Global fire sensitivity assessment method based on climate factors

Abstract

The invention discloses a global fire sensitivity assessment method based on climate factors, relates to the field of climate science and disaster management, and solves the technical problem that an existing fire sensitivity assessment method does not consider a nonlinear relationship between climate variables and a synergistic effect to affect an assessment effect. The method comprises the following steps: S1, constructing a dynamic time-delay effect model; s2, constructing a space-time interaction effect model; s3, constructing a correlation coefficient of climatic change intensity measurement and fire response intensity measurement; s4, considering the complex influence of climate change, combining the dynamic time-lag effect, the space-time interaction effect and the correlation coefficient, and adopting a mathematical modeling method to construct a space-time sensitivity index STSI; s5, STSI is used for clustering the sensitivity of the fire disaster to the climate change into five grades; according to the method, a plurality of climate variables including temperature, precipitation, wind speed and humidity can be integrated, and the time-space interaction among the factors is considered, so that the fire sensitivity can be identified and classified more accurately.

Description

Technical Field

[0001] The present invention relates to the fields of climate science and disaster management, and particularly to a method for assessing global fire sensitivity based on climate factors. Background Art

[0002] Global climate change has become one of the major challenges faced by modern society, and it has a profound impact on the frequency and intensity of natural disasters, especially fires. With the increase in the earth's surface temperature, extreme weather events have become more frequent, including droughts and heatwaves, which have significantly increased the risk of forest fires. Fires not only cause huge economic losses but also have long-term negative impacts on ecosystems and human life. Therefore, accurately understanding and predicting the fire sensitivity of different regions is crucial for formulating effective fire prevention strategies and disaster reduction plans. In traditional fire sensitivity analysis methods, basic climate data and relatively simple statistical models are often relied on to evaluate fire risks. When evaluating fire sensitivity, traditional methods usually analyze climate variables such as temperature, humidity, and wind speed as independent factors, ignoring their non-linear relationships and synergistic effects. For example, the combination of high temperature and low humidity may significantly increase the fire risk, and analyzing these variables separately cannot accurately reflect their combined effects. In addition, traditional methods are mostly based on data from static or fixed time periods and do not fully consider the dynamic characteristics of climate variables changing over time, such as seasonal changes and the increasing frequency of extreme climate events. Such static models cannot reflect the dynamic trends of climate change, resulting in limitations in their application effects and accuracy under the background of global warming.

[0003] For example, in the Chinese patent "A Forest Fire Prediction Method Based on Deep Learning and Multi-Source Remote Sensing Data" (Patent Application No.: CN202210458913.2, Publication No.: CN114998719A). This patent uses data samples of influencing factors to train a ResNet residual neural network, and then uses the trained model for forest fire prediction.

[0004] However, the technical solution of this patent only uses specific numerical values within a certain time period as samples for model training and does not fully consider the dynamic characteristics of climate variables changing over time. For forest fire situations under climate conditions similar to the data samples of influencing factors, the expected prediction effect can be achieved. But when facing completely different climate conditions for forest fire prediction, the prediction effect will be less than satisfactory. Summary of the Invention

[0005] To address the problems existing in the above-mentioned prior art, the present invention provides a global fire sensitivity assessment method based on climate factors, which can comprehensively consider multiple climate variables, including temperature, precipitation, wind speed, and humidity, and take into account the spatio-temporal interaction between these factors. In addition, considering that the effects of climate change may have a time lag, a dynamic time lag model can be used to capture this lag effect. Through this comprehensive and systematic analysis, fire sensitivity can be more accurately identified and classified, providing a more scientific basis for policy-making and resource allocation.

[0006] A global fire sensitivity assessment method based on climate factors, comprising: S1. Construct a dynamic time lag effect model for capturing the lag of the impact of climate change parameters on fires; S2. Construct a spatio-temporal interaction effect model for examining the variability of the impact of different ecological regions and time on fires; S3. Construct a correlation coefficient between the measurement of climate change intensity and the measurement of fire response intensity for quantifying the association between climate change and fires; S4. Considering the complex impact of climate change, combining the dynamic time lag effect, spatio-temporal interaction effect, and correlation coefficient, and using a mathematical modeling method to construct a spatio-temporal sensitivity index STSI; S5. Use STSI to cluster the sensitivity of fires to climate change into 5 levels.

