A method for quantifying the increment of yield loss of food crops caused by alien invasive species based on scenario difference

Abstract

This invention discloses a spatial quantification method for incremental yield loss in food crops caused by invasive alien species based on scenario differentiation, belonging to the field of biosafety and agricultural risk assessment. It obtains a climate suitability raster based on GIS raster operations and a species distribution model, constructing a candidate set of driving factors. Regularized regression is used to screen key factors and assign weights, coupling suitability with anthropogenic factors to generate a single-species risk raster, which is then integrated into a multi-species comprehensive risk according to rules. The baseline and future scenario comprehensive risks are calculated and differencing, with negative differences reclassified as 0 to extract newly added risk areas. This is then multiplied by the crop yield raster to obtain the yield loss increment raster corresponding to the newly added risk areas, outputting the global / regional spatial distribution. This method can be extended to multiple species, multiple crops, and multiple scenarios, providing quantitative decision support for food security and invasion control.

Description

Technical Field

[0001] This invention belongs to the field of biosafety and agricultural risk assessment technology, specifically relating to a spatial quantification method for incremental yield loss of food crops caused by invasive alien species based on scenario differencing, which uses a Geographic Information System (GIS) on a computer platform to process spatially registered raster data. More specifically, this invention relates to generating a climate suitability raster for invasive species based on a species distribution model, combining it with a raster of human-driven factors, and determining weight parameters through regularized regression screening to construct a comprehensive invasion risk raster; further, by risk difference between the baseline scenario and the future scenario and reclassifying negative difference pixels as zero, only newly added risk areas are extracted; the newly added risk raster is then multiplied by the food crop yield raster to output the spatial distribution result of the incremental yield loss corresponding to the newly added invasion risk areas. Background Technology

[0002] Food security is one of the fundamental issues of national security and global sustainable development. Globalization, denser transportation networks, and climate change are jointly driving the cross-regional spread and diffusion of invasive alien species, exacerbating the risk of biological disasters to food crops. At the macro level, invasive species can reduce crop yields and increase control costs through direct predation, disease transmission, and resource competition. At the spatial level, the spatial mismatch between invasion risk and food yield distribution, as well as the dynamic expansion of risk with changing scenarios, make it difficult to pinpoint risk hotspots and quantify potential yield losses on a large spatial scale using only traditional statistics or local assessments.

[0003] In existing technologies, spatial assessment of the risk of invasive alien species often employs Species Distribution Models (SDMs) to predict potential suitable habitats for invasive species, outputting a suitability raster or binarized potential distribution map, which is then overlaid to obtain species richness or a risk index. While this type of method has advantages such as relatively convenient data acquisition and the ability to perform scenario projection, it still has the following shortcomings when applied to the spatial quantification of crop yield loss: 1. Risk characterization is biased towards climate suitability alone, failing to reflect the moderating effects of human activities on actual occurrences. The occurrence and spread of invasive species depend not only on climate and environmental suitability but are also significantly influenced by factors such as the intensity of human activities, trade and transportation accessibility, land use, and population and economic activities. Using only climate suitability as a risk characterization can easily lead to overestimation of risk in areas with high suitability but low input opportunities or weak human interference; conversely, it can lead to underestimation of risk in areas with high input pressure, strong interference, or significant propagation channels. On the other hand, climate variables and human variables often exhibit collinearity. Directly incorporating a large number of human factors into the model can lead to parameter instability, variable importance drift, and overfitting, thereby affecting the repeatability and generalizability of the risk grid.

[0004] 2. Loss quantification is often expressed as absolute risk or absolute loss, lacking a mechanism for incremental identification of newly added risk areas. Existing spatial loss assessments typically couple the suitability or risk index under a specific scenario directly to a crop distribution / yield grid to obtain the total loss or risk intensity distribution. Such results struggle to distinguish which areas represent risks already present under historical or baseline scenarios and which represent risks newly expanded under future scenarios. Without this distinction, historically risky areas are easily double-counted in future assessments, thus exaggerating the spatial extent and magnitude of new threats and losses, hindering accurate identification of risk increments and prioritized resource allocation.

