Introduction

One of the major causes of biomass burning is forest fire which impacts the environment, depending on their intensity and frequency (Crutzen & Andreae, 1990; Fearnside, 2000; Darmawan et al., 2001; Rabade & Aragoneses, 2008; Gao et al., 2018; Mendez-Espinosa et al., 2019). The principal causes of forest fires are natural viz. volcanic eruption, earthquake, lightning, etc., and anthropogenic viz. shifting agriculture, inadvertent fires, planned to burn, burning of firewood and agricultural residues, etc. Emissions from a forest fire can significantly decline the quality of air and population health over a region. A massive portion of pollutants like carbon dioxide (CO2), carbon monoxide (CO), methane (CH4), nitrogen oxides (NOx), ammonia (NH3), particulate matter (PM), non-methane volatile organic compounds (VOCs), sulfur dioxide (SO2), Black Carbon (BC) and various chemical substances are released in the smoke (Hao & Liu, 1994; Miranda et al., 2005; Yokelson et al., 2007, 2011). It also affects the radiation budget globally as pollutants remain loaded in the atmosphere. In addition to its environmental and climatic impact and reducing visibility, around 350,000 deaths per year have been reported worldwide due to inhalation from wildfires (Johnston et al., 2012).

The North Eastern Region (NER) of India is one of the 18 biodiversity hotspots globally. It has some of the largest tropical and sub-tropical forests reserves of wet evergreen, semi-evergreen, moist deciduous, coniferous forests, mixed forests, and shrubs (Roy & Joshi, 2002), covering almost 64.66% of its geographic area. Every year forest cover of NER of India undergoes significant changes due to shifting (called “Jhum”) cultivation. Some people burn forest areas to create new agricultural fields, leading to fires from those fields (Puri et al., 2011) and causing forest fires. Historical data analysis of forest fire counts in NER of India shows an average of 100 thousand fire events annually (Sarma et al., 2017). From 2015 to 2017, this region experienced a forest loss of 630 sq km because of slashing and burning activity related to shifting cultivation (FSI state of forest report, 2017). A decrease of 1020 sq km of forest area in the region has been reported due to the shifting cultivation, natural calamities, and development activities etc., (FSI state of forest report, 2021a; b). With the increase in population and agriculture demands in NER, the cycle of “Jhum” cultivation is reduced from 25–30 years to 2–3 years, resulting in more destruction of forest cover in the area (Puri et al., 2011). Badarinath et al. (2004) reported the massive loading of pollutants and higher turbidity in Mizoram during the period of forest fire events in comparison to the pre-burning and post-burning periods. During the burning phase, a considerable reduction in direct solar radiation and lower visibility was reported in the same study. Kundu et al. (2018) demonstrated that the Brahmaputra river valley in NER is burdened with a lot of natural and anthropogenic aerosol during local winter and is modified by the prolonged dispersion of pollutants from the Indo-Gangetic plain (IGP). Dahutia et al. (2018) reported the aerosol features and their climatic consequences over the region by employing numerous satellites and reanalysis datasets. The natural occurrences, as well as the anthropogenic sources, play a pivotal role in the evolution of aerosol particles over the region (Barman et al., 2018). Barman et al. (2019) carried out a study at Agartala city (the capital of Tripura, a state in NER) and found that in pre-monsoon and monsoon seasons, biomass-burning aerosols are dominant, whereas fossil fuel combustion predominated in other seasons. Borgohain et al. (2019) investigated aerosol’s physical and optical properties using ground-based instruments under campaign mode and mentioned significant biomass burning along the sub-Himalayan terrains of NER during winter and pre-monsoon seasons. Barman and Gokhale (2019) reported enhanced BC mass concentration (BCC) over Guwahati city, mainly contributed by vehicular emissions. Saikia et al. (2019) stated that open biomass burning is the region’s dominant source of BC and CO.

