Large-scale forest fire spreading simulation method based on FARSIDE model and WRF-SFIRE model
Through the coordinated use of FARSITE and WRF-SFIRE models, combined with satellite observation data to optimize the combustible material model parameters, the problems of discomfort in parameter default values and human factors in large-scale forest fire spread simulation are solved, and a higher precision forest fire spread is achieved Simulation and fire behavior characteristic prediction provide reliable information support for forest fire prevention and control.
Patent Information
- Application Number
- CN202411161352.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-22
AI Technical Summary
The existing technology is difficult to accurately reflect the fire spread under natural conditions in forest fire spread simulations, especially in large-scale forest fires. The default value of the model parameter is not applicable to southwest China, and the accuracy evaluation based on combustion profiles is susceptible to artificiality. Factors influence, resulting in great uncertainty in simulation.
The large-scale forest fire spread simulation method of collaborative satellite observation based on FARSITE and WRF-SFIRE models was adopted to determine the global sensitivity of the combustible material model parameters through the Sobol algorithm, and the sensitive parameters were optimized using SC coefficients and relative root mean square errors, combined with non-dominant sorting The genetic algorithm performs multi-objective optimization, improves simulation accuracy, and obtains more reliable forest fire spread results through WRF-SFIRE model coupling.
The accuracy of spatial simulation and fire behavior characteristics prediction of forest fire spread simulation has been improved, the reliability of large-scale forest fire spread simulation has been enhanced, and key information support is provided for precise prevention and control of forest fires.
Smart Images

Figure CN119004837A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of forest fire early warning, and in particular to a multi-objective optimization method for simulating large-scale forest fire spread in coordination with satellite observation. Background Art
[0002] Using fire behavior models to simulate forest fire spread can not only timely assess the development trend of forest fires to better serve fire rescue decisions, but is also an important means of predicting potential forest fire behavior characteristics (such as fire spread speed, fire intensity, etc.) and extreme fire behaviors (such as crown fire, flying fire, etc.) at the landscape scale. Accurately simulating the spread of forest fires is of great significance for clarifying the process of fire occurrence and development. It also helps to build a reliable forest fire behavior case library to improve the accuracy of forest fire risk assessment and serve the precise prevention and control of forest fires.
[0003] Fuel load and moisture content are key factors affecting the development of forest fires. They have been incorporated into the parameters of the fuel model and can be used in fire behavior models such as FARSITE and WRF-SFIRE to simulate fire spread. At present, there are two types of commonly used fuel models: the Anderson fuel model (a total of 13 types) and the Scott and Burgan fuel model (a total of 43 types). The latter has a more detailed classification of fuel types. These standard fuel models correspond to a set of fuel parameters or attributes, but the default values given are not necessarily applicable to southwest China. Therefore, the fuel model parameters can be preliminarily determined in combination with field fuel surveys to improve the reliability of forest fire spread simulation results as much as possible. In terms of fire spread simulation accuracy evaluation, most studies only consider the consistency between fire spread simulation profiles based on satellite observations of fire burning profiles. This accuracy evaluation method based on burning profiles is easily affected by human factors, resulting in the fire spread situation observed by satellites not reflecting the fire spread situation under natural conditions. In addition to the fire burning profile, the thermal radiation released during the fire spread process is also worthy of attention and is closely related to the size of the fuel load. Satellite remote sensing technology has the ability to monitor fire radiation intensity over a large range and with high precision, which provides a new reference for evaluating the accuracy of forest fire spread simulation and can be used for optimizing fuel model parameters.
[0004] In addition, the FARSITE model is more suitable for simulating the spread of mesoscale forest fires, but if it is used for large-scale forest fire spread, the large amount of water vapor and heat released by the combustion of combustibles will be enough to change the local atmospheric field, resulting in greater uncertainty in the simulation of forest fire spread. The WRF-SFIRE model takes into account the atmosphere-fire coupling effect and is suitable for large-scale forest fire spread simulation, but the fire behavior module involved is relatively simple. Combining the respective advantages of the two, the fire spread results simulated by the FARSITE model are used to drive the WRF-SFIRE model to carry out forest fire spread simulations, in order to obtain more reliable large-scale forest fire spread simulation results, which can provide key information decision-making support for the precise prevention and control of major and high-intensity forest fires. Summary of the invention
[0005] The technical problem to be solved by the present invention is to provide a multi-objective optimization method for simulating large-scale forest fire spread in coordination with satellite observation, which has strong scalability and simple calculation.