[0005] 3. The computational chain from risk maps to yield loss maps usable for decision-making is not clearly defined, making it difficult to establish a reproducible spatial quantification process. Under conditions of large spatial scales and multiple data sources, risk rasters, crop yield rasters, and human-factor rasters often exhibit inconsistencies in spatial resolution, projection, and masking range. Without clear rules for registration, pruning, and raster operations, it is difficult to guarantee the repeatability and cross-scenario comparability of the results, and it also hinders the further application of the results to national or regional-scale food security risk assessment and policy analysis.

[0006] In addition, existing methods for risk assessment of invasive alien species and quantification of food loss generally suffer from the following problems: 1. Risk characterization relies heavily on a single climate suitability index, which is insufficient to reflect the modulating effect of human activities, trade and transportation accessibility, and other anthropogenic factors on the occurrence and spread of invasion; 2. Candidates are often strongly correlated with climate variables as driving factors. Directly including them in the model can easily lead to instability and overfitting of coefficients due to collinearity, which affects the repeatability and generalizability of the risk grid. 3. Existing loss assessments are mostly expressed as "absolute risk / absolute loss" under a certain scenario, lacking a mechanism for extracting new risk areas based on the comparison between the baseline scenario and future scenarios. This easily leads to the repeated inclusion of historical risk areas in future assessments, resulting in an overestimation of new threat areas and losses. 4. The spatial computation chain is unclear, making it difficult to form a reproducible large-scale spatial quantization process.

[0007] For the reasons mentioned above, current technologies lack a unified methodology capable of: constructing a comprehensive invasion risk raster that couples climate suitability with weighted anthropogenic driving factors in a computer / GIS environment; extracting newly added risk areas based on risk difference between baseline and future scenarios and reclassifying negative difference pixels to zero; and coupling the newly added risk raster with a crop yield raster to obtain a spatially quantified output of yield loss increments. Therefore, it is necessary to propose a reproducible spatial quantification method applicable to global and regional scales, multiple species, and multiple scenarios, in order to more accurately identify newly added invasion threat areas and quantify their incremental impact on food crop yields, providing more targeted decision-making support for biological invasion prevention and control and food security management. Summary of the Invention

[0008] This invention proposes a spatial quantification method for performing calculations on spatially registered raster data using a Geographic Information System (GIS) on a computer platform: at the single-species level, a comprehensive invasion risk raster coupled with "climate suitability - human-driven" is constructed; at the multi-species level, a crop-level comprehensive risk raster is formed; and new risk areas are extracted by setting the baseline / future risk difference and negative difference pixels to zero. Finally, the method is coupled with a crop yield raster to output the spatial distribution of yield loss increments in the new risk areas.

[0009] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0010] A spatial quantification method for incremental crop yield loss caused by invasive alien species based on scenario differentiation includes the following steps (each step can be implemented through raster operations in GIS software or a geocomputing framework):

[0011] Step S1: Acquisition of multi-source data, unified spatial benchmark and raster registration S1-1 Obtain a large-scale spatial distribution raster of the target food crop's yield. And a list of invasive alien species associated with the crop; for each invasive species To obtain its distribution location data and origin range data.

[0012] S1-2 Obtain the environmental predictor raster for modeling, including the climate factor raster and the set of candidate driving factor raster. .

[0013] S1-3 performs unified spatial reference processing on the above raster data, including: unified projection coordinate system, unified spatial resolution, unified cell alignment method, unified mask range (effective cell range), and performs spatial clipping and coordinate consistency checks on the point data in the same way as the raster to ensure the consistency of cell-level registration in subsequent raster operations.

[0014] Step S2: Generating a raster of climate suitability for invasive species S2-1 For each invasive species A species distribution model is established based on its distribution point data and climate factor raster, and the distribution of the species in each pixel is output. Climate suitability grid Its values ​​are normalized and mapped to .

[0015] The species distribution model described in S2-2 can be MaxEnt, generalized linear model, random forest, ensemble model, etc.; preferably, a model that can output continuous fitness is used so as to continuously couple with human driving factors.

[0016] Step S3: Screening of artificial driving factors and determination of weight parameters S3-1 Constructing a raster set of candidates as driving factors Furthermore, each factor raster is standardized within the effective pixels (e.g., zero-mean unit variance standardization) to eliminate the influence of dimensional differences on the regression penalty term.

