Temporal and Spatial Patterns of Biomass Burning Fire Counts and Carbon Emissions in the Beijing–Tianjin–Hebei (BTH) Region during 2003–2020 Based on GFED4

: Biomass burning (BB) plays an important role in the formation of heavy pollution events during harvest seasons in the Beijing–Tianjin–Hebei (BTH) region by releasing trace gases and particulate matter into the atmosphere. A better understanding of spatial-temporal variations of BB in BTH is required to assess its impacts on air quality, especially on heavy haze pollution. The fourth version of the Global Fire Emissions Database (GFED4)’s ﬁre counts and carbon emissions data were used in this research, which shows the varying number of ﬁre counts in China from 2003 to 2020 demonstrated a ﬂuctuating but generally rising trend, with a peak in 2013. Most ﬁre counts were concentrated in three key periods: March (11%), June–July (33%), and October (9.68%). The increase in ﬁre counts will inevitably lead to the growth of carbon emissions. The four major vegetation types of the ﬁres were agriculture (58.1%), followed by grassland (35.5%), and forest (4.1%), with the fewest in peat. In addition, a separate study for the year 2020 found that the ﬁre counts and carbon emissions were different for this year, with the overall average trend in the study time. For example, the monthly peak ﬁre counts changed from June to March. The cumulative emissions of carbon, CO, CO 2 , CH 4 , dry matter, and particulate matter from BB in BTH reached 201 Gg, 39 Gg, 670 Gg, 2 Gg, 417 Gg, and 3 Gg in 2020, respectively.


Introduction
Biomass burning (BB), such as wildfire, agricultural open-burning, residential biofuel burning, forest fire, grass burning, and peatland fire [1], is an important source of atmospheric particulate matter and trace gases [2][3][4][5]. It has an important impact on regional and even global air quality, the chemical composition of particulate matter, climate system, and human health [6,7]. Gaseous pollutants, such as CO, CO2, CH4, NOx, and NMOC (nonmethane hydrocarbons) released by BB, as well as particulate matter, such as BC (black carbon) and OC (organic carbon), affect local, regional, and global air quality, climate forcing, and carbon-nitrogen cycle, and promote the formation of ozone and acid rain environmental problems [8][9][10].
China is one of the greatest sources of BB emissions, predominantly due to anthropogenic burning, such as post-harvest agricultural open biomass burning (OBB) [11]. The main form of BB in China is crop straw burning. With the continuous increase in comprehensive agricultural production levels, China's total straw output is increasing to become The BTH region is located in northern China (Figure 1a), which is one of the most economically developed, populated, and polluted regions in the country. The region is surrounded by the Bohai Sea in the east, the Taihang Mountains in the west, Yanshan Mountains in the north, and the North China Plain in the south. The terrain is high in the northwest and low in the southeast, inclined from northwest to southeast. The rich geomorphic types include plateaus, mountains, hills, plains, grasslands, and coastal geomorphic types. The region has a temperate continental monsoon climate, with hot and rainy summers and cold, dry winters when temperature inversion occurs, while the spring and autumn seasons are short, windy, and rainless. The climate is affected by the structure of the Yanshan-Taihang Mountains, which are characterized by terrain that gradually decreases from northwest to southeast and act as a barrier to the dominant wind direction in the region. The frequent occurrence of calm winds and inversion weather is not conducive to the diffusion of atmospheric pollutants. The BTH region serves as the core area of the Bohai Rim Economic Circle and is characterized by high energy consumption, high pollution emissions, and complex air pollution.

GFED4 Data Set
The fourth version of the Global Fire Emissions Database (GFED4) [17] was analyzed in this study, which was downloaded from http://www.globalfiredata.org/index.html (accessed on 26 December 2021). GFED4 provides monthly burned area, emissions such as fire carbon (C) and dry matter (DM), and the contribution of different fire types to these emissions in order to calculate trace gas and aerosol emissions using emission factors [18,19]. The burned area data set provides global, monthly burned area at 0.25 • spatial resolution from mid-1995 through December 2016 and higher temporal resolution daily burned area for a subset of the time series extending back to August 2000. Note that GFED4 was primarily derived from the now-obsolete Collection 5.1 MODIS MCD64A1 burned area product, which was superseded by the Collection 6 MCD64A1 product in early 2017 [19,20].