According to Statistical Year Book India (2018), the decadal population growth rate of NER shows a 17.71% increase from 2001 to 2011, equal to India’s growth rate (17.64%). The population growth rate of some of the states of NER of India like Assam (19.93%), Arunachal Pradesh (25.92%), Meghalaya (27.92%), Mizoram (22.78%), and Manipur (18.65%) was more than the national growth rate. In comparison, Nagaland (− 0.47%) had negative population growth and Sikkim (12.36%) according to the 2011 census (2011–2021 census not completed). The number of registered vehicles increased from 878 to 4045, and factories rose from 1995 to 4457 in a span of 15 years (2001–2015) over the NER. Figure S1 (Supplementary Figure 1 in Supplementary Material) depicts the national, regional, and state-wise decadal population growth rate and the regional changes in the number of registered vehicles and factories in NER of India from 2001 to 2015. Sheel et al. (2010) found a rising trend in the tropospheric NO2, which is produced mostly by combustion over different regions of India as a result of rapid industrial and vehicular growth. Ghosh et al. (2017) also reported variations in tropospheric NO2 due to biomass burning and combustion of fossil fuels in industries and vehicles over IGP. Similarly, Sur et al. (2021) studied the variation of TCN over IGP during the lockdown period due to the outbreak of COVID-19 and observes a significant decrease in NO2 concentration during the period.

The states in NER are unique depending upon forest types, orographic conditions, land use, “Jhum” practice, development, and anthropogenic activities. The influence of biomass burning/forest fire on air quality in the area is one of the essential topics of study. A few studies reported significant aerosol and trace gas loading over the NER of India. However, a thorough investigation of the possible effects of forest fires biomass burning on aerosol and trace gas loading is still lacking. This paper investigated the impact of fire counts on BC and TCN over the NER using satellite data for 15 years of fire counts, BC and TCN. The space and time variability of the mentioned parameters over the study domain is also discussed.

Study Domain and Synoptic Meteorology

The study region is between latitude 20° N–30° N and longitude 88° E–98° E with particular emphasis on the NER of India. North East India comprises eight states viz., Assam, Arunachal Pradesh, Manipur, Meghalaya, Mizoram, Nagaland, Tripura, and Sikkim (Fig. 1). Approximately a quarter of the forest area of India i.e., 23.75% (FSI state of forest report, 2021a, b) is found in NER of India with approx. 1.7 lakh km2 within reserved, protected, and unclassified forests (Ningthoujam & Mutum, 2010). As per the percentage of the total geographical area, the top five forest cover states are in North East India, viz. Mizoram (84.53%), Arunachal Pradesh (79.33%), Meghalaya (76%), Manipur (74.34%), and Nagaland (73.9%). Figure S2 shows the total Geographical Area (GA) and Forest Area (FA) of NE states of India. These areas are affected by numerous forest fires each year during the dry seasons. NER has a sub-tropical climate due to its unique topography. The Himalayan Mountains exist in the north direction of NER of India, while the Indo-Myanmar ranges of foothills and hills in the east direction, and Mizoram and Tripura in the south direction (Fig. 1). In the winter (Dec–Feb) and pre-monsoon (Mar–April) seasons westerly winds play a crucial role in the atmospheric processes of the region as it is open in the west towards the IGP. The southwest monsoon is the primary source of rain in the area. Cherrapunjee receives the world’s highest rainfall owing to the orographic lifting of the southwest monsoon winds by the Meghalaya plateau. The mean temperatures vary from 15 °C (January) to 28 °C (August) in the plain area, while 9–21 °C in the hilly region during the same periods. The mean relative humidity varies between 30 and 50% during pre-monsoon seasons and 80–95% during monsoon. Averaged wind vector at 850 hPa from the ECMWF Reanalysis ERA-5 spanning 15 years (2006–2020) during the interval from January to April over 20° N–30° N and 84° E–98° E is shown in Fig. S3 which illustrates that westerly wind is dominant during the months. The wind speed ranges from 2 to 3 ms−1 in January and February, ranging from 4 to 5 ms−1 in March and April. The westerly wind converts to southwesterly over the Brahmaputra valley during January and February. Further, it converts to southeasterly over the eastern part of NER during the same periods. Again, it converts to southwesterly/southeasterly over Nagaland, part of Manipur, and other eastern borders.