[0017] S3-2 uses regularized regression to screen candidates as driving factors, determining the stable set of factors and their weight parameters. .

[0018] Step S4: Calculation of Integrated Invasion Risk for Single Species using a Raster S4-1 For any invasive species After completing steps S2 and S3, the pixel value of this species is obtained. Climate suitability value And the key artificial driving factor grid obtained by regularized regression screening. and its weighting coefficients .in, This indicates the sequence number of the key human driving factor after screening (determined according to the screening results). , (Number of factors after screening).

[0019] set up For species In pixels The single-species integrated invasion risk value; This is the weighting coefficient for climate suitability. To map the coupling results within the parentheses to The normalization function (activation function).

[0020] S4-2 Normalized Weighted Coupling In each pixel Above, climate suitability is weighted and coupled with key anthropogenic driving factors, and then... Mapping yields the integrated invasion risk value for a single species. The calculation formula is as follows: ;

[0021] in, Characterize the promoting or inhibiting effect and intensity of key anthropogenic driving factors on intrusion risk; Used to adjust the contribution of the climate suitability term in the coupling.

[0022] Optionally, to characterize the interaction modulation effect between climate suitability and anthropogenic driving factors, a factor-by-factor interaction term weighting coefficient is introduced based on Implementation Method 1. And calculate the integrated invasion risk value of a single species. Its formula is:

[0023] in, Used to quantify climate suitability for the first Modulation strength of the risk contribution of each key human factor: when This indicates that the risk contribution of this anthropogenic driving factor is amplified in areas with high biodiversity; when The presence of a time interval indicates the existence of an inhibition or saturation effect.

[0024] S4-4 Implementation method The To map the calculation results within the parentheses to The function can be the Logistic function, the linear truncation normalization function, or the quantile normalization function, etc. The Logistic function is preferred. ;

[0025] S4-5 Output Perform step S4-2 (or step S4-3) on all valid pixels within the study area to obtain the invasive species. Single-species integrated invasion risk grid Its pixel value The range of values ​​is This serves as an input for the multi-species integrated risk integration in the subsequent step S5.

[0026] Step S5: Multi-species integrated risk grid integration S5-1 Risk raster for all invasive species associated with the target food crop Raster integration is performed according to preset rules to obtain the target crop in pixels. Comprehensive risk grid .

[0027] The multi-species integration rule described in S5-2 can be adopted in one of the following ways:

[0028] (a) Summation, truncation, and normalization: ;

[0029] (b) Merging probability rule: ; S5-3 Perform steps S2-S5 under both the baseline and future scenarios to obtain the comprehensive risk grid for the baseline scenario. Integrated Risk Grid with Future Scenarios .

[0030] Step S6: Extracting Newly Added Risk Areas S6-1 Calculate the risk differential grid: ; S6-2 Reclassifies the difference results, retaining only newly added risk areas: When When, the corresponding pixel is reclassified as 0; when At that time, its original value remains unchanged, thus obtaining a risk increment raster that only represents newly added risk areas. .

[0031] S6-3 Optionally, to reduce the impact of the potential suitable habitat area being too large relative to the actual occurrence area, a partitioning benchmark subtraction mask can be constructed during the extraction of newly added risk areas. Determining the origin mask based on invasive species origin range data With Intrusion Zone Mask Within the intrusion zone, the actual locations of the intrusion are rasterized to generate an occurrence mask. Within the original production area, the climate suitability grid is reclassified under the current scenario to generate a suitability mask. ; and order ; pass ; Or pixel correspondence Assigning a value of 0 yields a risk increment raster that only represents newly added risk areas.

[0032] Step S7: Calculation and output of incremental production loss in newly added risk areas S7-1 Calculates the incremental production loss raster corresponding to newly added risk areas at the pixel scale: ; S7-2 Spatial aggregation and mapping output are performed to generate spatial distribution results of the incremental loss of grain crop yield caused by the new invasion risk at a large scale.

[0033] Preferably, the regularized regression described in S3 is an elastic network regularized regression, in which the regularization strength parameter is selected through cross-validation, and the non-zero regression coefficient is used as the criterion for factor selection.

[0034] Furthermore, to improve robustness under spatial autocorrelation, the cross-validation adopts spatial block cross-validation and can perform repeated block fitting to calculate the selection frequency, and select factors with high selection frequency and consistent sign direction as the final selection factors.