Land Use and Land Cover
The International Geosphere-Biosphere Project (IGBP) classification scheme of MCD12Q1 data at a spatial resolution of 500 m

GFED4 Data Set
The fourth version of the Global Fire Emissions Database (GFED4) [17] was analyzed in this study, which was downloaded from http://www.globalfiredata.org/index.html (accessed on 26 December 2021). GFED4 provides monthly burned area, emissions such as fire carbon (C) and dry matter (DM), and the contribution of different fire types to these emissions in order to calculate trace gas and aerosol emissions using emission factors [18,19]. The burned area data set provides global, monthly burned area at 0.25 • spatial resolution from mid-1995 through December 2016 and higher temporal resolution daily burned area for a subset of the time series extending back to August 2000. Note that GFED4 was primarily derived from the now-obsolete Collection 5.1 MODIS MCD64A1 burned area product, which was superseded by the Collection 6 MCD64A1 product in early 2017 [19,20].

Land Use and Land Cover
The International Geosphere-Biosphere Project (IGBP) classification scheme of MCD12Q1 data at a spatial resolution of 500 m (https://lpdaac.usgs.gov/products/mcd12q1v006/ (accessed on 18 January 2022)) was selected to assist in obtaining the underlying surface type of the vegetation sources. Figure 1b covered the research area and was mosaiced and reprojected using HEG tools. The major land use reclassification vegetation types were reclassified as forests (evergreen needleleaf forest, evergreen broadleaf forest, deciduous needleleaf forest, deciduous broadleaf forest, mixed forest), savannas (woody savannas and savannas), shrublands (closed shrublands and open shrublands), croplands (croplands and cropland-natural vegetation mosaic), water bodies (snow and ice), wetlands, urban areas, and barren or sparsely vegetated in this study.

Digital Elevation Model
The digital elevation model (DEM) data set from the Shuttle Radar Topography Mission (SRTM) was used to analyze the impact of elevation on dust aerosols and may be downloaded from https://srtm.csi.cgiar.org/srtmdata/, accessed on 26 December 2021, then preprocessed with ArcGIS.

Meteorological Data
In order to analyze the change of environmental parameters and influence factors, we obtained the key meteorological data (temperature, precipitation, wind speed, wind direction, and sunlight between 2003 and 2020) collected by 22 meteorological stations in the BTH region from the China Meteorological Data Service Center (http://data.cma.cn (accessed on 26 December 2021), as shown in Figure 2.

Temporal and Spatial Patterns of BB Fire Counts in BTH
The number of BTH fire counts from 2003 to 2020 showed a fluctuating upward trend, which can be divided into two periods, as shown in Figure 3a