[0007] Furthermore, the impact of climate change on fires is not only immediate but also has a time lag effect. To capture this dynamic time lag effect, the dynamic time lag effect model described in S1 is expressed as follows:

[0008] is the fire index at time t and ecological region s, is the lag time is the climate index at time and are weights, is the error term. is the climate index at time t and ecological region s, is the lag time is the fire index at time Furthermore, considering the complex association between climate change and fire activities: the spatio-temporal interaction effect measure considers the non-linear interaction between climate factors and fires. The spatio-temporal interaction effect model described in S2 is expressed as follows:

[0009] represents the spatio-temporal interaction effect model, is the climate index at time t and ecological region s, is the fire index at time t and ecological region s, is the spatial weight of different ecological regions S. The spatial weights of different ecological regions are determined by Geographically Weighted Regression (GWR), which reflects the influence degree of climate in different regions on fire, and is the adjustment coefficient, which measures the influence degree of temperature change on the fire burning area, is determined by linear regression, is determined by non - linear fitting (logarithmic regression), is the non - linear response index. By analyzing the data distribution of the climate index and the burning area and their interaction, the best non - linear response index is determined using a non - linear regression model (polynomial regression).

[0010] Furthermore, the S3 includes: S31. Construct a measure of the intensity of climate change; S32. Construct a measure of the intensity of fire response; S33. Construct the correlation coefficient between the measure of the intensity of climate change and the measure of the intensity of fire response.

[0011] Furthermore, the measure of the intensity of climate change reflects the changes of climate factors at different time points and spatial points; the intensity of climate change is measured by calculating the change amount of climate factors and is standardized to eliminate the influence of seasonal fluctuations; the formula for calculating the measure of the intensity of climate change in S31 is as follows:

[0012] represents the measure of the intensity of climate change, and respectively represent the i - th and value, is the climate index at time t and ecological region s, is the lag time climate index at the moment, and represent the mean and standard deviation, is the adjustment coefficient, which is determined by linear regression analysis.

[0013] Furthermore, the measure of the intensity of fire response is used to describe the change of the burning area (BA) in the time dimension. By calculating the annual change amount of the burning area and further standardizing it, the intensity of fire response is obtained. The formula for calculating the measure of the intensity of fire response in S32 is as follows:

[0014] Represents the fire response intensity metric, and respectively represent the j-th and values, is the fire index at time t and ecological region s, is the lag time The fire index at the moment, and represent the mean and standard deviation, is the adjustment coefficient, determined by linear regression analysis.

[0015] Furthermore, in order to reveal the connection between climate change and fire intensity, the correlation between the climate change intensity metric and the fire response intensity metric is calculated. The formula for calculating the correlation coefficient of the climate change intensity metric and the fire response intensity metric in S33 is as follows:

[0016] represents the correlation coefficient, and respectively represent the i-th and values, represents the fire response intensity metric, represents the climate change intensity metric, and respectively represent and means.

[0017] Furthermore, in order to comprehensively reflect the complex relationship between climate change and fire, a comprehensive model considering the correlation coefficient of climate change intensity and fire response intensity, spatiotemporal interaction effects, and time lag effects is designed. This model considers the immediate and delayed impacts between climate factors and fire combustion area, provides a multi-dimensional perspective to analyze their dynamic relationship, and constructs the Spatiotemporal Sensitivity Index (STSI) formula as follows:

[0018] Spatiotemporal sensitivity index, is The climate change intensity metric at the moment, is The fire response intensity metric at the moment, represents the attenuation of the time lag effect, is the attenuation rate, determined by the least squares method, and is the adjustment coefficient, represents the measurement of climate change intensity, represents the measurement of fire response intensity, which is determined by linear regression analysis.