[0035] Preferably, This can be achieved using the Logistic function, linear truncation normalization, or quantile normalization, among other methods. This is the fitness weighting coefficient. The selected individuals are factor weighting coefficients.

[0036] Furthermore, a partitioned baseline subtraction mask is constructed in S6. Among them, the pixels of the intrusion area Occurrence mask obtained by rasterizing the actual occurrence point of the intrusion site Confirmed, the original region of origin pixels The suitability mask obtained from the current climate suitability grid reclassification Determine; and based on

[0037] ; Or The corresponding pixel Assigning a value of 0 yields a risk increment raster that only represents newly added risk areas.

[0038] Compared with the prior art, the present invention has at least the following beneficial effects: 1. Existing technologies often address single problems in isolation (e.g., only assessing risk, only calculating loss), lacking a systematic connection between risk changes, new areas, and yield increases. This invention completes the technology chain through grid coupling, scenario differentiation, and multiplication operations. Existing yield loss quantification is mostly based on regional totals or average loss rates, lacking spatially refined representation. This invention uses GIS grid operations to output the spatial distribution of loss increments at global and regional scales, meeting the needs of precise prevention and control. Existing multi-species risk integration ignores crop differences. This invention targets crop type (yield grids are differentiated by crop type), making the correlation between risk and loss more accurate.

[0039] 2. The risk characterization of this invention more closely reflects the actual occurrence mechanism. By coupling climate suitability with regularly selected anthropogenic driving factors through weighting, a repeatable comprehensive invasion risk raster is formed, reducing the systematic bias caused by relying solely on climate suitability. The robustness of variable selection in this invention is improved. Regularized selection is used to determine stable inclusion factors and weight parameters in the context of collinearity, suppressing coefficient drift and overfitting, and improving the generalizability of cross-scenario and cross-species migration. The identification of newly added risk areas obtained by this invention is more accurate. By using baseline / future risk difference and reclassifying negative difference pixels as zero, only newly added risk areas are extracted, avoiding the repeated inclusion of historical risks in future loss assessments. A zonal baseline subtraction mask is used to subtract the actual occurrence area in the invasion area and the current suitable area in the origin area, reducing the underestimation of incremental potential suitable areas caused by an overestimation of potential suitable areas in the invasion site. The loss results of this invention can be directly used for spatial decision-making. The method of this invention outputs a pixel-level loss increment raster consistent with yield registration, which can directly identify hotspots and support regional priority prevention and control and resource allocation. Attached Figure Description

[0040] Figure 1 This refers to the number of invasive corn species obtained from data in this embodiment of the invention; Figure 2 This refers to the number of maize invasive species after partition correction in this embodiment of the invention; Figure 3 This represents the number of potential invasive species of maize under the SSP1 climate scenario in 2050 in this embodiment of the invention. Figure 4 This refers to the number of newly added invasive species of maize under the SSP1 climate scenario in 2050 in this embodiment of the invention; Figure 5 This is the incremental production loss (in tons) in newly added risk areas under the SSP1 climate scenario in 2050 in this embodiment of the invention.

[0041] The world map used in this invention is a standard world map, which does not involve any national borders or administrative boundaries, and only shows land boundaries. The world map used in the attached drawings was downloaded from the "Standard Map Service System" (http: / / bzdt.ch.mnr.gov.cn / ), and the map approval number is "GS (2016) 1561". Detailed Implementation

[0042] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the invention will be further described below with reference to embodiments. This embodiment uses maize as the target food crop to assess the spatial distribution of maize yield loss increments caused by invasive alien species on a global scale. These embodiments are for illustrative purposes and do not constitute a limitation on the scope of protection of this invention.

[0043] Example 1: Spatial Quantification Method for Incremental Maize Yield Loss Caused by Invasive Alien Species

[0044] (1) Spatial reference and data preparation (S1) S1-1: Establish a unified spatial reference. Unify all subsequent raster data to a spatial resolution of 100 km × 100 km, and unify them to the same projection / coordinate reference and cell alignment method. Unify the mask range to the global land area or the non-zero area of ​​the maize yield raster, ensuring that each raster is fully registered at the cell level.