Temporal and Spatial Patterns of BB Fire Counts in BTH
The number of BTH fire counts from 2003 to 2020 showed a fluctuating upward trend, which can be divided into two periods, as shown in Figure 3a. The first period is a continuous fast-rising stage from 2003 to 2013, peaking in 2013 with the large amount of 6700 BB fire counts. In 2003 and 2004, the low amount of BB fire counts was less than 2000 and then continued to increase. There was only one downward trend, which occurred in 2008, with 2976 BB fire counts. In this year, due to holding the 2008 Olympic Games, the surrounding urban environment was well controlled. The total number of fire counts in the first period was 40,682, accounting for 49.79% of the total fire counts. The second period is the fluctuating decline period, and the time range is 2014-2020 (50.21%), which maintains a high amount of BB fire counts above 5000 and a rise and fall between 5088 and 6519 by 1-2 years. The average number of BB fire counts in the second period (5851) is 1.58 times that in the first period (3698). Overall, it can be seen that from 2003 to 2020, the number of BB fire counts generally showed an increasing trend; the annual data of BB fire counts can be seen in Table 1.
The monthly BB fire counts time series for each region is shown in Figure 3b. The BB fire counts mostly appeared in March (accounting for 11%), June-July (33%), and October (9.68%). The minimum BB fire counts were distributed in December (2%) and January (2%). The winter season remained at a low level, with no significant change. Climate, terrain, weather, and other factors affect the amount of BB fire counts, which responded differently at different times and seasons. From 2003 to 2020, the fewest monthly BB fire counts were in January 2006, with only 21 BB fire counts. In contrast, June had the largest monthly BB fire counts each year, and 2017 is the most prominent, with 2237 BB fire counts, which is almost more than a full year of data in 2003 and 2004. The BB fire counts increased steadily from January to March, and the BB fire counts in April and May were relatively stable. The growth rate increased rapidly from May to June. At this time, the temperature increased, the thunderstorm season increased, and the winter wheat harvest season arrived. The number of fire counts decreased significantly from June to September, then increased again in October, and decreased again in November and December. At this time, winter starts, the weather becomes cold, and the number of fire counts decreases significantly. Figure 3c shows the changing trend of the daily average from 2003 to 2020. There are three peaks in March, June, and October, which is consistent with the conclusion of the monthly analysis.
Moreover, at the subregion scale in BTH, the total number of BB fire counts in Hebei is the largest, with far more than the other two cities in BTH, as shown in Figure    We also analyzed the data of burned area in the same period, which began in 2003 and ended in 2016, because GFED4 did not process later years, so the information is not available. The burned area product is a digital map, at full resolution, of the extent of surfaces burned during a period of time and includes the burned area itself and information about the temporal pattern of the fire activity.  Table 1. Based on the monthly burned area data, as shown in Figure 4b, we know that the burned area in January was the monthly lowest, with a burned area of 174 km 2 , accounting for 0.63% of the total burned area for the whole period. The burned area in June was the monthly highest, with a burned area of 8517 km 2 , accounting for 30.74% of the total monthly burned area. The burned area in June 2007 was the largest, 1165 km 2 , accounting for 55.24% of the total annual burned area in 2007. For the season, the burned area had the same trend as BB fire counts, which was low in winter and high in summer. The burned area was 5021 km 2 in spring, accounting for 18.12% of the total burned area, 15,478 km 2 (55.87%) in summer, 6360 km 2 (22.96%), and 846 km 2 (3.03%) in winter. For the BTH subregional analysis, the annual largest burned area in Beijing was in 2011, with a burned area of 524.4 km 2 , accounting for 13.16% of the total burned area. The annual smallest burned area in Beijing was in 2003, with a burned area of 40.65 km 2 , accounting for 1.03% of the total burned area. The largest annual burned area in Tianjin was in 2016, with a burned area of 622.68 km 2 , accounting for 14.21% of the total burned area in Tianjin from 2003 to 2016, and the annual smallest burned area in Tianjin was in 2004, accounting for 1.59% of the total burned area. The largest burned area in Hebei was 3550.8 km 2 , accounting for 14.83% of the total burned area. The annual smallest burned area in Hebei was in 2003, with a burned area of 400.19 km 2 , accounting for 1.67% of the total burned area. According to the data, it can be seen that the burned area of Hebei far exceeds that of Beijing and Tianjin, and is even greater than the sum of the burned areas of Beijing and Tianjin. Hebei has the largest burned area, followed by Tianjin and Beijing. The annual data of burned areas can be seen in Table 1.     The volatility is basically in line with the timing of agriculture. Winter wheat (planted in mid-October, harvested at the end of May) and summer maize (planted in mid-June, harvested at the end of September) are the two most important crops in the zone. Among them, the peak of the fires in June resulted from the harvest of winter wheat. To increase the soil fertility for the next cultivation, the wheat residue is burned after harvest. The biomass burning in October-November may be attributed to the maturity of the corn and subsequent burning of corn stalks, but it was not as concentrated as in June and lower than June in the total area of combustion.

