1. Introduction
To meet the growing demand for food, nations across the world have implemented state-of-the-art biotechnology and agricultural expansion methods [
1,
2]. Yet, these efforts concurrently resulted in an upsurge in crop residue outputs (maize stalks, rice straw, wheat straw, etc.) [
3,
4]. Despite concerted efforts aimed at maximizing the utilization of crop residues (e.g., as compost, natural fertilizer, livestock feed, edible fungus substrates, and biomass fuel pellets) and promoting their integration into soil ecosystems, the substantial volume of these residues presents a formidable challenge in achieving complete clearance from croplands. As a result, farmers often employ open-field burning, the most cost-effective and convenient method, to clear crop residues before tillage. The concentrated and unregulated practice of open-field crop residue burning (OCRB) directly releases large amounts of pollutants into the atmosphere, e.g., nitrogen oxide (NOx), volatile organic compounds (VOCs), and aerosols, which, in turn, results in severe air pollution events [
5,
6,
7].
In order to protect the environment and residents’ health, numerous countries have implemented a series of policies and adopted heterogenous control measures to mitigate fire events [
8,
9]. These control measures primarily include traditional regular patrols [
10,
11], the installation of advanced electronic monitoring equipment [
12], the utilization of unmanned aerial vehicles [
13,
14], and the establishment of watchtowers [
15]. However, the above methods have inherent limitations, i.e., they are time-consuming and labor-intensive. The management challenges are particularly prominent in remote areas or at the rural–urban interface, where the frequency of inspections is often inadequate, resulting in an elevated risk of fires [
16]. Furthermore, these measures often lack the ability to offer a macroscopic perspective on regional fire management and prevention, hindering the effective development of collaborative strategies across nearby regions.
With cutting-edge advancements in technology, the combination of geographic information systems with satellite remote sensing has opened up innovative possibilities for precise fire monitoring and control. Many countries and regions around the world have established fire management platforms (
Table 1), which mainly integrate multisource satellite and meteorological forecast datasets. Based on their functionalities, these platforms can offer helpful information to both government agencies and the public via the visualization of fire spots, fire emissions, and burned areas and enable the assessment of fire weather conditions and air quality status. For instance, the Canadian Wildland Fire Information System and European Forest Fire Information System provide near-real-time fire information on, e.g., fire spots, fire weather, and fine fuel moisture. However, these platforms are mainly utilized for monitoring forest fires (also called wildfires) and fail to distinguish between different types of fire spots or specifically manage OCRB. Unlike wildfires, it is important to note that OCRB is mainly influenced by farmers’ agricultural practices; that is, OCRB is easily impacted by human behavior and hard to predict. Despite the existence of platforms designed for OCRB monitoring, their functionality typically centers around the visualization of fire spots and is regionally constrained [
17,
18,
19]. Furthermore, the lack of open-source availability restricts broader public access and applicability.
To assist local governments and relevant stakeholders in understanding the spatiotemporal variations of fire spots and emissions in near real time or real time, as well as provide comprehensive information (e.g., regarding weather and air quality conditions, crop residue coverage, burning windows, and fuel loads) for the scientific and precise control of OCRB, this study is dedicated to integrating multiple-source data and developing an open-source OCRB-monitoring platform, which can also be utilized for other countries. This technical note is organized as follows:
Section 2 describes the architecture of the platform and different data sources and preprocessing methods;
Section 3 showcases the visual analyses and major functions of the platform; and
Section 4 presents the concise conclusions and future perspectives.
2. Data and Methods
2.1. Architecture of the Platform
AgriFireInfo (version 1.0) is an online webpage based on Web-GIS technology. The architecture of this platform is illustrated in
Figure 1, which includes five modules. The data acquisition module retrieves diverse datasets, encompassing satellite fire spot data from multiple sources, fire emission data, satellite images, and Global Forecast System (GFS) data. The numerical simulation module integrates the Weather Research and Forecasting (WRF) model along with the Fertilizer Emission Scenario Tool (FEST), generating outputs used for forecasting burning windows and calculating crop residue yields, respectively. The data-preprocessing module extracts fire spots associated with croplands using land cover data (accessible at:
https://www.resdc.cn/DOI/DOI.aspx?DOIid=54, last accessed: 26 November 2023). Additionally, it leverages the GDAL and GeoPandas libraries to resample Sentinel satellite products and fire emission data and convert the resampled data, GFS data, and model-simulated results into GeoJSON format. Furthermore, the module employs the matplotlib and pyecharts libraries for visualization, facilitating interactive operations on web pages. The data storage module is responsible for managing data operations such as storage, retrieval, and updating while also ensuring data security. It plays a crucial role in guaranteeing timely access to the database when users submit their requests. The data needed for this platform are stored using Alibaba Cloud Storage Service. The visualization module leverages HTML, CSS, and JavaScript as the primary languages for performing the demanding task of data visualization. The interactive display function was mainly created via the JavaScript toolkits OpenLayers and Cesium. The platform was deployed on
https://agrifireinfo.github.io (accessed on 15 January 2023).