Fig. 1
figure 1

Study domain and monthly averaged NDVI of pre-monsoon season (2006–2020) obtained from MODIS-Terra

Full size image

Data and Methodology

Fire Count, BC Surface Mass Concentration and TCN

In the present study, data is utilized from MODIS (Moderate Resolution Imaging Spectroradiometer), MERRA-2 (Modern Era-Retrospective Analysis for Research and Applications), and OMI (Ozone Monitoring Instrument) as listed in Table 1. MODIS Collection 6 provides active fire count data with a minimum 30% confidence level. MODIS fire products are obtained from a detection algorithm that uses brightness temperatures and reflectance for daytime observations (Giglio et al., 2003). The algorithm identifies the fire pixels and assigns them to their respective class of missing data, cloud, water, non-fire, fire, or unknown. Collection 6 also provides improved data sets by eliminating false alarms caused by small forest clearings and identifying thick smoke from the cloud, etc., the details are available in Giglio et al. (2003, 2016). MERRA-2 includes Goddard Earth Observing System (GEOS) model and Grid point Statistical Interpolation (GSI) assimilation system, which provides continuous long-term earth system datasets. OMI gives global observations of ozone and other atmospheric pollutants and aerosols by observing the Earth’s backscattered radiation with a hyperspectral imaging spectrometer that employs Charged Coupled Device (CCD) detector. It includes three separate sensors for different spectral ranges, out of which NO2 is detected in the visible spectrum of 405–465 nm. Further details on the MERRA-2 model and OMI are available in Gelaro et al. (2017) and Krotkov et al. (2016). MODIS fire counts, BC surface mass concentration, and TCN are collected from 2006 to 2020 over the study domain, and analyzed for each state of NER during the months of fire events.

Table 1 Data source and its resolution
Full size table

Possibility of Long-Range Transport

Variability of air mass trajectories can change the chemical and loading characteristics of atmospheric airborne particles and gases. So, to understand the possibility of a distant source and long-range transport of aerosols, the HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) model of the National Oceanic and Atmospheric Administration (NOAA) has been used with the help of Trajstat software (Wang et al., 2009). Five days of isentropic back trajectories ending at 500 m and 3500 m altitude over the west, central and east regions (27.5° N, 88.5° E; 26.5° N, 92.5° E; and 24.5° N, 94.0° E) of NER for winter and pre-monsoon from 2006 to 2020 is shown in Fig. 2. Daily trajectories have been clustered into seasonal trajectories by combining the nearest trajectories in Euclidean distance (Harris & Kahl, 1990; Sirois & Bottenheim, 1995) to obtain seasonal mean pathways.

Fig. 2
figure 2

Cluster mean backward wind trajectories ending over three places over NEI during winter and pre-monsoon seasons at 500 m and 3500 m AGL (only the first three significant clusters are shown here)

Full size image

Results and Discussions

Forest Fire Counts Over the Region

The study years have been classified into five periods viz., T1 (2006–2008), T2 (2009–2011), T3 (2012–2014), T4 (2015–2017), and T5 (2018–2020) for the months January, February, March, and April. Figure S4 shows fire events over the time intervals over the study domain. During these months, aerosol loading is higher due to the higher fire counts, scarcity of rain, and the long-range transport of pollutants from IGP and Bangladesh (Pathak et al., 2015). Badarinath et al. (2011) reported intense forest fire episodes in Kerala during the year 2004, which started in January and increased even more in February and March due to a decrease in rainfall and high temperatures than normal years. Jain et al. (2021) observed the occurrence of increased forest fires associated with severe heatwave, drought and strong El Niño during 2015 July–January and 2018 February–June over central India. Chand et al. (2006) found that the spatial variation in peak fire seasons was attributed to the varying forest types and climatic conditions over the Indian subcontinent. They reported intense fire episodes over Himalayan zone forests during May and June every year due to moist conditions in April and prevailing dry weather conditions in May. Whereas, high fire counts were recorded in the tropical evergreen forests of North East India during February, March, and April with a maximum number of ~ 1200 fires per day. Kharol and Badarinath (2008) reported the high forest fire incidents over NER of India due to the burning of agricultural lands in March and April. In the same study, the author also records high forest fire incidents in Mizoram. It is evident from Fig. S4 and Table 2 that there are always large numbers of fire events in March, which is on average 2–15 times more than any other month like January, February, or April. Fire counts vary from 64,092 to 107,506 in March, 4972–6204 in January, and 12,268–28,393 in February. From 2009 to 2011 encountered maximum fire count in winter and the third maximum fire count in pre-monsoon, related to the drought episodes (NRAA, 2009, 2013). In the drought period, less availability of moisture in vegetation makes them more flammable, thereby increasing the probability of forest fire (Littell et al., 2016).

