Wild fire risk evaluation method
By building the optimal wildfire susceptibility model and evaluating environmental vulnerability and rescue capabilities, and using a variety of machine learning models to optimize, the problem of incomplete wildfire risk assessment in the existing technology is solved, and the accuracy and scientificity of the evaluation is improved. More effective preventive measures were made.
Patent Information
- Application Number
- CN202411341356.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-25
- Publication Date
- 2024-12-31
AI Technical Summary
The existing wildfire risk research fails to fully consider wildfire susceptibility, environmental vulnerability and rescue capabilities, resulting in inappropriate model parameters setting, affecting the accuracy and effectiveness of the assessment results.
By constructing an optimal wildfire susceptibility model, combining the environmental data of the target area and the spatiotemporal laws of historical wildfire distribution, computing environmental vulnerability and rescue capabilities, using a variety of machine learning models for optimization, and selecting the optimal model for wildfire risk assessment.
It improves the accuracy and scientific nature of wildfire risk assessment, provides more objective prevention suggestions, and reduces the possibility of wildfire occurrence.
Smart Images
Figure CN119228127A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of forest wildfire risk assessment, and in particular to a wildfire risk assessment method. Background Art
[0002] As an important part of the ecosystem, forests play an important role in the carbon cycle system and biodiversity conservation. Fire is one of the main natural disasters faced by the forest system. It has occurred frequently under the influence of extreme weather conditions in recent years. The occurrence of fire will quickly release the carbon stored in plants, reduce the carbon sink function of the ecosystem, and cause great damage to environmental health. The occurrence of forest fires poses a huge threat to the ecological environment and the safety of people's lives and property. In the face of wildfire threats, corresponding measures should be formulated to reduce wildfire risks.
[0003] Current wildfire risk research has not taken into account the following: ① Most studies assess wildfire risk from a single perspective. ② Most studies lack spatial analysis of wildfires themselves. When setting model parameters, most of them are set based on experience or use the model default parameter values. Model parameters have a great impact on model convergence speed and performance. If the appropriate hyperparameter optimization method is not used to tune the model hyperparameters, the model performance may not be optimized. The final model result may not be the optimal solution, which is not conducive to the assessment of wildfire risk. Summary of the invention
[0004] In order to overcome the defects existing in the above-mentioned prior art, the purpose of the present invention is to provide a wildfire risk assessment method, aiming to provide a better method for wildfire prevention and thus reduce the occurrence of wildfires.
[0005] In order to achieve the above-mentioned object of the present invention, the present invention provides a wildfire risk assessment method, comprising the following steps:
[0006] An optimal wildfire susceptibility model is constructed based on the environmental data of the target area and the temporal, spatial and geographical laws of historical wildfire distribution, and the wildfire susceptibility of the target area is evaluated using the optimal wildfire susceptibility model;
[0007] Calculate the environmental vulnerability of the target area;
[0008] Calculate rescue capabilities within the target area;
[0009] A wildfire risk model is constructed based on the wildfire susceptibility assessment results, environmental vulnerability assessment results and rescue capabilities of the target area to evaluate the wildfire risk in the target area.
[0010] This wildfire risk assessment method not only considers wildfire susceptibility but also the environmental vulnerability and rescue capabilities of the target area, making the final evaluation results more in line with objective facts, providing more accurate suggestions and guidance for wildfire prevention, and reducing the possibility of wildfire occurrence.
[0011] Optionally, the steps to construct an optimal wildfire susceptibility model are:
[0012] Collect climate characteristics, geographic characteristics, vegetation characteristics, and historical wildfire information in the target area, and pre-process all data;
[0013] The kernel density calculation is performed using historical wildfire data, and the temporal, spatial and geographical laws of historical wildfire distribution are analyzed based on the calculation results.
