CN105631526A - Forest pest disease outbreak risk prediction method and device - Google Patents
Forest pest disease outbreak risk prediction method and device Download PDFInfo
- Publication number
- CN105631526A CN105631526A CN201410602522.9A CN201410602522A CN105631526A CN 105631526 A CN105631526 A CN 105631526A CN 201410602522 A CN201410602522 A CN 201410602522A CN 105631526 A CN105631526 A CN 105631526A
- Authority
- CN
- China
- Prior art keywords
- forest
- leaf area
- area index
- envirment factor
- variable
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a forest pest disease outbreak risk prediction method and device, and the method comprises the steps: obtaining a known distribution region of forest pest disease outbreak; obtaining an environment factor variable which comprises a biology climate variable and a leaf area index; and predicting the forest pest disease outbreak risk according to the known distribution region and the environment factor variable through employing a pre-built ecological niche model. According to the invention, the method and device firstly employ the leaf area index to predict the forest pest disease outbreak risk, and improve the prediction accuracy of the forest pest disease outbreak risk.
Description
Technical field
The present invention relates to agriculture field, break out method and the device of risk profile in particular to a kind of forest disease and pest.
Background technology
Forest disease and pest drastically influence the existence of forest ecosystem and species as common disaster always, and the Forecasting Methodology accuracy in correlation technique is not high.
For the problem that forest disease and pest outburst risk profile accuracy in correlation technique is not high, effective solution is not yet proposed at present.
Summary of the invention
For the problem that forest disease and pest outburst risk profile accuracy in correlation technique is not high, the invention provides the method for a kind of forest disease and pest outburst risk profile and device, at least to solve the problems referred to above.
According to an aspect of the invention, it is provided the method for a kind of forest disease and pest outburst risk profile, including: obtain forest disease and pest outburst known distribution region; Obtaining envirment factor variable, wherein, described envirment factor variable includes: bioclimate variable and leaf area index; According to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk.
Alternatively, according to described known distribution region and described envirment factor variable, before utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk, also include: according to the spatial resolution of described leaf area index, described known distribution region is carried out distributed points duplicate removal process; According to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: process, according to described envirment factor variable and duplicate removal, the distributed areas obtained, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
Alternatively, obtain envirment factor variable, including: utilize the month highest temperature, lowest temperature and precipitation data to solve and obtain bioclimate variable, and obtain leaf area index; Make the spatial resolution of described bioclimate variable consistent with the spatial resolution of described leaf area index by bilinear interpolation.
Alternatively, obtain leaf area index, including: obtain the leaf area index in scheduled time sequence; Screen, from the leaf area index obtained, the leaf area index that described damage by forest-insects is corresponding according to ground mulching categorical data; The leaf area index that obtains of screening is averaged synthesis according to Preset Time granularity, obtains for predicting that described damage by forest-insects breaks out the leaf area index of risk.
Alternatively, after obtaining envirment factor variable, also include: utilize PCA that described envirment factor variable is carried out dimension-reduction treatment; According to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: the envirment factor variable obtained according to described known distribution region and dimension-reduction treatment, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
According to another aspect of the present invention, it is provided that the device of a kind of forest disease and pest outburst risk profile, including: the first acquisition module, it is used for obtaining forest disease and pest outburst known distribution region; Second acquisition module, is used for obtaining envirment factor variable, and wherein, described envirment factor variable includes: bioclimate variable and leaf area index; Prediction module, for according to described known distribution region and described envirment factor variable, utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk.
Alternatively, described device also includes: deduplication module, for according to described known distribution region and described envirment factor variable, before utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk, according to the spatial resolution of described leaf area index, described known distribution region is carried out distributed points duplicate removal process; Described prediction module, for processing, according to described envirment factor variable and duplicate removal, the distributed areas obtained, utilizes the Niche Model pre-build to predict described damage by forest-insects outburst risk.