Currently, AgriFireInfo v1.0 is tailored to delivering five core functionalities: (1) an interactive presentation of multi-source satellite fire spots and fire emissions data, (2) an interactive visualization of meteorological conditions and air quality status, (3) an interactive presentation of fire radiative energy across various prefecture-level cities, (4) an interactive presentation of Prevention Alarming Index data, and (5) an interactive display of “burning windows” and crop residue yields.
2.2. Data Sources
Near-real-time fire spot data: The AgriFireInfo v1.0 incorporates the near-real-time active fire products obtained from both sun-synchronous (Terra/Aqua, Suomi-NPP, NOAA-20) and geostationary satellites (Himawari-9). The near-real-time sun-synchronous satellites’ fire products can be obtained from the NASA FIRMS website (
https://firms.modaps.eosdis.nasa.gov/active_fire, last access: 26 November 2023) via a registered account. To access fire spot data from Himawari-9, users need to register an account and obtain permission through the official website (
https://www.eorc.jaxa.jp/ptree/index.html, last access: 26 November 2023).
To ensure timely updating of the platform, shell scripts were utilized to automate the downloading of active fire spot data from relevant websites. As certain fire spots were not associated with OCRB, we first deleted non-agricultural fire spots from the original CSV file based on land cover data. Secondly, each fire spot in the file was assigned a low, medium, or high confidence value. We further removed fire spots with low confidence levels from various satellite data sources, including those labeled “l” in VIIRS data, “<30%” in MODIS data, and “1” in the Himawari-9 data. Thirdly, the observation times in the files were converted from Coordinated Universal Time (UTC) to the local time. It should be noted that solely presenting fire spots on a webpage cannot offer a precise representation of their geographical locations. Therefore, an additional step was taken, leveraging the Python language and Baidu API, to match the latitude and longitude information with specific geographic locations (township scale). This allows users to directly obtain detailed geographic information on the targeted fire spots.
Near-real-time fire emission inventory data: To help government authorities understand the spatial distribution characteristics of OCRB emissions, the Fire INventory from NCAR (FINN) was incorporated into the AgriFireInfo v1.0 platform. The daily emission file can be directly downloaded from the cited website (
https://www.acom.ucar.edu/acresp/MODELING/finn_emis_txt/, last access: 26 November 2023). We first extracted fire emissions data related to OCRB based on the GENVEG information of the file. Then, the GeoPandas library was used to generate a fishnet with a grid cell size of 0.25°, and OCRB emissions were merged into the respective grid cells for each day. Finally, the data were stored in the GeoJSON format.
Meteorological and air quality data: Real-time surface temperature, relative humidity, wind speed, wind direction, and precipitation data were retrieved from GFS and can be accessed via the cited website (
https://nomads.ncep.noaa.gov/dods/, last access: 26 November 2023). The daily averaged snow cover data, with a spatial resolution of 5 km, were obtained from the National Snow and Ice Data Center (available at:
https://nsidc.org/home, last access: 26 November 2023). Real-time air quality tile maps can be accessed using the following link (
https://aqicn.org/api/, last access: 26 November 2023) by applying for a token ID. This platform currently integrates seven air quality variables, namely, air quality index (AQI), PM2.5, PM10, CO, NO
2, SO
2, and O
3.
Numerical simulation data: The AgriFireInfo v1.0 platform also incorporates the forecasted results from the WRF and FEST-C models. Utilizing the WRF outputs, the “burning windows” were calculated, considering various meteorological factors, and then visualized in PNG format. Moreover, the FEST-C model is capable of generating annual crop yields. Using the specific straw-to-grain ratios for each crop, this platform calculated crop residue yields across different regions.