[0014] All data are divided into data sets according to time units, wildfire data are matched with other feature data one by one, and all data except fire points are normalized;
[0015] Wildfire susceptibility models were constructed based on a variety of machine learning models;
[0016] Based on the sci-kit-opt imi ze framework, the wildfire susceptibility models constructed by various machine learning models were optimized to achieve the optimal solution for each model. The optimal solutions of all models were compared, and the model with the highest accuracy was selected as the optimal wildfire susceptibility model.
[0017] Optionally, historical wildfire information of the target area is obtained through wildfire data observed by the fire department and / or various satellites in the target area, and bad data in the historical wildfire data is screened out to obtain the historical wildfire information of the target area.
[0018] Optionally, the climate characteristic data of the target area is obtained through the meteorological station in the target area, and the obtained information includes specific values and data time;
[0019] Obtain the geographic feature data of the target area by querying the National Geographic Information Public Service Platform;
[0020] The historical images of the target area were downloaded through the GEE platform, and the vegetation characteristic data of the target area were calculated in the ENVI software.
[0021] Optionally, the kernel density calculation tool in ArcMap is used to calculate the historical wildfire data, and the analysis results are used to obtain the spatiotemporal and geographic patterns of the historical wildfire distribution in the target area.
[0022] Optionally, an environmental vulnerability model of the target area is constructed according to the ecological remote sensing index of the target area; and the RSE I value is calculated based on the remote sensing image in the target area and the environmental vulnerability model of the target area to evaluate the environmental vulnerability of the target area.
[0023] Optionally, the steps for calculating the rescue capability within the target area are:
[0024] Five factors affecting rescue capability, namely, the distance to the road, the distance to the river, the distance to the fire station, the distance to the hospital and the distance to the densely populated area, were selected, and the rescue capability model of the target area was constructed using the GIS-MCDM method.
[0025] The rescue capability in the target area is calculated based on the rescue capability model of the target area.
[0026] Optionally, wildfire risk = target area wildfire susceptibility × target area environmental vulnerability × rescue capacity within the target area.
[0027] The beneficial effects of the present invention are:
[0028] The wildfire risk assessment method provided by the present invention comprehensively evaluates the wildfire risk from three aspects: wildfire susceptibility of the target area, environmental vulnerability of the target area, and rescue capabilities within the target area. It is more comprehensive, more practical, and more scientific than the existing technology. In addition, the method uses machine learning to build a model, adopts multiple machine learning models for construction and optimization, and selects the optimal model to evaluate wildfire susceptibility, thereby further improving the accuracy of wildfire risk assessment.
[0029] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0031] Figure 1 It is a schematic diagram of the principle of the present invention;
[0032] Figure 2 It is a flow chart of the first embodiment. DETAILED DESCRIPTION
[0033] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0034] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.
[0035] Embodiment 1
[0036] like Figure 1 and Figure 2 As shown, this embodiment provides a wildfire risk assessment method, comprising the following steps:
[0037] Assess wildfire susceptibility in target areas
[0038] An optimal wildfire susceptibility model is constructed based on the environmental data of the target area and the spatiotemporal and geographical laws of historical wildfire distribution, and the wildfire susceptibility of the target area is evaluated using this optimal wildfire susceptibility model.
[0039] Specifically, the environmental data of the target area include but are not limited to: climate characteristics, geographical characteristics, and vegetation characteristics in the target area. Among them, the acquisition of climate characteristics: the climate data of the target area, such as temperature, rainfall, humidity, wind speed, etc., can be obtained through the meteorological station in the target area, and the information obtained includes specific values and data time; the acquisition of geographical characteristics: the geographical data of the target area, such as DEM, slope, slope direction, curvature, etc., can be obtained by querying the National Geographic Information Public Service Platform; the acquisition of vegetation characteristics: the historical images of the target area can be downloaded through the GEE platform, and the vegetation data, such as the Normalized Difference Vegetation Index NDVI, can be calculated in the ENVI software, and the tree classification information in the area can be collected.