[0019] Furthermore, in S5, STSI is used to cluster the sensitivity of fire to climate change into five levels, including L1 low sensitivity, L2 relatively low sensitivity, L3 medium sensitivity, L4 relatively high sensitivity, and L5 high sensitivity.

[0020] The beneficial effects of the present invention include: First, by integrating the dynamic time-lag effect, spatio-temporal interaction effect, and correlation coefficient between climate change and fire, the complex effects between climate variables and their sensitivity to fire can be captured and analyzed more accurately. Second, by constructing the spatio-temporal sensitivity index (STSI) and classifying the fire sensitivity of each region in the world into five levels, a scientific decision-making support tool is provided for the government and relevant agencies to help them conduct more effective resource allocation and risk management, so as to prevent and respond to fire disasters more effectively. In addition, the application of the present invention helps to improve public safety, reduce the life and property losses caused by fires, and at the same time promote the implementation of environmental protection and sustainable development strategies. In short, the present invention is not only innovative in technology, but also has extensive social, economic, and environmental benefits in practical applications. Description of the Drawings

[0021] Figure 1 is a flowchart of a global fire sensitivity assessment method based on climate factors according to an embodiment of the present application.

[0022] Figure 2 is a statistical chart of the spatio-temporal sensitivity index according to an embodiment of the present application.

[0023] Figure 3 is a statistical chart of the sensitivity classification of fire to climate change according to an embodiment of the present application. Among them, (a) shows the sensitivity classification statistical results of the spatio-temporal clustering model constructed based on the present invention, and (b) shows the sensitivity classification statistical results obtained by using the traditional K-means clustering model. Specific Embodiments

[0024] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part rather than all of the embodiments of this application. Therefore, the detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative efforts fall within the scope of protection of this application.

[0025] Embodiment 1 The global ecological regions were selected as the spatial analysis units. The division of ecological regions was based on the research of Dinerstein et al. (2017), and these ecological regions were used to describe the biogeographical trends in extreme wildfire events. Based on MODIS data, the average burned area (BA) was calculated with ecological regions as the spatial units. The present invention also used the ERA5 dataset as the main data source. ERA5 provides high-resolution meteorological data, including multiple meteorological elements such as air temperature, humidity, wind speed, and precipitation. Through these data, the present invention further calculated meteorological indices such as the Fire Weather Index (FWI), Standardized Precipitation Index (SPI), Standardized Land Surface Temperature Index (SLTI), and wind speed (W). The preprocessing of the aforementioned data included: resampling, outlier removal, image registration, etc.

[0026] A global fire sensitivity assessment method based on climate factors, as Figure 1 shown, includes: S1. Construct a dynamic time-lag effect model to capture the delay of climate change parameters on fire; S2. Construct a spatio-temporal interaction effect model to examine the variability of the impact of different ecological regions and time on fire; S3. Construct the correlation coefficient between the climate change intensity metric and the fire response intensity metric to quantify the association between climate change and fire; S4. Considering the complex impact of climate change, combining the dynamic time-lag effect, spatio-temporal interaction effect, and correlation coefficient, use a mathematical modeling method to construct a spatio-temporal sensitivity index STSI; S5. Use STSI to cluster the sensitivity of fire to climate change into 5 levels.

[0027] In another embodiment, the impact of climate change on fire is not only immediate but also has a time-lag effect. Specifically Figure 2Shows the statistical charts of the spatio-temporal sensitivity indices of four climate factors: the Fire Weather Index (FWI), the Standardized Precipitation Index (SPI), the Standardized Land Surface Temperature Index (SLTI), and wind speed (W); To capture this dynamic time-lag effect, the dynamic time-lag effect model described in S1 is expressed as follows:

[0028] Is the fire index at time t and ecological region s, calculated by obtaining the average value of BA according to different ecological regions. Is the lag time The climate index at time And Are weights Is the error term Is the climate index at time t and ecological region s, calculated by obtaining the average value of the meteorological index according to different ecological regions. Specifically, C represents different climate factors, and in turn, C = FWI, C = SPI, C = SLTI, C = W. Is the lag time The fire index at time In another embodiment, considering the complex association between climate change and fire activities: The spatio-temporal interaction effect measure considers the non-linear interaction between climate factors and fire. The spatio-temporal interaction effect model described in S2 is expressed as follows:

[0029] Represents the spatio-temporal interaction effect model Is the climate index at time t and ecological region s Is the fire index at time t and ecological region s Is the spatial weight of different ecological regions S, determined by Geographically Weighted Regression (GWR) to reflect the influence degree of climate on fire in different regions. And Are adjustment coefficients, measuring the influence degree of temperature change on the fire burning area. Determined by linear regression Determined by non-linear fitting (logarithmic regression) Is the non-linear response index, determined by analyzing the data distribution and interaction of the climate index and the burning area, and using a non-linear regression model (polynomial regression) to determine the optimal non-linear response index.

[0030] In another embodiment, S3 includes: S31. Construct a measure of the intensity of climate change; S32. Construct a measure of the intensity of fire response; S33. Construct the correlation coefficient between the climate change intensity metric and the fire response intensity metric.

[0031] In another embodiment, the climate change intensity metric reflects the changes of climate factors at different time points and spatial points; the climate change intensity is measured by calculating the change amount of climate factors and is standardized to eliminate the influence of seasonal fluctuations; the formula for calculating the climate change intensity metric in S31 is as follows:

[0032] represents the climate change intensity metric, and respectively represent the i-th and value, is the climate index at time t and ecological region s, is the lag time of the climate index at time and represent the mean and standard deviation, is the adjustment coefficient determined by linear regression analysis. Among them, C represents different climate factors (one of the climate factors of FWI, SPI, SLTI, W). C = FWI, C = SPI, C = SLTI, C = W are calculated in a loop. Specifically, when C = FWI, C i represents the value of the i-th FWI; when C = W, C i represents the value of the i-th W.

[0033] In another embodiment, the fire response intensity metric is used to describe the change of the burned area (BA) in the time dimension. By calculating the annual change amount of the burned area and further standardizing it, the fire response intensity is obtained. The formula for calculating the fire response intensity metric in S32 is as follows:

[0034] represents the fire response intensity metric, and respectively represent the j-th and value, is the fire index at time t and ecological region s, is the lag time of the fire index at time and represent the mean and standard deviation, is the adjustment coefficient determined by linear regression analysis.

[0035] In another embodiment, to reveal the connection between climate change and fire intensity, the correlation between the climate change intensity metric and the fire response intensity metric is calculated. The formula for calculating the correlation coefficient of the climate change intensity metric and the fire response intensity metric in S33 is as follows:

[0036] represents the correlation coefficient, and respectively represent the i-th and values, represents the fire response intensity metric, represents the climate change intensity metric, and respectively represent and means.

[0037] In another embodiment, to comprehensively reflect the complex relationship between climate change and fire, a comprehensive model considering the correlation coefficient of climate change intensity and fire response intensity, spatiotemporal interaction effects, and time lag effects is designed. This model considers the immediate and delayed impacts between climate factors and fire combustion area, provides a multi-dimensional perspective to analyze their dynamic relationship, and constructs the Spatiotemporal Sensitivity Index (STSI) formula as follows:

[0038] Spatiotemporal Sensitivity Index, is the climate change intensity metric at time, is the fire response intensity metric at time, represents the decay of the time lag effect, is the decay rate, determined by the least squares method, and are adjustment coefficients, represents the climate change intensity metric, represents the fire response intensity metric, determined by linear regression analysis.

