, 757 forest fire incidence was observed in Bara district, 509 forest fire incidence was observed in Rautahat district, 451 forest fire incidence was observed in Sarlahi district, 339 forest fire incidence was observed in Mahottari district, 257 forest fire incidence was observed in Dhanusha district, 90 forest fire incidence was observed in Siraha district and 388 forest fire incidence was observed in Saptari district. Out of total detection, about 4991 fires were detected with greater than 30% confidence.
Research shows that highest forest fire incidence in Province weeds, and undergrowth having low moisture content only because of high temperature, low relative humidity and precipitation and high wind velocity in this season [20]. This research study shows that the lowest fire incidence was observed from the month of June to October which is supported by the study by Khanal (2015) [21] which states that low fire activity was observed in the period between July to October in Nepal. Comparing this study with the study of forest fire in other countries in South Asia it was found that the fire occurrence month in other South-Asian countries were almost similar to Nepal. The study by Tian (2013) [22] shows that among the forest fires in China, March was the month with the most fires (60.0%), followed by April (

Major Causes of Forest fire and its Management Practices
According to the study conducted it was analyzed that most of In the paper Report on Fires in South Asian Region it has been mentioned that Bhutan's climate conditions during winter (lack of rainfall, and high wind velocities) very much favor ignition of fires.
Also, the end of the dry winter season is used to prepare the fields with burnings and it is very common that these fires escape and cause damage. Based on this research it has been found that long dry period, delay in rainfall and uncontrolled burning of accumulated litter were major causes of forest fire in Province-2 which shows similarity in research by Benndorf (2008) [28]. This shows that Parsa district. This study shows that community involvement in management of forest fire is one of the most important management strategies to control forest fire in Province-2 which is supported by the study done by Padillah (2006) [27].
Community involvement in monitoring and management of forest fire must be considered effective as they possess valuable knowledge of place, fire history, traditional management practices, fuel loading etc. and having ownership makes these communities feel more inclined and responsible for the management practices.
Our study also depicts the increasing need of public awareness programs as a management strategy which is also supported different studies in Gambia, Honduras and India [29]. Public awareness can be used as an effective tool for fire management as increase in public awareness program can lead towards increase in active participation of forest users in forest fire management and forest biodiversity enhancement activities. Our study clarifies that the creation of fire lines and fire breakers can be used as an efficient tool for forest fire management which is consistent to the study by Padillah (2006) [27]. Fire-lines are effective management approach as construction of fire lines around and inside the forest helps to align the fuels which ultimately helps to segregate, stop and control forest fire.
Similarly regular forest fire patrolling, removing unwanted weeds and litter and strict laws and fine against carelessness to forest fire are other important practices as shown by this research study that must be taken in consideration for management of forest fire in Province-2, which is also supported by the study done by Benndorf (2008) [28]. The study by Benndorf [30][31][32] states that practices as provision of a legal and financial basis of fine, strict law enforcement, provision of basic tools and materials for fire patrolling and launching of forest fire management programs must be reinforced for management of forest fire which utterly supports the findings of this research study for management of forest fire in Province-2.

Conclusion
The fire incidence was found varying in different year in different districts of Province-2. The highest forest fire occurrence