Alternatively, described second acquisition module, including: the first acquiring unit, it is used for utilizing the month highest temperature, lowest temperature and precipitation data to solve and obtains bioclimate variable, and obtain leaf area index; First processing unit, for making the spatial resolution of described bioclimate variable consistent with the spatial resolution of described leaf area index by bilinear interpolation.
Alternatively, described second acquisition module, including: second acquisition unit, for obtaining the leaf area index in scheduled time sequence; Screening unit, for screening, from the leaf area index obtained, the leaf area index that described damage by forest-insects is corresponding according to ground mulching categorical data; Synthesis unit, for the leaf area index that obtains of screening being averaged synthesis according to Preset Time granularity, obtains for predicting that described damage by forest-insects breaks out the leaf area index of risk.
Alternatively, described device also includes: dimensionality reduction module, for, after obtaining envirment factor variable, utilizing PCA that described envirment factor variable is carried out dimension-reduction treatment; Described prediction module, for according to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: the envirment factor variable obtained according to described known distribution region and dimension-reduction treatment, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
By the present invention, use LAI data to carry out forest disease and pest outburst risk profile, analyze the precision showing to the method increase forest disease and pest outburst risk profile.
Accompanying drawing explanation
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, and the schematic description and description of the present invention is used for explaining the present invention, is not intended that inappropriate limitation of the present invention. In the accompanying drawings:
Fig. 1 is according to correlation technique;
Fig. 2 is according to embodiments of the present invention;
Fig. 3 is the whole world robur sudden death disease bursting point geographical distribution location drawing;
Fig. 4 a is the U.S./Europe SOD bursting point bioclimate variable comparison diagram one;
Fig. 4 b is the U.S./Europe SOD bursting point bioclimate variable comparison diagram two;
The Tu5Shi U.S./Europe SOD bursting point LAI comparison diagram;
Fig. 6 is that whole world SOD exists point and puppet is absent from a scattergram;
Fig. 7 a is based on the global SOD of MaxEnt Model B io and breaks out risk profile result;
Fig. 7 b is based on the global SOD of MaxEnt Model B io+LAI and breaks out risk profile result;
Fig. 7 c is based on the global SOD of GARP Model B io and breaks out risk profile result;
Fig. 7 d is based on the global SOD of GARP Model B io+LAI and breaks out risk profile result;
Fig. 7 e is based on the global SOD of GARP Model B io and breaks out risk profile result;
Fig. 7 f is based on the global SOD of GLM Model B io+LAI and breaks out risk profile result;
Fig. 7 g is based on the global SOD of SVM Model B io and breaks out risk profile result; And
Fig. 7 h is based on the global SOD of SVM Model B io+LAI and breaks out risk profile result.
Detailed description of the invention
Below with reference to accompanying drawing and describe the present invention in detail in conjunction with the embodiments. It should be noted that when not conflicting, the embodiment in the application and the feature in embodiment can be mutually combined.
The embodiment of the present invention uses leaf area index (LAI) data to carry out forest disease and pest outburst risk profile, and for a kind of forest disease and pest robur sudden death disease, combine other remotely-sensed datas, meteorological data and outburst data etc., predicting the outcome and on the basis of precision at the multiple Niche Model of relative analysis, whether the comprehensive introducing analyzing LAI can improve risk profile precision.
Fig. 1 is the flow chart of method of forest disease and pest outburst risk profile according to embodiments of the present invention, as it is shown in figure 1, the method comprising the steps of 101 to step 103:
Step 101, obtains forest disease and pest outburst known distribution region;
Step 102, obtains envirment factor variable, and wherein, described envirment factor variable includes: bioclimate variable and leaf area index;
Step 103, according to described known distribution region and described envirment factor variable, utilizes the Niche Model pre-build to predict described damage by forest-insects outburst risk.
In embodiments of the present invention, leaf area index is satellite remote sensing date.