2.3. Fire Radiative Energy
Fire radiative energy (
FRE) is a value estimated by temporally integrating fire radiative power (
FRP). Laboratory and field measurements show that
FRE has a significant linear relationship with fuel consumption and burning intensity [
20,
21]. To understand the near-real-time OCRB practices and the hourly burning intensity of OCRB in different prefecture-level cities, Himawari-9 Level 3 fire products were adopted to calculate hourly
FRE values. We extracted the fire spots within each prefecture-level city using a shapefile and calculated the hourly
FRE by integrating
FRP over the occurrence of OCRB events in that hour; we then summed the hourly
FRE in city c using Equation (1).
Here, FRP (mean) is the mean FRP value of the monitored agricultural fire spots (MW), N is the number of detections in one hour (1 ≤ N ≤ 6), and 10 min represents the temporal resolution of the fire datasets, which is then translated to seconds (600 s). φ_(hr,c) represents the number of fire spots observed per hour (h) and city (c).
The FRE data from various prefecture-level cities in a province over the past 48 h were merged into a CSV file for subsequent charting. We utilized the pyecharts library for visual rendering and the depiction of the charts. This platform adopts two types of charts to provide a dynamic interactive display: a bar chart with a timeline and a line chart with tab labels.
2.4. Prevention Alarming Index
A near-real-time understanding of crop residue coverage (
CRC) plays a crucial role in improving the precision of OCRB management and optimizing resource allocation. In this study, we introduce a novel Prevention Alarming Index (PAI) based on
CRC to evaluate OCRB risk levels, aiming to help local governments enhance management efficiency. With the advancement of satellite remote sensing technology, satellite imagery enables the estimation of
CRC [
22,
23]. The unique spectral absorption characteristics of crop residue at 2100 nm can effectively differentiate it from bare soil surfaces [
24]. By establishing a regression relationship between
CRC and the Normalized Difference Tillage Index (
NDTI) [
25], a spatial distribution map of
CRC can be easily generated.
To estimate
CRC, we adopted a regression equation from [
26].
Upon determining the spatial distribution of
CRC, the levels of PAI were linearly classified based on the
CRC values. A higher
CRC value indicates a greater PAI level. Then, we used crop-type data [
27] along with the GDAL and GeoPandas libraries to extract cropland area grids. Finally, the data were processed into a GeoJSON format.
3. Results and Discussion
Considering the actual needs of frontline administrative personnel, the design of a platform should prioritize simplicity and cost-effectiveness. This will contribute to enhancing practicality and accessibility, better addressing the challenges arising from OCRB.
Section 3.1,
Section 3.2,
Section 3.3,
Section 3.4 and
Section 3.5 provide a detailed introduction to the functionalities of the platform in question, using Chinese regions as examples.
3.1. Displaying Crop Residue Fire Spots and Emission Patterns
To assist environmental departments in identifying areas with high OCRB activity and prioritizing efforts to mitigate air pollution, this platform integrates and presents near-real-time data on fire spots and pollutant emission amounts. It facilitates the identification of cropland areas at risk of experiencing fires and enables the prediction of fire spread patterns. This useful function significantly enhances the efficiency of OCRB management efforts, offering valuable insights for policymakers to quickly respond to OCRB events.
The interface for the visualization of near-real-time and/or historical fire spots is illustrated in
Figure 2a–d. It features a div-based tooltip in the bottom-right corner of the webpage, providing crucial information such as image types and the sources of the satellite data (like NOAA-20, Suomi-NPP, Terra/Aqua, and Himawari-9). This webpage includes two types of online tile layers: a street view map from Gaode (
Figure 2a) and a satellite image map from OpenStreetMap (
Figure 2b). Users have the flexibility to choose their preferred map source, and the default mode is the Gaode street view map. Moreover, users can easily zoom in and out to focus on their targeted prevention regions. Considering the practical needs of the relevant authorities during OCRB management processes, this platform not only showcases near-real-time fire spots but also offers a feature allowing users to dynamically visualize the historical fire spot data for the past seven days.
Figure 2c depicts the platform’s ability to simultaneously display fire spot data from multiple satellites. This function enables a combined exhibition of fire data both from polar-orbiting and geostationary satellites and offers the geographical coordinates of fire spots to administrative personnel for supervision, inspection, and the provision of evidence for penalties, as shown in
Figure 2d.