[0045] S1-2: Obtain the list of harmful organisms and the list of invasive alien species of maize from the CABI database, match the two, and form a list of invasive alien species related to maize.

[0046] S1-3: Obtain global distribution location data of the invasive species (e.g., public databases such as GBIF), and obtain the origin range data of the invasive species (e.g., national-scale origin range).

[0047] S1-4: Obtaining a raster of global maize yield spatial distribution. It is then resampled / aggregated to 100 km resolution and fully registered with a uniform grid.

[0048] S1-5: Obtain environmental predictor raster, including: climate factor raster (Bioclim variable set), and candidate driving factor raster set. (GDP, Human Footprint Index, Population Density, Transport Accessibility or Trade Connectivity). All factor rasters are processed to 100 km resolution according to the S1-1 rule and compared with... Registration.

[0049] Note: In this embodiment, the SSP1 scenario for 2050 uses a time window of "2041-2060" to represent the climate conditions around 2050.

[0050] (2) Generation of climate suitability grid for invasive species (S2) S2-1: For each invasive species related to maize A species distribution model (preferably the MaxEnt model) was established using its distribution point data and climate factor raster. This model was then projected under both the current climate and the 2050 SSP1 scenario to obtain the species' distribution in each pixel. Climate suitability grid .

[0051] S2-2: Will Normalization to And ensure that the output raster is registered with a uniform 100 km grid.

[0052] (3) Screening of artificial driving factors and determination of weight parameters (S3) S3-1: For candidates, which are driving factor grid sets Standardization should be performed within the effective pixel range (e.g., zero-mean, unit variance standardization) to eliminate dimensional differences. For artificial variables with obvious long-tailed distributions, logarithmic transformation followed by standardization can be performed to improve numerical stability.

[0053] S3-2: Use regularized regression to screen for stable human-driven factors and determine their weight parameters. This embodiment preferably employs elastic network regularized regression: the regularization strength parameter is selected through cross-validation. Elastic mesh hybrid parameters Determined by cross-validation; preferred method To obtain a more robust model with simpler variables; using non-zero regression coefficients as the inclusion criterion, a stable set of included factors and their... .

[0054] S3-3: To control the overestimation of model performance caused by spatial autocorrelation, spatial block cross-validation is adopted, and repeated block fitting can be performed to statistically analyze the selection frequency of variables, thereby screening a set of factors with high selection frequency and consistent sign direction.

[0055] (4) Calculation of integrated invasion risk of single species using a grid (S4) S4-1: In each pixel Above, invasive species were calculated based on climate suitability and selected anthropogenic driving factors. Overall intrusion risk value This embodiment adopts the following form: ; in This is the normalization function. To ensure feasibility, this embodiment fixes... For the Logistic normalization function: ; In step S4, when weighting and coupling climate suitability and key anthropogenic factors, an interaction modulation term between the two is introduced.

[0056] ; in , which is the interaction term weight coefficient, used to characterize the modulating effect of human-driven factors on climate adaptation risks.

[0057] S4-2: Calculate the combined risk grid for each invasive species under the current scenario and the 2050 SSP1 scenario, respectively. And ensure that the output is registered with a uniform 100 km grid.

[0058] (5) Crop-level multi-species integrated risk grid integration (S5) S5-1: Integrated risk grid for all maize-related invasive species The data is integrated according to preset rules to obtain a multi-species integrated risk grid for maize crops. .

[0059] S5-2: To ensure the uniqueness of the implementation, this embodiment consistently adopts the merger probability integration rule: ; above Alternatively, the summation and truncation normalization rule can be used: ; S5-3: Obtain the comprehensive risk grid for maize crop under the current scenario. Integrated risk grid for maize crop level under the 2050 SSP1 scenario .

[0060] (6) Extraction of newly added risk areas (S6) S6-1: Calculate the crop-level integrated risk differential grid: ; S6-2: Reclassify the difference results, retaining only newly added risk areas: when When, the corresponding pixel is assigned a value of 0; when By keeping the original values ​​unchanged, a risk increment raster representing only newly added risk areas is obtained. .

[0061] This step avoids the duplicate inclusion of existing risk areas in the baseline scenario in future assessments and prevents the negative values ​​of risk reduction areas from offsetting the incremental losses caused by new risks, ensuring that subsequent loss calculations strictly correspond to the "newly added risk areas".