In an embodiment of the embodiment of the present invention, above-mentioned steps 103 is according to described known distribution region and described envirment factor variable, before utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk, also include: according to the spatial resolution of described leaf area index, described known distribution region is carried out distributed points duplicate removal process. Above-mentioned steps 103 is according to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: process, according to described envirment factor variable and duplicate removal, the distributed areas obtained, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk. By this embodiment, it is ensured that distributed areas are consistent with the spatial resolution of envirment factor variable.
In an embodiment of the embodiment of the present invention, above-mentioned steps 102 obtains envirment factor variable and may include that utilizing the month highest temperature, lowest temperature and precipitation data to solve obtains bioclimate variable, and obtains leaf area index; Make the spatial resolution of described bioclimate variable consistent with the spatial resolution of described leaf area index by bilinear interpolation.
Alternatively, above-mentioned steps 102 obtains leaf area index, including: obtain the leaf area index in scheduled time sequence; Screen, from the leaf area index obtained, the leaf area index that described damage by forest-insects is corresponding according to ground mulching categorical data; The leaf area index that obtains of screening is averaged synthesis according to Preset Time granularity, obtains for predicting that described damage by forest-insects breaks out the leaf area index of risk.
In an embodiment of the embodiment of the present invention, after above-mentioned steps 102 obtains envirment factor variable, it is also possible to utilize PCA that described envirment factor variable is carried out dimension-reduction treatment. Above-mentioned steps 103 is according to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: the envirment factor variable obtained according to described known distribution region and dimension-reduction treatment, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk. By this embodiment, it is possible to reduce the complexity of data operation.
It should be noted that do not have sequencing, step 101 and step 102 that random order can be adopted to perform between above-mentioned steps 101 and step 102, also include executed in parallel.
Fig. 2 is the structured flowchart of the device of forest disease and pest outburst risk profile according to embodiments of the present invention, as in figure 2 it is shown, this device includes: the first acquisition module 10, is used for obtaining forest disease and pest outburst known distribution region; Second acquisition module 20, is used for obtaining envirment factor variable, and wherein, described envirment factor variable includes: bioclimate variable and leaf area index; Prediction module 30, is connected with the first acquisition module 10 and the second acquisition module 20, for according to described known distribution region and described envirment factor variable, utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk.
In an embodiment of the embodiment of the present invention, said apparatus can also include: deduplication module, for according to described known distribution region and described envirment factor variable, before utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk, according to the spatial resolution of described leaf area index, described known distribution region is carried out distributed points duplicate removal process. Prediction module 30, for processing, according to described envirment factor variable and duplicate removal, the distributed areas obtained, utilizes the Niche Model pre-build to predict described damage by forest-insects outburst risk.
Alternatively, above-mentioned second acquisition module 20, it is possible to including: the first acquiring unit, is used for utilizing the month highest temperature, lowest temperature and precipitation data to solve and obtains bioclimate variable, and obtain leaf area index; First processing unit, makes the spatial resolution of described bioclimate variable consistent with the spatial resolution of described leaf area index by bilinear interpolation.
Alternatively, above-mentioned second acquisition module 20, it is possible to including: second acquisition unit, for obtaining the leaf area index in scheduled time sequence; Screening unit, for screening, from the leaf area index obtained, the leaf area index that described damage by forest-insects is corresponding according to ground mulching categorical data; Synthesis unit, for the leaf area index that obtains of screening being averaged synthesis according to Preset Time granularity, obtains for predicting that described damage by forest-insects breaks out the leaf area index of risk.
In an embodiment of the embodiment of the present invention, said apparatus can also include: dimensionality reduction module, for, after obtaining envirment factor variable, utilizing PCA that described envirment factor variable is carried out dimension-reduction treatment. Prediction module 30, for according to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: the envirment factor variable obtained according to described known distribution region and dimension-reduction treatment, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
For robur sudden death disease, the embodiment of the present invention is described below.