Besides showcasing fire spots, this platform has the ability to render near-real-time pollutant emissions from OCRB events. The species of pollutants involved include PM2.5, PM10, CO, NO2, CH4, and VOCs. AgriFireInfo v1.0 adopts self-adaptive color gradients to provide accurate and region-specific representations of pollutant emission amounts. This assists users in rapidly identifying high-OCRB-emission areas (
Figure 2e,f). Furthermore, near-real-time wind direction data can also be incorporated into its visualizations. Government authorities can use these data to identify potential polluted downwind areas and formulate plans for control and prevention measures. To enhance its utility, users have the option to add more fire emission inventories based on their requirements, such as GFAS or GFED.
3.2. Visualizing Meteorological Conditions and Air Quality
Meteorological conditions not only influence the occurrences of OCRB but also significantly affect regional air pollution levels [
28,
29,
30]. Moreover, air quality status is a critical contributory factor for shaping OCRB control strategies due to its direct impact on environmental and human health [
31,
32]. A real-time understanding of meteorological conditions and air quality status is essential for developing effective strategies for preventing illegal burning activities and reducing exposure to adverse air quality conditions for sensitive individuals, such as those suffering from chronic obstructive pulmonary disease (COPD). The AgriFireInfo v1.0 platform meets these practical needs by providing real-time visualization capabilities for both meteorological conditions and air quality status.
This platform leverages global GFS data to provide users with detailed meteorological information. Users have the flexibility to incorporate additional high-spatiotemporal-resolution meteorological data, such as data derived from advanced weather models like the WRF model, the Regional Atmospheric Modeling System (RAMS), and the Fifth-Generation Penn State/NCAR Mesoscale Model (MM5). To ensure the adaptability of the platform, one can change the meteorological parameters that affect the occurrence of OCRB in a specific region, e.g., snow cover or wind speed. Moreover, users can also set different thresholds to explicitly analyze meteorological impacts on OCRB. For instance, areas with less snow cover (<20%) are excluded from visualization (
Figure 3a) due to the limited impact of snow cover on burning activities.
Currently, seven air quality parameters, AQI, PM2.5, PM10, CO, NO2, SO2, and O3, are considered in the AgriFireInfo v1.0 platform, and the Air Quality Health Index (AQHI) will be incorporated in upcoming versions. With this platform’s timely updates, users are ensured access to the latest air pollution levels of their selected variables (as shown in
Figure 2c,d). Unlike the Prescribed Fire Information System developed by the authors of [
33], who utilized an air quality model to forecast and assess the impacts of prescribed burning on regional air quality and human health in the southeastern United States, the AgriFireInfo v1.0 platform does not yet possess predictive capabilities for assessing air quality during OCRB periods. The occurrence of OCRB strongly depends on anthropogenic factors such as farmers’ burning behaviors and the enforcement of burning ban policies, adding further complexity and uncertainty to air quality forecasting work. Fortunately, recent studies have indicated the promising potential of using machine learning to predict OCRB occurrences [
34,
35]. Future developments for the platform aim to provide an API to integrate numerical forecasts from both machine learning and Eulerian models (e.g., CMAQ and WRF-Chem).
3.3. Evaluating the Daily Variations in Burning Intensities
While various platforms can exhibit the spatial distribution of fire spots and fire emissions [
36,
37], solely displaying fire spots or emissions cannot provide detailed evaluations of farmers’ daily burning behaviors and burning intensities in different regions. This limitation underscores the necessity for adopting more comprehensive solutions. In response, AgriFireInfo v1.0 enriches its functionality by integrating near-real-time
FRE data, thereby improving the ability to assess and understand agricultural burning practices. By addressing this gap, this platform not only aids in the implementation of effective control measures but also facilitates collaboration with adjacent regional governments, ultimately contributing to a more informed and coordinated approach to mitigating the environmental impacts of OCRB.
The AgriFireInfo v1.0 features a bar chart that displays the hourly variations in
FRE across various cities in a province/region (
Figure 4a). This interface allows for the swift evaluation of burning intensity within a single city domain and facilitates comparative analysis to pinpoint which cities are facing severe OCRB. Moreover, this platform also offers a time navigation tool that enables users to retrospectively examine
FRE data over the past 48 h and analyze
FRE trends. In addition to comparing the burning intensities across different cities, it is essential to visually understand the characteristics of
FRE within each city or region. This facilitates the timely monitoring and tracking of changes in local farmers’ burning behaviors and further supports efforts to prohibit OCRB.