[0062] S6-3: To reduce the impact of the potential suitable habitat area being too large relative to the actual occurrence area, a partitioned baseline subtraction mask is constructed during the extraction of newly added risk areas. Determining the origin mask based on invasive species origin range data With Intrusion Zone Mask Within the intrusion zone, the actual locations of the intrusion are divided into 100 km grids to generate an intrusion mask. (That is, if an occurrence point exists in the pixel, it is assigned a value of 1; otherwise, it is assigned a value of 0); within the original production area, the climate suitability raster under the current scenario is reclassified according to the threshold to generate a suitability mask. (The fixed threshold in this embodiment is) Set to 1 otherwise set to 0), and let ; Therefore, it can be done through ; Or pixel correspondence Assigning a value of 0 yields a risk increment raster that only represents newly added risk areas.

[0063] (7) Calculation and output of incremental production loss in newly added risk areas (S7) S7-1: Increase the risk increment grid for newly added risk areas Corn yield grid Pixel-by-pixel coupling calculates the incremental raster of production loss corresponding to newly added risk areas: ; S7-2: To Mapping and spatial aggregation are performed to output the spatial distribution of yield loss increments caused by new maize invasion risks at the global scale under the 2050 SSP1 scenario. Optionally, the pixel-level results can be aggregated by national or regional boundaries to obtain national / regional loss increment indicators, which can be used for food security risk assessment and identification of priority prevention and control areas.

[0064] (8) Results and Analysis The embodiment obtained Characterizes the potential increase in maize yield loss due to newly added invasion risk areas under the 2050 SSP1 scenario compared to the current scenario (e.g., Figure 1-5 (As shown). Compared with the absolute loss result obtained by directly coupling future risks and outputs without differential zeroing, the present invention uses differential and zeroing of negative difference pixels to make the output result reflect only the loss increment of newly added threat areas, which is more conducive to identifying future new risk hotspots and supporting the priority allocation of prevention and control resources.

[0065] This invention overcomes the limitations of traditional loss assessment methods, which can only perform ex-post accounting for losses that have already occurred. It constructs a multi-scenario comparative analysis system, enabling dynamic identification, spatial separation, and quantitative expression of the incremental yield loss of food crops caused by invasive alien species. Through this system, it is possible not only to identify the spatial differentiation characteristics and risk ceilings of potential loss increases under different diffusion intensities and suitable habitat conditions, but also to extend traditional static loss accounting into a forward-looking assessment oriented towards diffusion processes, changing trends, and future early warnings. This provides quantifiable, visualized, and implementable scientific evidence for prioritizing the control of invasive alien species, setting early warning thresholds, and formulating loss reduction strategies.

[0066] Compared with existing technologies, the innovation of this invention lies not only in the improvement of a single link, but also in the construction of a system integration and analysis framework for spatial decision-making. Existing technologies often focus on solving a single problem in isolation, such as only conducting invasion risk assessments or only calculating the total amount of crop yield loss. They lack a systematic connection of the entire process from risk change to the emergence of new risk areas to the output of incremental loss, thus making it difficult to accurately reveal the new sources and spatial patterns of food loss under future diffusion conditions. This invention, through a technical chain of grid coupling, scenario differencing, and loss conversion, organically connects climate suitability, anthropogenic driving factors, multi-species integrated risk, scenario differencing results, and crop yield grids, forming a logically complete, hierarchically clear, and repeatable spatial quantitative analysis system, demonstrating significant systemic and methodological innovation.

[0067] This invention also more closely reflects the actual risk characterization process. By coupling climate suitability with key anthropogenic driving factors obtained through regularization screening and weighting them, a comprehensive single-species invasion risk grid is constructed. This overcomes the systematic bias caused by relying solely on climate suitability for risk characterization, making the risk expression more consistent with the actual mechanism by which the spread of invasive alien species is driven by both natural conditions and human activities. Simultaneously, a regularization method is introduced in the variable selection stage to identify stable selection factors and their weight parameters in a collinear context, effectively suppressing coefficient drift and model overfitting, and improving the robustness and generalizability of the method in cross-scenario, cross-regional, and cross-species applications.