The embodiment of the present invention provides a kind of outburst risk predicting Global Forests pest and disease damage based on LAI data product, the influence degree to modeling accuracy by kinds of risks forecast model relative analysis and LAI, whether the introducing of assessment LAI data product can improve risk profile precision, solves remotely-sensed data product and not can apply to the present situation of forest disease and pest outburst risk profile.
Based on LAI data product, combining the multi-source datas such as other satellite remote sensing date products, meteorological data, outburst data, based on kinds of risks forecast model, for a kind of forest disease and pest robur sudden death disease, analyzing LAI affects modeling result.
One, robur sudden death pathogen envirment factor data set is set up
Remotely-sensed data is collected and is processed
Owing to robur sudden death disease (SuddenOakDeath, referred to as SOD) is to endanger so far after finding the beginning of the nineties in last century, the global remote sensing Monitoring Data product in this time span should be obtained. The present invention collects the LAI data in whole world long-term sequence (1981-2011), global seismic cover classification data on Global Scale.
LAI data product is whole world long-term sequence GIMMSLAI3g product, and temporal resolution is 15 days/every half ten days, and spatial resolution is 0.083 ��. In the long-term sequence Remote Sensing Data Processing of the embodiment of the present invention, in order to improve computational efficiency, reduce data fluctuations and reduce data dimension, use for reference first quarter moon MVC (MaximumValueComposites) and synthesize thought, by every two ten days LAI Data Synthesis be monthly LAI data, and border is averaged synthesis per year, obtains 12 monthly LAI data between 1981-2011.
Robur sudden death pathogen host covers the vegetation such as Qiao, shrub, draft, only has SOD to break out risk in vegetative coverage region, and other does not have outburst risk without vegetative coverage region. Ground mulching categorical data chooses GlobCover2009, its use " the ground mulching categorizing system of the United Nations's food and agricultural organization " (UNFoodandAgricultureOrganisation ' sLandCoverClassificationSystem, LCCS) taxonomic hierarchies, totally 22 kinds of ground mulching types, spatial resolution is 300m.
Meteorological data collection and process
Whole world meteorological data have employed Dong Yingge Leah university's climatic study center CRU (ClimateResearchUnit) the Global land earth's surface month meteorological data provided, and the meteorological measuring interpolation that it is provided by more than 4000, whole world meteorological observation website gets. The latest edition provided at present is CRU3.21, contains cloud desk amount, day and night temperature, frost day rate, potential transpiration amount, precipitation, (average daily/monthly) highest temperature/lowest temperature and vapour pressure data etc. For consistent with the time range of GIMMSLAI and meet subsequent bio Climatic extract, obtain the monthly average value year after year of the highest temperature of in January, 1981 in December ,-2012, lowest temperature, precipitation, pass through bilinear interpolation, it is 0.083 �� by CRU meteorological data interpolation, consistent with LAI data spatial resolution.
Robur sudden death disease bursting point data collection and process
Robur sudden death disease only breaks out in the U.S. and European Region at present, but its area of origin is unknown, and along with traffic between various countries, trade frequently, once a large amount of host plants will be caused crushing blow by invasion, has a strong impact on the forest ecosystem of locality. The embodiment of the present invention collects the SOD bursting point data of U.S. 2000-2013 from SODMAP and the OakMapper project of U.S.'s Berkeley University's forest pathology Yu fungus laboratory, from European Union robur sudden death pathogen risk analysis group (RiskAnalysisforPhytophthoraRamorum, RAPRA) obtain 2004-2006 all Europe and infect vegetation point position information, have collected the state all over Britain of Britain 2010-2013 from Committee on Forestry of Britain (UKForestryCommission) and infect vegetation information, as shown in Figure 3.
Robur sudden death pathogen Analysis of Environmental Factors
When vegetation is infected in monitoring, station acquisition personnel collect to infect vegetation point position information as much as possible so that original bursting point data are excessively intensive. Again because the data resolution of Global Scale is relatively low, most of bursting points drop on a pixel interior (spatial resolution is 0.083 ��) of envirment factor, cause data redundancy.