Figure 4b shows a time series of
FRE from April 5 to April 6 2023 in Suihua (a typical agricultural area in Heilongjiang Province). One can quickly glean the variation trend of the
FRE values, including the peak burning time and duration patterns. Besides OCRB management, the hourly variations in
FRE values can also be utilized as a time allocation profile in the preprocessing of fire emission inventories, contributing to enhancing the forecasting accuracy of air quality models. Future enhancements that will add a function allowing the modeling community to download historical and near-real-time
FRE data are planned.
While platforms like Virtual Fire (
http://virtualfire.aegean.gr/, accessed on 22 January 2024) effectively utilize line charts to visualize meteorological parameters, there is an opportunity to enhance user experience by adding interactive display features. Similarly, platforms such as EFFIS (
https://effis.jrc.ec.europa.eu/apps/effis.statistics/, accessed on 22 January 2024) and CWFIS (
https://gwis.jrc.ec.europa.eu/apps/gwis.statistics/, accessed on 22 January 2024) provide certain interactive visualizations of burned areas and fire spots. However, expanding their temporal resolution beyond weekly or seasonal scales could further enhance management capabilities. To the best of our knowledge, besides AgriFireInfo v1.0, no previously developed platforms offered interactive visualization of hourly
FRE changes, especially for OCRB. It should be noted that this module relies on retrieved data from geostationary satellites. One can set satellite data sources based on local availability, for instance, the MSG satellite data for Europe and Africa or the GOES satellite data for the Americas.
3.4. Assessing the Regions with High Prevention Alarming Index VALUES
Traditional inspection methods for prohibiting OCRB are often constrained in their ability to provide comprehensive insights into regions at risk of OCRB occurrences and are also time-consuming. Considering the limitations associated with these traditional inspection methods, this platform innovatively introduces a parameter named the Prevention Alarming Index (PAI). This index is calculated based on CRC, offering additional insights into regions more prone to OCRB. It can also be used by agricultural departments to check and fix the comprehensive utilization ratio of crop residues. The display of this value enhances governments’ awareness of OCRB prevention levels and supports more-targeted measures for promoting comprehensive utilization.
Figure 5a illustrates a demonstration of the PAI map over Northeast China on 20 November 2018. On this webpage, the open-source Cesium library is used for the rendering of PAI layers. The AgriFireInfo v1.0 classifies the PAI into five levels according to
CRC. These levels are categorized as Low, Medium-low, Medium, Medium-high, and High, respectively. The PAI layer dynamically adapts to a terrain’s elevation, allowing for an immersive 3D visual representation. A binocular view function is featured in the bottom-right corner of the interface. This function offers users the ability to navigate regions in need of heightened inspections using VR headsets or AR glasses (
Figure 5b). By rotating their head-mounted headsets (
Figure 5c), users can interactively change their viewpoints, gaining a thorough, immersive understanding of the spatial distribution of PAI values in different areas. However, the current integration of WebVR into Cesium scenes is limited to three degrees of freedom (DOF) and cannot support holographic displays with six DOF. Future works intend to optimize the Cesium library to support six DOF, thereby improving the sense of spatial immersion and interaction.
In order to enhance layer-rendering performance, we resampled the PAI layer into various resolutions and analyzed the corresponding file sizes and load times across different browsers. As shown in
Table 2, the results indicate that as the layer resolution increases, the load time for different browsers progressively lengthens. For example, when rendering the PAI layer at a 1000 m resolution, Chrome exhibits a load time of 4.63 s, while at a 200 m resolution, the load time increases to 111.68 s. Moreover, the load times observed on mobile devices are substantially longer compared to those on desktop browsers. At higher resolutions (300 m and 200 m), mobile devices fail to load the layers. Considering the balancing of file size, resolution, and browser load times, this platform adopts a resolution of 500 m for the PAI layer. At this resolution, the file size is manageable at 39.1 MB, and while the load times increase compared to those for lower resolutions, they remain within the acceptable limits for efficient rendering and a positive user experience. Other users also have the flexibility to adjust the layer resolution based on server performance and specific requirements.
3.5. Regulating Controlled Crop-Residue-Burning Activities
Scientific controlled burning is an effective method for dealing with crop residues [
38,
39]. However, the implementation processes require the consideration of meteorological conditions and the quantity of crop residues. To aid relevant authorities in taking scientific OCRB management measures and ensuring regional air quality at good or excellent levels, this platform offers burning window and crop residue yield (fuel load) data for carrying out controlled OCRB activities. The integration of these data provides decision-makers with the information required to conduct controlled OCRB activities, thereby contributing to improving air quality and sustainable agricultural practices.