Table 2 Fire counts, total surface BC mass concentration, and TCN over the study domain in Jan, Feb, Mar, and Apr
Full size table

Spatial Variation of BCC and TCN

Spatial variation of surface BC mass concentration (BCC) and TCN over the study domains are shown respectively in Figs. 3 and 4 for all the study months from the year 2006 to 2020 are split into five periods T1, T2, T3, T4, and T5. Western parts of Assam and Meghalaya, the entire Tripura, and parts of Mizoram show more surface BCC during January and February for all the periods. The increase in BCC towards the western territory of NER can be attributed partially to native biomass burning activity and mainly to the transport of BC from IGP by westerly, as represented in Fig. S3. BC hotspots were shifted towards the eastern area of NER during March and April and covered most parts of Mizoram and a few parts of Manipur, Tripura, and Assam due to increased forest fire events in those areas. The total average surface BCC and TCN during all 4 months over NER from 2006 to 2020 are provided in Table 2. Spatially averaged BCC varies from 1.80 to 2.76 µg m−3 in January and February and 1.82–2.36 µg m−3 in March. BCC is observed more in Mizoram, Tripura, Manipur, Nagaland, and Brahmaputra valley of Assam than in the rest of the NER. Babu et al. (2013) studies long-term aerosol optical depth (AOD) trend over 8 ARFINET stations (Trivandrum, Visakhapatnam, Mysore, Anantapur, Hyderabad, Patiala, and Dibrugarh) spread across India and found that Dibrugarh, (eastern part of Assam) shows a highest increasing trend of aerosol loading during the dry months (DJFM) than the rest of India with a value of ~ 0.02 year−1. Badarinath et al. (2007a), also studies the variation of BC aerosol, carbon monoxide, and ozone over an urban area of Hyderabad during the forest fire season and found high values of BC during active forest fire events over the region. Although NER is one of the remote locations of India, less than 20% occurrence frequency of BCC above 20 µg m−3 was observed during winter and pre-monsoon season which was comparable to some of the high BCC values of the winter season of some urban/metropolitan cities like Delhi, Hyderabad, Kolkata, Chennai etc., (Gogoi et al., 2017). Over the study region, we have observed that BC is higher during January and February, but Mizoram exhibits more BCC in March than in January and February, where fire counts are more. In addition to the fire counts, more local household burnings in the region during winter might contribute to the more BCC recorded during January and February. Recently, in a study by Gogoi et al. (2021) BC levels were observed to decline by > 30% during the second phase of the lockdown over NER of India, however, the absolute magnitude remained high over the region compared to the rest of India. A week before the lockdown, they recorded more than 10 µg m−3of BCC over IGP and NER and observed a significant decrease of < 4 µg m−3 in BCC during the first lockdown. Meanwhile, Pathak et al. (2021) recorded a substantial decrement of NO2 from 7.65 ppb in 2015–2019 to 4.62 ppb in 2020 during lockdown i.e., about a 43% decrease over NER of India. A 14% decadal increase in TCN was recorded in South Asia with the highest increase of 77% over Dhaka (ul-Haq et al., 2015) which is the capital of Bangladesh, located south of the NER. In a study by Haridas et al. (2019), the highest mean value of TCN was observed over IGP throughout all the seasons and highlights the presence of high TCN levels over NER during the months of pre-monsoon season contributed by the high biomass burning in the region. Kalita and Bhuyan (2011), found the highest concentration of TCN over NER during the dry months from February to May, contributing to the increased levels of tropospheric ozone (O3) even higher than IGP. However, the authors elaborated that high TCN level during the dry months is not solely responsible for the increased concentration of tropospheric O3 and mentioned more lighting and thunderstorm activities during the season as one of the major contributors of upper tropospheric O3. In our present study, TCN also shows the max value in March month over the study domain in the five periods. Maximum TCN has been observed over Mizoram, Manipur, and Nagaland, followed by Assam, Tripura, and Meghalaya. Spatially averaged TCN varies from 18.79 × 1014 to 29.08 × 1014 cm−2 in March, 10.76 × 1014–15.81 × 1014 cm−2 in January and February, and 12.67 × 1014–14.2 × 1014 cm−2 in April. More BC and TCN are observed during 2009–2011, which also happens to be the period of drought. TCN shows 5–10% more during March than in other months for all the periods.