Known SOD bursting point is screened by the first step, and deletion record point is the data of nursery record, because of its by seedling transportation, cultivate and cause, easy to control and do not had field vegetation big by natural environment influence; Bursting point after Preliminary screening is utilized the regular grid in ArcGIS10.1 (Fishnet) to carry out distributed points duplicate removal in grid by second step, grid is sized to 0.083 �� �� 0.083 ��, obtain pathogenic bacteria bursting point data totally 216 after duplicate removal, wherein the U.S. has 105, and Europe has 111.
Bioclimate variable analysis
Bioclimate variable by the moon value temperature data and the moon value precipitation data calculate and obtain, there is good biological significance, be the most frequently used environmental variable in Ecological niche modeling. Generally comprise 19 factors, have Annual variations (such as year-round average temperature, annual precipitation), seasonal variations (such as temperature, precipitation annual range) and extreme meteorological variables (as the coldest/the hottest month temperature, the most dry/most rainy season precipitation) etc.
By the whole world 1981-2012 CRU month highest temperature, lowest temperature and precipitation data year after year, 19 bioclimate variablees can be solved. Specifically solve and carry out under ' dismo ' package of R language. Utilize biovars function therein, input 12 months maximum temperature, minimum temperature, the meteorological data such as precipitation, data type can be vector, matrix or grid, can calculate and obtain 19 bioclimate variablees (Bio1-Bio19).
Envirment factor relative analysis
It is extracted bioclimate variable corresponding to the U.S./Europe bursting point and LAI data respectively, contrasts (Fig. 4 a and Fig. 4 b) by the box traction substation (Boxplot) of two places data, it is possible to contrast two places envirment factor statistical information intuitively. Box traction substation describes five attribute of data set: median, minima, maximum, first quartile, the 3rd quartile. Another meansigma methods of adding is by box traction substation. Box traction substation can be observed data intuitively and whether have symmetry, point spread of distribution and multisample and compare.
There is larger difference in two places bioclimate variable major part. Bioclimate variable to temperature correlation, U.S. locations SOD bursting point year-round average temperature (Bio1) the value mean height 4 DEG C than European Region SOD; The former day and night temperature month (Bio2) ratio the latter mean height 3 DEG C; Isothermal (Bio3) is equal to Bio2 and year samming excursion (Bio7) ratio, and clearly, the distribution of two places Bio3 is not overlapping for two places Bio3 distributional difference, average difference nearly 20; Other difference comparatively significantly factor is such as most dry season degree mean temperature (Bio9), and the extreme value distribution scope difference in two places is relatively big, and average differs nearly 10 DEG C. The bioclimate variable that precipitation is relevant, the differences such as the most dry monthly total precipitation (Bio14), the precipitation coefficient of variation (Bio15), most dry season degree precipitation (Bio17), most warm season degree precipitation (Bio18) are clearly, U.S. locations SOD bursting point in the most dry moon, most dry season degree and most warm season degree precipitation far below European Region, and other the moon/season, between two places, precipitation difference is not obvious especially.
The box traction substation of the U.S./two places, Europe SOD bursting point correspondence LAI value is shown in Fig. 5, has only taken four seasons LAI value here and has contrasted, and U.S. locations bursting point correspondence LAI Distribution value scope is much larger than European Region correspondence LAI value, and the former LAI average is also greater than the latter. Host's list is provided from APHIS, the host types that NA pedigree in U.S. locations infects from broad-leaf forest to Coniferous forest, shrub, herbosa etc. have distribution, and in the EU pedigree of European Region, its host types is fewer, it is mainly the shrub such as Cuculus polioephalus, pod and larch-tree.