[0006] The technical solution adopted by the present invention to solve the above technical problems is: a large-scale forest fire spread simulation method based on the FARSITE model and the WRF-SFIRE model, comprising the following steps:
[0007] Step 1: Determine the fire area and collect the parameters required for the FARSITE model and the WRF-SFIRE model to simulate the spread of forest fires. The input parameters include air temperature, relative humidity, wind speed and direction, rainfall, elevation, slope, aspect, forest canopy structure parameters and combustible model;
[0008] Step 2: Multi-objective optimization of forest fire spread simulation based on FARSITE model;
[0009] Step 2.1: FARSITE model parameterization;
[0010] Determine the fuel model of the target area based on the GLC_FCS30 land cover product; use the Sobol algorithm to determine the global sensitivity of each parameter of the fuel model; set the interval range of sensitive parameters based on forest stand and fuel information survey data, and set weakly sensitive or insensitive parameters as default values;
[0011] Step 2.2: Extraction of forest fire behavior features based on satellite observations;
[0012] Based on the Global Fire Atlas dataset, the initial fire location is determined, and the burning contour information of the fire is extracted. The SC coefficient is used to evaluate the spatial simulation accuracy of forest fire spread. The calculation method of the SC coefficient is formula 1. Based on the VIRSNRT active fire point product, the fire radiation intensity characteristics of the burning area are extracted. The relative root mean square error is used to evaluate the simulation accuracy of fire behavior characteristics during forest fire spread. The calculation method of the relative root mean square error rRMSE is formula 2.
[0013]
[0014] Among them, S obs is the fire burn area extracted from the GFA dataset, S s is the fire burning area simulated by FARSITE model, S o is the area of overlapping fire burning areas;
[0015]
[0016] in, and are the fire radiation intensities simulated by the FARSITE model and observed by satellite, respectively, corresponding to the i-th pixel, and n is the number of satellite fire point pixels falling within the FARSITE model simulated burning contour;
[0017] Step 2.3: Optimize sensitive parameters;
[0018] The SC coefficient and rRMSE were used to simultaneously evaluate the accuracy of forest fire spread simulation under specific parameters; based on the non-dominated sorting genetic algorithm II, the number of individuals in the population was set to m, the maximum evolutionary generation was set to n, and the sensitivity parameters of the FARSITE model were optimized to obtain the optimal results of forest fire spread simulation with a higher SC coefficient and a lower rRMSE;
[0019] The optimal forest fire spread simulation results are shown in SC opt Coefficients and rRMSE opt Formula (3) should be satisfied;
[0020]
[0021] Among them, d i is a comprehensive evaluation index that considers the SC coefficient and rRMSE during the spread of the i-th forest fire;
[0022] Step 3: Simulate large-scale forest fire spread based on the optimized WRF-SFIRE model;
[0023] Step 3.1: WRF-SFIRE model parameterization;
[0024] According to the GLC_FCS30 land cover product, the fuel model of the target area is divided into grassland and forest, and the optimal fuel model parameters calibrated in step 2.3 are used as the fuel model parameters of WRF-SFIRE; the domain of the WRF-SFIRE model is set to 4 layers of nesting, the spatial resolution of the fourth layer of atmospheric grid is 300 meters, and the spatial resolution of the fire grid is 30 meters;
[0025] Step 3.2: Simulate large-scale forest fire spread based on the WRF-SFIRE model;
[0026] The optimal forest fire spread simulation results obtained based on the FARSITE model are used as the initial fire location information of the WRF-SFIRE model. The WRF-SFIRE model is used to carry out large-scale forest fire spread simulation, and forest fire behavior characteristics such as fire spread speed and fire intensity are output. To ensure the reliability of forest fire spread simulation, the above process is repeated every hour.
[0027] Furthermore, the sensitive parameters described in step 2.1 are 1-h / 10-h / 100-h surface dead combustible moisture content and load, live herb and wood combustible moisture content and load, and combustible bed thickness, and the insensitive parameters are combustible surface area to volume ratio, combustible calorific value, and combustible extinction moisture content.
[0028] Furthermore, the initial fire pixel described in step 2.2 should satisfy the requirement that its burning date is the smallest relative to other pixels in the fire scene, and the fire start time and end time are determined by the earliest and latest times of the VIIRS NRT fire point pixels.