Fig. 3
figure 3

Surface BC mass concentration in January, February, March, and April during a 2006–2008 b 2009–2011 c 2012–2014 d 2015–2017 and e 2018–2020 over the study domain

Full size image
Fig. 4
figure 4

TCN in January, February, March, and April during a 2006–2008 b 2009–2011 c 2012–2014 d 2015–2017 and e 2018–2020 over the study domain

Full size image

Figure 5 depicts monthly fire counts over the domain of interest, spatial area-weighted average of surface BCC and TCN from 2006 to 2020 over NER. As mentioned earlier, Forest fire is significantly high in January, February, March, and April over the NER, which justifies the enhanced BCC and NO2 concentration over the region. Maximum fire counts have been observed in March (12,556–45,498) and April (4864–23,912) every year, followed by February (2882–15,438) and January (1154–3112). TCN showed a maximum (10.98 × 1014–30.48 × 1014 cm−2) in March and April and a minimum in August and September (6.59 × 1014–8.88 × 1014 cm−2), which is in satisfactory correlation with the monthly fire count changes. Surface BCC shows a maximum (1.42–3.15 µg m−3) during January–March and a minimum (0.34–0.48 µg m−3) in July and August in each year of the study. A considerable amount of BCC over NER is observed during winter and pre-monsoonal months. During the winter months, a lot of firewood burning at the domestic level occurs throughout the NE region, particularly over the hilly areas where people burn firewood to combat the severe cold.

Fig. 5
figure 5

Month-to-month variation of fire points (top), spatial average of surface BC concentration (middle), and spatial average TCN (low) over NER of India during 2006–2020

Full size image

BCC Contribution by Various Sources

Production of high absorbing aerosol (BC) over different parts of Assam during the winter season has been reported by Kundu et al. (2018), due to rice crop residue burning, small-scale forest fires, shrubs etc. Transportation of aerosol from IGP by strong westerly wind also contributes to high BC concentrations during the January and February months. The westerly wind turns southwesterly and southerly over the western part of NER during January and February. The southeasterly winds over the Brahmaputra valley of Assam and the eastern part of NER could make a differential contribution from west to east in high BC aerosol loading over the valley (Fig. S3). The same westerly wind curls southwesterly or southeasterly over Nagaland, part of Manipur, and other eastern bordering states during March and April. BC also rises simultaneously. Cluster analysis of seasonal wind back trajectories showed that the majority of them are passed through the IGP and Bangladesh region, which is relatively polluted from industrial emissions, brick kiln releases, burning of fossil fuels, firewood, and cow dung burning and stubble (Barman et al., 2018; Kundu et al., 2018; Pathak et al., 2015). As evident by seasonal mean trajectories, gaseous pollutants, including BC, are being imported by the winds. Borgohain et al. (2019) characterized BC over 3 rural mountainous stations in Northern Arunachal Pradesh, and demonstrated high BC loading due to biomass burning over all the sites and transport of air mass from the IGP, Nepal, Bhutan, and the Brahmaputra valley due to the westerly wind. Many previous studies like Pathak et al. (2012, 2014), Bhuyan et al. (2014), Gogoi et al. (2017) and Barman et al. (2018, 2019) etc. have also shown air mass trajectories originating from the IGP, Bangladesh and West Asia which contributes to the increased level of pollution in the north eastern states. Figure 3 exhibits the trajectory ratios of three significant clusters reaching three chosen places over NER; a remarkable number of trajectories are falling in the cluster groups coming over IGP, Bangladesh, and East Indian states. Consequently, the high amount of BCC over NER during the winter season is dependent on native sources from forest fires, local-level biomass burning, and protracted movement from the IGP, which seems to be aided by westerly winds.