As mentioned in the Data and Methods section, the crop production values output by FEST-C were utilized to calculate residue yields.
Figure 6a,b displays the crop residue yields of four widely cultivated crops: maize, rice, soybean, and wheat. This information, combined with the spatial distribution of burning windows, enables stakeholders to precisely plan controlled burning activities. Users are provided with the option to select specific areas for visualizing the forecasted available burning windows over the next 48 h. Furthermore, a swipe control function was applied in AgriFireInfo v1.0, allowing users to slide left or right to display either the gridded or county-level burning windows. In this platform, the burning windows at the county level are computed based on the ratio of cultivated land area within gridded burning windows to the total cultivated land area in a county. Using this ratio, AgriFireInfo v1.0 categorizes values into five levels, set at 0.0, 0.25, 0.50, 0.75, and 1.0. When the ratio is 0.0 (depicted in red), the implication is that a county is unfavorable for conducting OCRB activities. For values above 0, the darker the color, the greater the suitability for controlled OCRB activities within a county. Users can also quickly access the time and suitability for controlled burning in their counties by sliding the timeline at the bottom of the webpage. Note that the meteorological parameters selected for calculating burning windows on a remote cloud server were specifically demoed for Northeast China. If the platform is to be applied in other areas, these parameters for computing the burning windows would require dynamic adjustments.
At present, platforms related to controlled burning primarily focus on forest/shrubland and offer valuable resources for fire management. For example, the PFIRS platform (
https://ssl.arb.ca.gov/pfirs/cb3/cb3.php, accessed on 22 January 2024) provides daily updates on burning windows, although it does not include details about fuel load. On the other hand, the Flint Hills Smoke Management Tool (
http://ksfire.sonomatechdata.com/view/details/, accessed on 22 January 2024) displays fuel conditions, categorizing them into three types: light, medium, and heavy. However, this tool can be further enhanced by incorporating more precise information about different fuel types. Taking into account the advanced functionalities of forest fire management platforms and the special demands for OCRB control, the AgriFireInfo v1.0 platform not only exhibits burning window information at the hourly scale but also offers precise crop residue yield data. A potential way of improving this platform is to incorporate trajectory models such as HYSPLIT. By setting areas for controlled burns as receptor points, users can rapidly determine the areas affected by controlled burning activities.
4. Conclusions
The direct burning of crop residues in open fields is a common and convenient practice that is favored by farmers, but it always precipitates severe air pollution events. This has exerted immense pressure on OCRB management work, necessitating the urgent adoption of advanced methods to enhance monitoring and prevention capabilities, optimize resource allocation, and ensure precise control.
Unlike former platforms or tools designed primarily for forest/shrubland fire control, the open-source AgriFireInfo v1.0 platform leverages Web-GIS technology and incorporates multi-source data and new functions specifically tailored for the monitoring and management of OCRB. The integration of fire spots and fire emission data from diverse satellites, along with meteorological conditions and air quality statuses, provides a holistic analysis of OCRB, facilitating effective decision making by government authorities and stakeholders. A key innovation of this platform is its incorporation of the Prevention Alarming Index (PAI), which enables the identification of areas with a high coverage of crop residues. This feature not only helps environmental departments to detect areas with a high risk of OCRB but also supports agricultural departments in checking the comprehensiveness of crop residue utilization work. Furthermore, this platform’s capacity to visualize burning windows and crop residue yields significantly contributes to the planning of scientific controlled-burning activities and effectively mitigates the harmful impacts of OCRB. Moreover, AgriFireInfo v1.0 allows users to analyze variations in farmers’ burning behavior and intensity across diverse regions. This feature is particularly beneficial in terms of understanding regional burning patterns and developing refined, region-specific strategies for managing OCRB.
The contribution of this platform extends beyond immediate OCRB management: its open-source architecture ensures that it can be adapted and utilized in other regions or countries, making it a versatile and valuable resource for governments and stakeholders worldwide. In future developments, our objective is to continuously enhance this platform by endowing it with practical features that bolster decision-making processes. One significant enhancement will entail the implementation of predictive scenario simulations via remote cloud computing and machine learning. These simulations aim to address the existing knowledge gap pertaining to optimal timing, geographical locations, responsible parties, and the permissible quantities of crop residue for controlled burning. In doing so, the objective is to facilitate the precise planning of controlled burning activities while minimizing adverse effects on air quality and the health of local inhabitants.