[0029] Furthermore, the confidence level of the VIIRS NRT fire point pixels used for rRMSE calculation in step 2.2 should be "h" or "n".
[0030] Further, m and n described in step 2.3 are 30 and 50 respectively;
[0031] Furthermore, h in step 3.2 is 6.
[0032] Beneficial effects of the invention: The multi-objective optimization method for simulating large-scale forest fire spread of the invention is simple to operate. First, the multi-objective optimization of model parameters is used to simultaneously improve the spatial simulation of forest fire spread and the prediction accuracy of fire behavior characteristics. Then, based on the coupling of two fire spread models, the reliability of large-scale forest fire spread simulation is further improved. The method can provide key information support such as fire situation for precise prevention and control of forest fires. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 It is a schematic diagram of the overall method flow of the present invention.
[0034] Figure 2 It is a forest fire case location map in a specific implementation manner of the present invention.
[0035] Figure 3 It is the result of sensitivity analysis of combustible model parameters in a specific implementation scheme of the present invention.
[0036] Figure 4 This is the optimal simulation result of forest fire spread by the FARSITE model in the specific implementation scheme of the present invention.
[0037] Figure 5 It is the simulation result of large-scale forest fire spread by WRF-SFIRE model in the specific implementation scheme of the present invention. DETAILED DESCRIPTION
[0038] The multi-objective optimization method for simulating the spread of large-scale forest fires with satellite observation provided by the present invention is further described below in conjunction with specific embodiments and the accompanying drawings of the specification:
[0039] A multi-objective optimization method for large-scale forest fire spread simulation based on satellite observations, such as Figure 1 As shown, the following steps are included:
[0040] Step 1: Fire behavior model input data preparation.
[0041] Select the forest fire in Qiaodi Township, Xichang City, Sichuan (2014.04.20) (see Figure 2 ) for case analysis.
[0042] It includes two parts:
[0043] Step 1.1: FARSITE model input data preparation (see Figure 1 ). Hourly air temperature, relative humidity, wind speed and direction, rainfall and other parameters were extracted from ERA5-Land meteorological reanalysis data; elevation, slope, aspect and other parameters were extracted from GDEMV3 products; forest canopy height was extracted from the global forest canopy height product published by Peter Potapov; canopy coverage was extracted from GFCC products; canopy height under branches was obtained by fitting canopy height and leaf area index; canopy bulk density was calculated by canopy fuel load and canopy height under branches; the spatial resolution of all parameters was resampled to 30 meters;
[0044] Step 1.2: Preparation of WRF-SFIRE model input data. The original meteorological data used is the ERA5 reanalysis data, with a temporal resolution of 1 hour and a spatial resolution of 0.25°. The main meteorological elements of each pressure layer, as well as the meteorological elements of the surface layer and the moisture content of the soil layer are used (see Figure 1 ); the required elevation (DEM) data is extracted from the GDEMV3 product (see Figure 1); the required soil type and other geographic data can be downloaded from https: / / www2.mmm.ucar.edu / wrf / src / wps_files.
[0045] Step 2: Multi-objective optimization of forest fire spread simulation based on FARSITE model;
[0046] Specifically, it includes 3 parts:
[0047] Step 2.1: Parameterization of FARSITE model. Based on the GLC_FCS30 land cover product, the fuel model of the study area was divided into two categories: grassland and forest. Using the Sobol algorithm, the sensitive parameters were determined as 1-h / 10-h / 100-h surface dead fuel moisture content and load, live herbaceous and woody fuel moisture content and load, and fuel bed thickness. The insensitive parameters were fuel surface area to volume ratio, fuel calorific value, and fuel extinguishing moisture content (see Figure 3 );
[0048] Step 2.2: Extraction of forest fire behavior characteristics based on satellite observations. Extract fire burning contour information from the Global Fire Atlas (GFA) dataset and determine the initial fire location. Based on the fire burning contour information observed by satellite, calculate the SC coefficient (Sorensen Coefficient, Formula 1) to evaluate the spatial simulation accuracy of forest fire spread. Based on the VIIRS NRT active fire point information with medium and high confidence, calculate the fire radiation intensity (RI, kW / km 2 ) to evaluate the simulation accuracy of fire behavior characteristics during forest fire spread.
[0049]
[0050] Among them, S obs is the fire burn area extracted from the GFA dataset (km 2 ), S s is the fire burning area simulated by FARSITE model (km 2 ), S o is the area of overlapping fire burning areas (km 2 ).