Temporal and Spatial Patterns of BB Carbon Emissions in BTH
Atmosphere 2022, 13, x FOR PEER REVIEW 10 of 15 October, harvested at the end of May) and summer maize (planted in mid-June, harvested at the end of September) are the two most important crops in the zone. Among them, the peak of the fires in June resulted from the harvest of winter wheat. To increase the soil fertility for the next cultivation, the wheat residue is burned after harvest. The biomass burning in October-November may be attributed to the maturity of the corn and subsequent burning of corn stalks, but it was not as concentrated as in June and lower than June in the total area of combustion. According to the analysis of three subregions in BTH, the total value of BB carbon emissions is 4.6321Tg, Beijing, Tianjin, and Hebei are 0.6178 Tg (accounting for 13.34%), 0.604 Tg (13.03%), and 3.4103 Tg (73.62%), respectively. It is inferred that the BB carbon emission is the highest in Hebei province and the lowest in Tianjin. The annual data can be seen in Table 1  According to the analysis of three subregions in BTH, the total value of BB carbon emissions is 4.6321Tg, Beijing, Tianjin, and Hebei are 0.6178 Tg (accounting for 13.34%), 0.604 Tg (13.03%), and 3.4103 Tg (73.62%), respectively. It is inferred that the BB carbon emission is the highest in Hebei province and the lowest in Tianjin. The annual data can be seen in Table 1. Among them, only Beijing reached the maximum carbon emission in 2011, and both Tianjin and Hebei reached the maximum emission in 2016. For the analysis of forests, Tianjin's carbon emissions from forests were not significant. Overall, Tianjin had roughly the same trend in carbon emissions as Hebei. The changing trend of Beijing was roughly the same as that of the other two provinces in 2011, and it entered a period of volatility after 2011. It is proved once again that the data of the three regions are consistent with the total region, with the highest emissions in June and fewer emissions in winter.
In addition to the analysis of the inter-annual variation of BB carbon emissions at specific periods and in specific regions, this paper also counted the four major vegetation types of fires and calculated the corresponding contribution ratios, as shown in Figure 5a. For an improved understanding of the quantitative relationship of annual variation among fire counts and BB carbon emissions, the linear regression was analyzed for four zones, as shown in Figure 6. The results revealed a positive correlation between the fire counts and carbon emissions, whether in a subregion or the total area of BTH. The correlation coefficients of Beijing, Tianjin, Hebei, and BTH are 0.79, 0.81, 0.84, and 0.89, respectively. Of course, it can be understood that higher carbon emissions are affected by more fire counts. In addition, the increase in emissions is related to the hot weather in the current month, the frequent occurrence of forest fires, and the emission of pollutants from straw burning.
Atmosphere 2022, 13, x FOR PEER REVIEW 11 of consistent with the total region, with the highest emissions in June and fewer emissio in winter.
In addition to the analysis of the inter-annual variation of BB carbon emissions specific periods and in specific regions, this paper also counted the four major vegetati types of fires and calculated the corresponding contribution ratios, as shown in Figure  Four  For an improved understanding of the quantitative relationship of annual variati among fire counts and BB carbon emissions, the linear regression was analyzed for fo zones, as shown in Figure 6. The results revealed a positive correlation between the f counts and carbon emissions, whether in a subregion or the total area of BTH. The cor lation coefficients of Beijing, Tianjin, Hebei, and BTH are 0.79, 0.81, 0.84, and 0.89, resp tively. Of course, it can be understood that higher carbon emissions are affected by mo fire counts. In addition, the increase in emissions is related to the hot weather in the cu rent month, the frequent occurrence of forest fires, and the emission of pollutants fro straw burning.