[0068] This invention also provides more accurate identification of newly added risk areas. By differentiating the comprehensive risks under the baseline scenario and future scenarios, only newly added risk areas are extracted, thus avoiding the duplication of historical risks into future loss assessments. Optionally, by setting a zoning baseline deduction mask to deduct the actual occurrence area from the invasion area and the current suitable growing area from the origin area, the problem of underestimating the incremental loss caused by the overestimation of the potential suitable growing area can be further reduced, improving the authenticity and reliability of the results of new risk identification and loss quantification.

[0069] In summary, this invention is not a simple aggregation of existing risk assessment or loss accounting technologies, but rather the construction of a complete analytical system encompassing risk formation mechanism characterization, key factor screening, multi-scenario differential identification, and loss increment spatial output. This system exhibits significant advantages in terms of the completeness of the technological chain, the interpretability of results, and decision support capabilities. It is particularly suitable for food security risk assessment and precise control of invasive alien species under multi-species, multi-crop, and multi-scenario conditions, demonstrating outstanding innovation and application value.

Claims

1. A spatial quantification method for incremental yield loss of food crops caused by invasive alien species based on scenario difference, characterized in that, Performing raster operations on spatially registered raster data on a computer involves the following steps: S1: Obtain the yield grid of the target food crop. The list of invasive alien species associated with the target food crop; for each invasive species, the distribution location data and the range of origin data are obtained, and the above data are registered and cropped under the same spatial resolution and mask range. S2: For each invasive species Based on its distribution location data and environmental predictor grid, a species distribution model is established and the species' distribution in the grid is generated. Climate suitability grid ; S3: Construct a raster set of candidates as driving factors Candidate factors are standardized, and regularized regression is used to screen candidates as driving factors to obtain the selected factors and their weight parameters. ; S4: In each grid Calculate the overall invasion risk value of this invasive species. The calculation is as follows: ； in This is the normalization function; S5: According to the rules, each invasive species... By integrating the grids, a comprehensive risk grid for the target food crop under the baseline scenario is obtained. and a comprehensive risk grid for future scenarios ; S6: Calculate the risk increment grid ； And The reclassified pixels are 0, resulting in a classification that only represents newly added risk areas. ; S7: Calculate the incremental production loss grid for newly added risk areas ； It also outputs the spatial distribution results of large-scale production loss increments.

2. The method according to claim 1, characterized in that, The regularization regression described in step S3 is an elastic network regularization regression. The regularization strength parameter is determined through cross-validation, and the non-zero regression coefficient is used as an artificial driving factor for selection.

3. The method according to claim 1 or 2, characterized in that, The screening of artificial driving factors in step S3 is carried out under a two-part framework, including: establishing an occurrence process model based on whether it is zero, and establishing an intensity process model on the subset of non-zero pixels, so as to obtain the corresponding set of selected factors and weight parameters respectively.

4. The method according to claim 2 or 3, characterized in that, The cross-validation adopts spatial block cross-validation to reduce the overestimation of model performance caused by spatial autocorrelation, and the selection frequency is calculated by repeated block fitting to screen stable factors.

5. The method according to claim 1, characterized in that, In step S4, when weighting and coupling climate suitability and key anthropogenic factors, an interaction modulation term is introduced, which is calculated according to the following formula: ； in , which is the interaction term weight coefficient, used to characterize the modulating effect of human-driven factors on climate adaptation risks.

6. The method according to claim 1, characterized in that, The multi-species integration rule in step S5 is one of the following formulas: (1) Summation and truncation to normalization: ;or (2) Probability of merger: .

7. The method according to claim 1, characterized in that, The integrated risk grid can be further combined with the average yield loss rate parameter of the target food crop for loss quantification. The average yield loss rate is obtained by averaging the total pest and disease loss rate of the target crop according to the number of invasive species.

8. The method according to claim 1, characterized in that, In step S6, a partitioned reference subtraction mask is constructed. Among them, the pixels of the intrusion area Occurrence mask obtained by rasterizing the actual occurrence point of the intrusion site Confirmed, the original region of origin pixels The suitability mask obtained from the current climate suitability grid reclassification Determine; and based on ； Or The corresponding pixel Assigning a value of 0 yields a risk increment raster that only represents newly added risk areas.

Images