[0051]
[0052] in, and are the fire radiation intensities simulated by the FARSITE model and observed by satellite, respectively, corresponding to the i-th pixel, and n is the number of satellite fire point pixels falling within the FARSITE model simulated burning contour;
[0053] Step 2.3: Optimization of sensitive parameters. Based on the non-dominated sorting genetic algorithm II (NSGA-II), the number of individuals in the population was set to 30, the maximum number of evolution generations was set to 50, and the SC coefficient and rRMSE were optimized. The optimal FARSITE model sensitivity parameters and the corresponding forest fire spread simulation results were selected in combination with formula (3). The results show that the forest fire spread simulation accuracy is higher when the optimized combustible model parameters are used (see Figure 4 ), i.e. SC = 0.807, rRMSE = 4.80%. Compared with the standard Scott and Burgan combustible model parameters, i.e. grassland (GR2) and forest (TU1), the accuracy of forest fire spread simulation using this combustible model is relatively low, i.e. SC = 0.756, rRMSE = 19.9%. Although the use of the standard combustible model can also obtain a fire burning profile consistent with satellite observations, there is still a large uncertainty in the prediction of fire radiation intensity, which shows that the present invention can obtain more reliable forest fire spread simulation results by comprehensively considering the SC coefficient and rRMSE indicators.
[0054]
[0055] Among them, d i is a comprehensive evaluation index that considers the SC coefficient and rRMSE during the spread of the i-th forest fire;
[0056] Step 3: Optimization of large-scale forest fire spread simulation based on WRF-SFIRE model;
[0057] Specifically, it includes 2 parts:
[0058] Step 3.1: Parameterization of WRF-SFIRE model. According to the GLC_FCS30 land cover product, the combustible model of the study area is divided into two categories: grassland and forest, and the optimal combustible model parameters calibrated in step 2.3 are used as the combustible model parameters of WRF-SFIRE; the domain of the WRF-SFIRE model is set to 4 layers of nesting, the spatial resolutions of the 4 layers of atmospheric grids are 27 km, 9 km, 3 km, and 300 m, respectively, and the spatial resolution of the fire grid is 30 m; the main physical process parameterization schemes of the WRF-SFIRE model are: microphysical process - WSM3 type scheme, longwave radiation - RRTM scheme, shortwave radiation - Dudhia scheme, near-surface layer - Monon-Obukhov scheme, land surface process - Noah-MP land surface process scheme, planetary boundary layer - YSU scheme, layer 4 is closed, cumulus parameterization - Kain Fritsch scheme, layer 3 / 4 is closed;
[0059] Step 3.2: Forest fire spread simulation based on WRF-SFIRE model. The results of FARSITE model simulation are used as the initial fire line and directly input into the WRF-SFIRE model to carry out large-scale forest fire spread simulation in the next 6 hours. The fire spread results simulated by FARSITE model at 18:00 on April 20 are used as the initial fire line, and the WRF-SFIRE model is used to simulate the fire spread characteristics from 18:00 on April 20 to 0:00 on April 21, such as Figure 5 As shown, the results show that the simulation results of large-scale forest fire spread based on the WRF-SFIRE model in a short period of time are relatively reliable.