[0040] The historical wildfire information of the target area can be collected through satellite product data and / or target area fire station data. If the historical wildfire data comes from satellite product data, such as MODIS, VII RS, etc., first, download the required wildfire data from the NASA official website according to the required time, then filter out the fire points with high confidence based on the confidence field; then filter out the fields that may be vegetation fires based on the type field. If the historical wildfire data comes from the target area fire station, obtain the wildfire location information and the time when the wildfire occurred.
[0041] After all the data are collected, they are preprocessed to filter out bad data. The screening criteria are as follows: if the historical wildfire data source is satellite product data, delete the wildfire samples at the boundary of the target area and those that do not belong to the target area, as well as the satellite product data with insufficient confidence or the fire type is not vegetation fire; if the historical wildfire data is fire station record data, delete the wildfire data at the boundary of the target area and those that do not belong to the target area.
[0042] As shown in Table 1, the slope, aspect, curvature, terrain wetness index, and water flow intensity index are all geographical characteristics of the target area and can be calculated in arcmap using DEM data. All the screened data are clipped according to the target area mask to obtain all the data that fully conforms to the target area.
[0043] Table 1
[0044]
[0045] The processed historical wildfire data is subjected to kernel density calculation, and the spatiotemporal and geographical laws of the historical wildfire distribution are obtained according to the calculation results. In this embodiment, the kernel density calculation tool of the arcmap software is used to calculate the historical wildfire data, and the analysis results obtain the spatiotemporal and geographical laws of the historical wildfire distribution in the target area.
[0046] All data were divided into data sets according to time units, and historical wildfire data were matched one-to-one with other characteristic data. Specifically, the Extract Value to Point tool was used in the ArcMap software to extract the climate characteristics, geographical characteristics, and vegetation characteristics of each fire point location, so that historical wildfire data were matched one-to-one with other characteristic data, and all data except the fire point were normalized.
[0047] Wildfire susceptibility models are constructed based on multiple machine learning models. In this embodiment, four representative machine learning models are selected to construct wildfire susceptibility models, namely RF, XGBoost, LR, and SVM. When constructing the wildfire susceptibility model, non-fire point samples that are equal to or close to 1:1 with the number of historical wildfire points are selected in the target area, and the random point selection tool is used in arcmap. Using the machine learning model, the wildfire sample is specified as 1 and the non-fire point sample is specified as 0. 80% of the samples are randomly selected as the training set, and the remaining 20% are used as the test set.
[0048] Based on the various tuning algorithms (Bayesian optimization, random search, and grid search) contained in the scikit-optimize framework, the hyperparameters of the wildfire susceptibility models constructed by each machine learning model are tuned in turn to achieve the optimal solution for each model. The optimal solutions of all models are compared, and the model with the highest accuracy is selected as the optimal wildfire susceptibility model.
[0049] Assess the environmental vulnerability of the target area
[0050] In this embodiment, the environmental vulnerability of the target area is evaluated by calculating the RSEI value based on the remote sensing image in the target area. Specifically, MODIS or other satellite remote sensing images can be downloaded from the GEE platform to calculate the normalized vegetation index NDVI of the target area as the greenness index, as well as the temperature index WET, the heat index LST and the dryness index NDBSI, and the PCA method is used to calculate the RSEI value, and the formula is as follows:
[0051] RSEI0=f PCA[f(NDVI,LST,NDBSI,WET)]
[0052]
[0053] Where RSEI0 is the preliminary RSEI value obtained by PCA; RSEI0 min is the minimum value among the preliminary RSEI values, RSEI0 max is the maximum value among the preliminary RSEI values.
[0054] Evaluate rescue capabilities within the target area
[0055] In this embodiment, the steps of calculating the rescue capability in the target area are:
[0056] Five factors affecting rescue capability, namely, the distance to the road, the distance to the river, the distance to the fire station, the distance to the hospital, and the distance to the population gathering area, were selected, and the rescue capability model of the target area was constructed by using the Geographic Information System-Multi-Criteria Decision Analysis (GIS-MCDM) method. Specifically, the AHP method of GIS-MCDM was used to determine the weights of each rescue capability factor to construct the rescue capability model, and the rescue capability of the target area was evaluated based on the rescue capability model to obtain an evaluation value, which reflects the rescue capability level in the target area.