BB Fire Counts and Carbon Emissions in 2020 in BTH
Through the introduction in the previous sections, we have a general understanding of the overall trend of fire counts and carbon emissions in BTH. Now we focus on analyzing the situation in BTH in 2020 during the COVID-19 outbreak. First, the time and space analysis of fire counts in 2020 was carried out. The monthly maximum fire counts in BTH were in July 2020, with 847, accounting for 16.65% of the total fire counts in 2020. It exceeded 700 in March, May, June, and July, which is closely related to crop harvesting and burning straw and the climate of high temperature and rain in the summer. The month with the lowest fire counts was December. The number of fire counts was 52, accounting for 1.02% of the total number of fire counts in the whole year. This is because the temperature is low in winter, the weather is cold, it is not easy to cause fire, the land is wet and cold in winter, and the straw is not easy to burn. By region, Tianjin is quite different from the general trend. The fire counts in Tianjin were the highest in March and April but lower in June and July. It shows that the fire counts in Tianjin are closely related to crop planting and straw burning. Beijing was controlled well in June, increased suddenly in July, and decreased from August to December. The fire counts trend of Beijing in June 2020 is different from the overall trend from 2003 to 2020. The overall trend is the peak of fire counts in June, which also shows that with progress over time, the problem of fire count concentration in June has begun to improve. Hebei is the region with the highest fire counts emissions in BTH, of which March, May, June, and July have the peak fire counts in a year. In the highest month, Tianjin fire counts do not exceed 100, Beijing does not exceed 40, while the highest value of Hebei fire counts is as high as 700. For space, most of the fire counts in BTH are in the south and east of Hebei and the east of Tianjin. The fire counts in Beijing are sparse, and most of them are in the North China Plain and Taihang Mountains, as shown in Figure 7.
Through the introduction in the previous sections, we have a general understanding of the overall trend of fire counts and carbon emissions in BTH. Now we focus on analyzing the situation in BTH in 2020 during the COVID-19 outbreak.
First, the time and space analysis of fire counts in 2020 was carried out. The monthly maximum fire counts in BTH were in July 2020, with 847, accounting for 16.65% of the total fire counts in 2020. It exceeded 700 in March, May, June, and July, which is closely related to crop harvesting and burning straw and the climate of high temperature and rain in the summer. The month with the lowest fire counts was December. The number of fire counts was 52, accounting for 1.02% of the total number of fire counts in the whole year. This is because the temperature is low in winter, the weather is cold, it is not easy to cause fire, the land is wet and cold in winter, and the straw is not easy to burn. By region, Tianjin is quite different from the general trend. The fire counts in Tianjin were the highest in March and April but lower in June and July. It shows that the fire counts in Tianjin are closely related to crop planting and straw burning. Beijing was controlled well in June, increased suddenly in July, and decreased from August to December. The fire counts trend of Beijing in June 2020 is different from the overall trend from 2003 to 2020. The overall trend is the peak of fire counts in June, which also shows that with progress over time, the problem of fire count concentration in June has begun to improve. Hebei is the region with the highest fire counts emissions in BTH, of which March, May, June, and July have the peak fire counts in a year. In the highest month, Tianjin fire counts do not exceed 100, Beijing does not exceed 40, while the highest value of Hebei fire counts is as high as 700.   According to the C emissions of BTH, the emissions of CO 2 were the highest, with the monthly average accounting for 8.3% of the cumulative CO 2 , the emissions of particulate matter were the lowest, and its monthly average also accounted for 8.3% of the cumulative emissions. Among them, the cumulative emissions of C were 201 (Gg), the cumulative emissions of CO were 39 Gg, the cumulative emissions of CO 2 were 670 Gg, the cumulative emissions of CH 4 were 2 Gg, the cumulative emissions of dry matter were 417 Gg, and the cumulative emissions of particulate matter were 3 Gg. The six substances are roughly proportional from the broken line diagram. The emissions of particulate matter were relatively stable, and the other five emissions were the highest in March. It showed a downward trend from March to May and a small peak in June. This is roughly the same as fire counts emissions in 2020. Combined with the start of 2020 in BTH, COVID-19 began to increase its emissions.
From a subregional perspective, in 2020, the fire counts in Beijing rose steadily from February to April, as shown in Figure 8a. In terms of CO 2 emissions, March was still the highest peak, and the emissions were proportional. The emissions of CO 2 were the highest, and the emissions of particulate matter were the lowest. June was the inflection point of the decline of fire counts, and its C emissions also fell to a low point. In September, C emissions ushered in an upward period again, which is connected with agricultural straw burning. From a sub-regional point of view, Tianjin's fire counts have a certain correlation with carbon emissions (Figure 8b). March was the highest for fire counts, and its C emissions also reached the highest value, of which CO 2 emissions were still high, and particulate emissions were the lowest. There are two inflection points in Tianjin. At the same time, the number of fire counts decreased in June, while C emissions are still increasing, and the fire counts increased in August, but C emissions are decreasing.
In 2020, the C emissions in Hebei were roughly the same as that in BTH Figure 8c. March had the highest value of CO 2 emissions, which decreased from March to May. In June, straw burning increased again due to hot weather, elevated temperature, and mature crops. In 2020, there was no significant correlation between fire counts and C emissions after March. The fire counts coincided with the C emissions and peaked in March. Hebei sits on the North China Plain, with flat terrain and hot summers, causing fires to flare up in June and July. In addition, comparing the three regions, the fire counts of Beijing and Tianjin showed a downward trend in June, while the fire counts of Hebei were still rising in June. For carbon emissions, the three provinces reached the highest value in March, in which CO 2 emissions accounted for the highest value of total carbon emissions, and particulate matter accounted for the lowest value of total carbon emissions.