[0039] In another embodiment, in S5, the STSI is used to cluster the sensitivity of fire to climate change into 5 levels, including L1 low sensitivity, L2 lower sensitivity, L3 medium sensitivity, L4 higher sensitivity, and L5 high sensitivity. Data-driven adaptive partitioning does not require setting a fixed threshold. The parameters to be set are as follows: n_slusters = 5. The number of clusters corresponds to 5 risk levels; Max_itec = 300. The maximum number of iterations to run; tol = 1e-4. Convergence threshold (distance of centroid movement between two iterations) N_init = 10. The number of times the algorithm is randomly initialized (the optimal result is taken).

[0040] Finally, the specific effect of hierarchical statistics is as Figure 3 shown.

[0041] Level 1 (very low sensitivity): The fires in these areas are less sensitive to climate change. Even if the climate changes (e.g., temperature increases), the change in the fire occurrence frequency in this area is small, and the fire risk is relatively low.

[0042] Level 2 (low sensitivity): The fires in these areas are less sensitive to climate change, but compared with Level 1 areas, climate change may affect the fire occurrence frequency to a certain extent, but the change range is small.

[0043] Level 3 (medium sensitivity): The fires in these areas are moderately sensitive to climate change. Climate change (such as temperature increase, precipitation decrease, etc.) may significantly change the fire risk in these areas, and the fire occurrence frequency may increase or change.

[0044] Level 4 (high sensitivity): The fires in these areas are highly sensitive to climate change. As climate change intensifies, the fire occurrence frequency increases significantly, and the fire risk is high.

[0045] Level 5 (extremely high sensitivity): These areas are extremely sensitive to climate change. Even a small climate change (such as temperature increase, precipitation decrease, etc.) may cause a significant increase in the fire occurrence frequency and intensity. Such areas have extremely high fire risks and require special attention.

[0046] In the present invention, step 5 specifically involves using the spatio-temporal clustering model developed by the present invention and the conventional K-means clustering model to conduct a five-level classification (L1 low sensitivity, L2 relatively low sensitivity, L3 medium sensitivity, L4 relatively high sensitivity, L5 high sensitivity) of the fire sensitivity in different ecological regions. To evaluate the performance of these two clustering methods, the present invention conducts an in-depth comparative analysis using three accuracy evaluation metrics: the silhouette coefficient, the Calinski-Harabasz index, and the Davies-Bouldin index. The silhouette coefficient is used to evaluate the compactness and separation of the clustering, and a higher value indicates better cohesion and distinctiveness of the clustering results. The Calinski-Harabasz index evaluates the effectiveness of the clustering by measuring the ratio of the within-cluster cohesion to the between-cluster separation, and a higher index value reflects better clustering quality. The Davies-Bouldin index evaluates the separability of the clustering by measuring the average similarity of the clusters, where a lower value indicates better clustering separation.

[0047] The evaluation results show that the method of the present invention is superior to the traditional K-means clustering in terms of the silhouette coefficient, the Calinski-Harabasz index, and the Davies-Bouldin index. Specifically, the average value of the silhouette coefficient of the method of the present invention is 0.65, showing a high degree of cohesion and good separation; while the average value of the silhouette coefficient of the K-means clustering is 0.45, indicating a lower clustering separation. In terms of the Calinski-Harabasz index, the present invention reaches a high score of 250, significantly superior to the 150 points of the K-means clustering, which further proves the significant advantage of the clustering method of the present invention in distinguishing between clusters and within clusters. In addition, the Davies-Bouldin index provides a score of 1.2 for the present invention, significantly lower than the 2.1 of the K-means clustering, thus confirming the superiority of the present invention in clustering quality.

[0048] The experimental results show that the method of the present invention can very effectively classify and analyze fire-sensitive areas, specifically: The Calinski-Harabasz index and the Davies-Bouldin index show that our method can clearly distinguish fire-sensitive areas, making the risk differences between different areas more obvious.

[0049] The silhouette coefficient indicates good classification results, with similar fire sensitivities within the area and significant differences from other areas.

[0050] The combination of these metrics proves that our method is not only applicable over a large range but also can efficiently and accurately identify fire-sensitive areas, thus providing accurate support for relevant decision-making. Especially in some complex areas, the method can still maintain a high classification accuracy.