State-wise Forest Fire Counts, BCC and TCN

Figure 6 shows the state-wise yearly variation of total fire count density, spatial averaged surface BCC, and TCN during the dominant fire event months (January, February, March, and April) from 2006 to 2020. BC and TCN show similar trends in the fire count density over NER states and areas other than NER. Fifteen years mean of spatially averaged fire counts density (in FC km−2) has been recorded maximum in Mizoram (0.398) followed by Nagaland (0.257), Tripura (0.251), Manipur (0.190), Meghalaya (0.138), Assam (0.065), Arunachal (0.027) and Sikkim (0.008). Badarinath et al. (2007b) reported the highest forest fires over Mizoram, followed by Manipur, Assam, and Meghalaya and observed a good correlation between aerosol loading and fire incidents over the NE states. Similarly, maximum surface BCC (in µg m m−3) over Mizoram, Nagaland, Tripura, Manipur, Meghalaya, Assam, Arunachal, and Sikkim were observed to be 3.54, 3.20, 4.16, 2.25, 3.26, 3.27, 1.61, and 1.24. Gogoi et al. (2017) reported a longitudinal gradient in BCC corresponding to fossil fuel or biomass burning during different seasons, where the author found that the western parts of the NER is highly influenced by the transport of anthropogenic aerosols from the IGP, whereas the southeastern part close to Myanmar is influenced more by the biomass burning emissions. They recorded 25% biomass burning fraction of the total aerosol absorption close to the surface during the pre monsoon season over the south eastern part of NER. At the same time, maximum TCN (in 1014 cm−2) over Mizoram, Nagaland, Tripura, Manipur, Meghalaya, Assam, Arunachal, and Sikkim were observed to be 25.41, 23.19, 26.67, 28.12, 22.06, 24.09, 10.89, and 7.16 respectively. Sikkim observed fewer fire counts density which also corresponded to low surface BC mass concentration and TCN over the region. BC concentration and TCN show more in the neighboring parts of NER, i.e., other than NER. Bhuyan et al. (2014) recorded the highest daily maximum value of O3 in March over Dibrugarh, Assam, because of the high solar insolation and the presence of large amounts of NOx and CO in the atmosphere due to the high burning activity.

Fig. 6
figure 6

Variation of fire count density (left), surface BC mass concentration (middle), and TCN (right) over each state of NER and other than NER within the study domain during the high fire count months (average of Jan, Feb, Mar, and Apr) from 2006 to 2020

Full size image

Conclusion

Biomass burning which includes burning of vegetation both by man-made and natural causes plays a pivotal role in driving the concentration of aerosol and trace gases in the atmosphere. It releases large amounts of gaseous pollutants and particulates which significantly influences the air quality, human health and climate change. In the present study, we have demonstrated the impact of fire counts on TCN and surface BCC over the NER of India using reanalysis data and satellite observations. Many fire events have been witnessed over the study site in March than any other month of January, February, or April. Fire counts have been found to vary from 64,092 to 107,506 in March and 4972–28,393 in January and February. TCN and surface BCC are higher in March and April where fire counts are also observed to be maximum during the respective months. The highest number of fire counts was recorded during the period of drought i.e., 2009–2011, where the concentration of BC and TCN was also observed to be high over the region. TCN showed maximum (10.98 × 1014–30.48 × 1014 cm−2) in March and April and minimum in August and September (6.59 × 1014–8.88 × 1014 cm−2), which is in good agreement with the monthly variations of fire events. Surface BCC varied from 1.42 to 3.15 µg m−3 from January to March in each year of the study. The significant value of surface BCC observed in January and February is due to the combined emissions from the forest fires, local level biomass burning, and long range transport from the IGP region. A longitudinal shift was observed in the maximum BCC over the region during winter and pre monsoon. It was observed that high BC loading was more towards the western part of NER due to major contributions from the IGP by westerly winds and partially from local biomass burning. However, BCC becomes more towards the eastern part during the pre monsoon season especially over Mizoram and a few parts of Manipur, Tripura and Assam due to  the occurrence of more forest fire incidents than the rest of the NER. Similarly, high TCN values were observed during March over the lower parts of NER i.e., Meghalaya, Tripura, Mizoram and some parts of Assam and Nagaland where large number of fire events were recorded during the month.