Conclusions
By using GFED4 data, this research attempted to explore the spatial and temporal distribution of biomass burning in the study area from 2003 to 2020, the corresponding proportion of different vegetation types, and the causes of inter-annual changes in BTH. Hence, we can draw the following conclusions: Based on the analysis of fires counts in BTH from 2003 to 2020, the number of fire counts in 2013 was the highest, and fire counts in 2003 were the lowest. In terms of months, due to the cold winter in the north, December and January had the lowest fire counts. The temperature increased in June, and the number of fire counts reached the maximum during the high incidence period of thunderstorm season. In terms of sub-regions, Hebei had the highest fire counts, Tianjin had the second highest, and Beijing had the least. Combining the characteristics of burned areas in BTH, the burned area in 2016 was the largest, accounting for 14.20% of the total burned area, and the burned area in 2003 was the smallest, accounting for 1.66% of the total burned area. The burned area is the lowest in January and the highest in June. In January, due to the winter in BTH, temperatures are low, the forests, grasslands, and agricultural lands are covered with ice and snow, which makes it not easy to cause fire, and the winter is cold. Farmers hardly burn straw in a large area in this season. In June, it is hot and rainy. There is not only a lot of straw burning, but heavy rain often brings lightning and fire.
According to the carbon emissions characteristics of BTH, the total annual emissions from 2003 to 2018 were 6.4679Tg, with the lowest emissions in 2003 and the highest emissions in 2011. According to the analysis of months and seasons, the most carbon emissions are in June and the least in January. Our results indicate that this is mainly due to the influence of seasons. March also shows a small peak of carbon emissions, which is closely related to factory emissions and agricultural straw burning in BTH.
Through a separate study for the year 2020, we found that because BTH is located in the North China Plain and the terrain is flat, it is easy to cause fire count aggregation, and its fire count peak areas are concentrated in the North China Plain and Taihang Mountains. Due to the influence of 2020 and the factors of factories in BTH, the carbon emissions still reached the highest value in March, and the highest fire counts were in the summer in July. The cumulative amount of fire counts in 2020 was 5088, and the cumulative emissions of carbon, CO, CO 2 , CH 4 , dry matter, and particulate matter from biomass burning in BTH reached 201 Gg, 39 Gg, 670 Gg, 2 Gg, 417 Gg, and 3 Gg in 2020, respectively.
The methodology and results from this research provide a useful reference for policymakers to better understand the characteristics and variations of biomass burning in BTH and lay the groundwork for simulating and predicting the impact of biomass burning on air quality in the next work. Accordingly, more effective measures to monitor and control biomass burning in these areas can be posed and carried out to enhance local air quality and protect human health.

Author Contributions:
The study was completed with cooperation among all authors: R.X. conceived and designed the research topic. Y.Z. and P.W. processed data and wrote the manuscript. Z.X. and L.W. collaborated in discussing the manuscript and providing editorial advice. All authors have read and agreed to the published version of the manuscript.