[0051] These experimental results verify the ability of the method of the present invention to effectively classify and accurately analyze fire sensitivity data globally, demonstrating its significant improvement in scientificity and accuracy compared with traditional methods.

[0052] The above-described embodiments merely represent the specific implementation manners of the present application. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the protection scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the technical solution of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application.

Claims

1. A global fire sensitivity assessment method based on climate factors, characterized in that, Including the following steps: S1. Construct a dynamic time-lag effect model to capture the delay of climate change parameters on fire; S2. Construct a spatio-temporal interaction effect model to examine the variability of the impact of different ecological regions and time on fire; S3. Construct the correlation coefficient between the climate change intensity measure and the fire response intensity measure to quantify the association between climate change and fire; S4. Considering the complex impact of climate change, combining the dynamic time-lag effect, spatio-temporal interaction effect and correlation coefficient, use the mathematical modeling method to construct the spatio-temporal sensitivity index STSI; S5. Use STSI to cluster the sensitivity of fire to climate change into 5 levels.

2. The global fire sensitivity assessment method based on climate factors according to claim 1, wherein, The dynamic time-lag effect model described in S1 is expressed as follows: ； is the fire index at time t and ecological region s, is the lag time is the climate index at time and are weights, is the error term, is the climate index at time t and ecological region s, is the lag time is the fire index at time 3. The global fire sensitivity assessment method based on climate factors according to claim 1, wherein The spatio-temporal interaction effect model described in S2 is expressed as follows: ； represents the spatio-temporal interaction effect model, is the climate index at time t and ecological region s, is the fire index at time t and ecological region s, is the spatial weight of different ecological regions S, and are adjustment coefficients, is the non-linear response index, which is determined using a non-linear regression model by analyzing the data distribution and their interaction of the climate index and the burned area.

4. The global fire susceptibility assessment method based on climate factors according to claim 1, characterized in that The said S3 includes: S31. Construct a measure of climate change intensity ; S32. Construct the measurement of fire response intensity ; S33. Construct the correlation coefficient between the climate change intensity metric and the fire response intensity metric .

5. The global fire sensitivity assessment method based on climate factors according to claim 4, wherein, The formula for calculating the climate change intensity measure in S31 is as follows: ； Indicates the intensity measure of climate change, and respectively represent the and values, is the climate index at time t and ecological region s, is the lag time of the climate index at that moment, and represent the mean and standard deviation, is the adjustment coefficient determined by linear regression analysis.

6. The global fire sensitivity assessment method based on climate factors according to claim 4, wherein The formula for calculating the fire response intensity measure in S32 is as follows: ； represents the measure of fire response intensity, and respectively represent the j-th and value, is the fire index at time t and ecological region s, is the lag time the fire index at the moment, and represent the mean and standard deviation, is the adjustment coefficient, determined by linear regression analysis.

7. A method for global fire sensitivity assessment based on climate factors according to claim 4, characterized in that The formula for calculating the correlation coefficient between the climate change intensity measure and the fire response intensity measure in S33 is as follows: ； represents the correlation coefficient, and represent the values of the i-th and respectively, represents the measure of fire response intensity, represents the measure of climate change intensity, and represent respectively and the means.

8. A method for evaluating global fire sensitivity based on climate factors according to claim 4, characterized in that The formula for constructing the spatio-temporal sensitivity index STSI is as follows: ； Spatio-temporal sensitivity index, is a measure of the intensity of climate change at time is a measure of the intensity of fire response at time indicating the attenuation of the time-delay effect, is the attenuation rate, and is the adjustment coefficient, indicating the measure of the intensity of climate change, indicating the measure of the intensity of fire response, determined by linear regression analysis.

9. A method for evaluating global fire sensitivity based on climate factors according to claim 1, characterized in that, In the said S5, using STSI to cluster the sensitivity of fire to climate change into 5 levels, including L1 low sensitivity, L2 relatively low sensitivity, L3 medium sensitivity, L4 relatively high sensitivity, L5 high sensitivity.

Images