Claims
1. A large-scale forest fire spread simulation method based on the FARSITE model and the WRF-SFIRE model, comprising the following steps: Step 1: Determine the fire area and collect the parameters required for the FARSITE model and the WRF-SFIRE model to simulate the spread of forest fires. The input parameters include air temperature, relative humidity, wind speed and direction, rainfall, elevation, slope, aspect, forest canopy structure parameters and combustible model; Step 2: Multi-objective optimization of forest fire spread simulation based on FARSITE model; Step 2.1: FARSITE model parameterization; Determine the fuel model of the target area based on the GLC_FCS30 land cover product; use the Sobol algorithm to determine the global sensitivity of each parameter of the fuel model; set the interval range of sensitive parameters based on forest stand and fuel information survey data, and set weakly sensitive or insensitive parameters as default values; Step 2.2: Extraction of forest fire behavior features based on satellite observations; Based on the Global Fire Atlas dataset, the initial fire location is determined, and the burning contour information of the fire is extracted. The SC coefficient is used to evaluate the spatial simulation accuracy of forest fire spread. The calculation method of the SC coefficient is formula 1. Based on the VIRSNRT active fire point product, the fire radiation intensity characteristics of the burning area are extracted. The relative root mean square error is used to evaluate the simulation accuracy of fire behavior characteristics during forest fire spread. The calculation method of the relative root mean square error rRMSE is formula 2. Among them, S obs is the fire burn area extracted from the GFA dataset, S s is the fire burning area simulated by FARSITE model, S o is the area of overlapping fire burning areas; in, and are the fire radiation intensities simulated by the FARSITE model and observed by satellite, respectively, corresponding to the i-th pixel, and n is the number of satellite fire point pixels falling within the FARSITE model simulated burning contour; Step 2.3: Optimize sensitive parameters; The SC coefficient and rRMSE were used to simultaneously evaluate the accuracy of forest fire spread simulation under specific parameters; based on the non-dominated sorting genetic algorithm II, the number of individuals in the population was set to m, the maximum evolutionary generation was set to n, and the sensitivity parameters of the FARSITE model were optimized to obtain the optimal results of forest fire spread simulation with a higher SC coefficient and a lower rRMSE; The optimal forest fire spread simulation results are shown in SC opt Coefficients and rRMSE opt Formula (3) should be satisfied; Among them, d i is a comprehensive evaluation index that considers the SC coefficient and rRMSE during the spread of the i-th forest fire; Step 3: Simulate large-scale forest fire spread based on the optimized WRF-SFIRE model; Step 3.1: WRF-SFIRE model parameterization; According to the GLC_FCS30 land cover product, the fuel model of the target area is divided into grassland and forest, and the optimal fuel model parameters calibrated in step 2.3 are used as the fuel model parameters of WRF-SFIRE; the domain of the WRF-SFIRE model is set to 4 layers of nesting, the spatial resolution of the fourth layer of atmospheric grid is 300 meters, and the spatial resolution of the fire grid is 30 meters; Step 3.2: Simulate large-scale forest fire spread based on the WRF-SFIRE model; The optimal forest fire spread simulation results obtained based on the FARSITE model are used as the initial fire location information of the WRF-SFIRE model. The WRF-SFIRE model is used to carry out large-scale forest fire spread simulation, and forest fire behavior characteristics such as fire spread speed and fire intensity are output. To ensure the reliability of forest fire spread simulation, the above process is repeated every hour.
2. A large-scale forest fire spread simulation method based on the FARSITE model and the WRF-SFIRE model as claimed in claim 1, characterized in that: The sensitive parameters described in step 2.1 are the 1-h / 10-h / 100-h moisture content and load of dead surface combustibles, the moisture content and load of living herbaceous and woody combustibles, and the thickness of the combustible bed. The insensitive parameters are the combustible surface area to volume ratio, the calorific value of the combustibles, and the moisture content of the combustibles at extinction.
3. A large-scale forest fire spread simulation method based on FARSITE model and WRF-SFIRE model as claimed in claim 1, characterized in that: The initial fire pixel described in step 2.2 should satisfy the requirement that its burning date is the smallest relative to other pixels in the fire field. The fire start and end time are determined by the earliest and latest time of the VIIRS NRT fire point pixels.
4. A large-scale forest fire spread simulation method based on FARSITE model and WRF-SFIRE model as claimed in claim 1, characterized in that: The confidence level of the VIIRS NRT fire pixels used for the rRMSE calculation described in step 2.2 should be "h" or "n".
5. The large-scale forest fire spread simulation method based on the FARSITE model and the WRF-SFIRE model as claimed in claim 1, characterized in that: m and n as stated in step 2.3 are 30 and 50, respectively.
6. A large-scale forest fire spread simulation method based on FARSITE model and WRF-SFIRE model as claimed in claim 1, characterized in that: The h mentioned in step 3.2 is 6.
Citation Information
Patent Citations
Forest fire risk assessment method based on Maxent and GIS
CN112712275A
Forest fire scene disposal risk assessment method
CN115841014A
Forest fire prevention situation spreading multivariate model prediction method and system
CN115907073A
Forest fire spreading real-time simulation method, fire extinguishing decision-making method and early warning system
CN116305832A
Forest outbreak fire risk assessment method based on forest fire spreading simulation and Bayesian network
CN118195322A
Cited By
Fire spread integrated prediction method and system coupled with smoke and weather feedback effects
CN121031138A
Forest fire spreading simulation system and method based on WRF and ForeFire bidirectional coupling
CN121920072A