[0057]
[0058] Finally, a wildfire risk model is constructed based on the wildfire susceptibility assessment results, environmental vulnerability assessment results and rescue capabilities of the target area, and the wildfire risk in the target area is calculated and evaluated based on the wildfire risk model.
[0059] In this embodiment, the wildfire risk model is: wildfire risk = target area wildfire susceptibility × target area environmental vulnerability × rescue capacity in the target area.
[0060] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.
[0061] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.
Claims
1. A wildfire risk assessment method, characterized in that: The following steps are involved: An optimal wildfire susceptibility model is constructed based on the environmental data of the target area and the temporal, spatial and geographical laws of historical wildfire distribution, and the wildfire susceptibility of the target area is evaluated using the optimal wildfire susceptibility model; Calculate the environmental vulnerability of the target area; Calculate rescue capabilities within the target area; A wildfire risk model is constructed based on the wildfire susceptibility assessment results, environmental vulnerability assessment results and rescue capabilities of the target area to evaluate the wildfire risk in the target area.
2. The wildfire risk assessment method according to claim 1, characterized in that: The steps to construct an optimal wildfire susceptibility model are: Collect climate characteristics, geographic characteristics, vegetation characteristics, and historical wildfire information in the target area, and pre-process all data; The kernel density calculation is performed using historical wildfire data, and the temporal, spatial and geographical laws of historical wildfire distribution are analyzed based on the calculation results. All data are divided into data sets according to time units, wildfire data are matched with other feature data one by one, and all data except fire points are normalized; Wildfire susceptibility models were constructed based on a variety of machine learning models; The wildfire susceptibility models constructed by various machine learning models were optimized based on the scikit-optimize framework to achieve the optimal solution for each model. The optimal solutions of all models were compared, and the model with the highest accuracy was selected as the optimal wildfire susceptibility model.
3. The wildfire risk assessment method according to claim 2, characterized in that: The historical wildfire information of the target area is obtained through the wildfire data observed by the fire department and / or various satellites in the target area, and the bad data in the historical wildfire data are screened and eliminated to obtain the historical wildfire information of the target area.
4. The wildfire risk assessment method according to claim 2, characterized in that: Obtain climate characteristic data of the target area through the meteorological station in the target area, and the acquired information includes specific values and data time; Obtain the geographic feature data of the target area by querying the National Geographic Information Public Service Platform; The historical images of the target area were downloaded through the GEE platform, and the vegetation characteristic data of the target area were calculated in the ENVI software.
5. The wildfire risk assessment method according to claim 2, characterized in that: The kernel density calculation tool in ArcMap is used to calculate the historical wildfire data, and the analysis results are used to obtain the spatiotemporal and geographical laws of the historical wildfire distribution in the target area.
6. The wildfire risk assessment method according to claim 1, characterized in that: An environmental vulnerability model of the target area is constructed based on the ecological remote sensing index of the target area. The RSEI value is calculated based on the remote sensing image of the target area and the environmental vulnerability model of the target area to evaluate the environmental vulnerability of the target area.
7. The wildfire risk assessment method according to claim 1, characterized in that: The steps to calculate the rescue capability within the target area are: Five factors affecting rescue capability, namely, distance to roads, distance to rivers, distance to fire stations, distance to hospitals, and distance to densely populated areas, were selected, and a rescue capability model of the target area was constructed using the GIS-MCDM method. The rescue capability in the target area is calculated based on the rescue capability model of the target area.
8. The wildfire risk assessment method according to claim 1, characterized in that: The wildfire risk model is: wildfire risk = target area wildfire susceptibility × target area environmental vulnerability × target area rescue capacity.