Envirment factor dimensionality reduction
Primal environment factor variable contains 19 bioclimate variablees, 12 monthly LAI data etc. year after year, and it is excessively complicated that too much envirment factor variable makes model calculate, and there is certain dependency between the factor, there is bigger dependency as between adjacent moon LAI value. PCA (PrincipalComponentAnalysis, referred to as PCA) it is a kind of Multielement statistical analysis method that multiple variablees are filtered out by linear transformation less number linear combination variable, namely study and how to disclose the internal structure between multiple original variable by a few main constituent, and retain the information of original variable, cancelling noise variable as much as possible.
Can the embodiment of the present invention, for being used alone bioclimate variable and using LAI data to add bioclimate variable two class data to carry out dimension-reduction treatment, be analyzed LAI data and added risk profile modeling and improve model prediction accuracy.
Two, the outburst risk profile of whole world robur sudden death disease
Utilize the envirment factor collection after dimensionality reduction, predict the outburst risk of Present Global robur sudden death disease based on multiple Niche Models. The general step utilizing Niche Model prediction pest invasion distribution comprises five parts: data collection and process, modeling, evaluation of result, drop shadow spread, applicable assessment. Niche Model is by species distribution data, can be divided into and there is-be absent from model (Presence-Absencemodel, PA) with there is model (Presence-Onlymodel, PO), the present invention chooses main model maximum entropy model (MaxEnt) in two class models, rule set genetic algorithm (GARP), generalized linear model (GLM), support vector machine (SVM), the above two are PO model, being PA model both rear, this just can analyze LAI data precision of prediction in different models all sidedly.
Utilize puppet to be absent from point (pseudo-absence) to replace being absent from a little; Prevalence rate (prevalence) is the ratio that there are some quantity and all somes quantity, elects 50% as and namely there is some quantity equal to when being absent from quantity, is proved to be most suitable for modeling. There are 216 points after point processes for duplicate removal, wherein, 105 points of the U.S., 111 points in Europe, correspondence has randomly selected 216 puppets (in environmental variable scope of data) in the world and has been absent from a little, as shown in Figure 6.
Model is chosen and Accuracy Assessment
MaxEnt model is the RobertE.Schapire 3.3.3k version issued, model need to input species distribution point bit data, relating environment values, here for the aforementioned main variables (environmental variable of following model input is all consistent) chosen after principal component transform. Randomly selecting in distributed points 25% as test set, all the other 75% are training set, and iterations is 500 times, and prevalence rate takes 0.5, remaining as default setting.
GARP is run under DesktopGARP1.1.6. First, environmental variable importing Datasetmanager in software, convert its .raw form that may identify which to, required species distribution point training data saves as .csv file. Carry out first having to carry out parameter setting when GARP analyzes, including the ratio of training data, operational rule, optimal models selection, the display predicted the outcome and storage position. Set the ratio of training data, with these data genaration forecast models, then by remaining data, the model of generation is verified. Training data sets ratio and is chosen as 75%, employs whole 4 kinds of operational rules of GARP, exports with the form of ARC/INFOGrids.
The GLM function in R language ' MASS ' package is utilized to build model, wherein error distribution (errordistribution) selects binomial distribution (binomial), establish Logic Regression Models (LR, logisticregressionmodel).
Use the function ksvm in R language ' kernlab ' package to set up SVM, type selecting C-classification, Selection of kernel function Radial basis kernel function, its only one of which hyper-function sigma, and can automatically calculate.
Outburst risk profile probability graph (probabilitymap) that can obtain whole world robur sudden death pathogen is calculated by model, by a threshold value, probit is divided into the risky region with devoid of risk, is converted to binary map (binarymap) by probability graph. By carrying out classification zoning in risky region, just can the outburst risk of robur sudden death pathogen be predicted.
Choose optimal threshold (optimalthreshold) as cut point (cut-offpoint), the probability of model prediction is converted to 0-1 value, obtain existence point and puppet is absent from a corresponding probit extracted, confusion matrix (confusionmatrix) can be obtained, in Table 1. A represents known distribution point and is correctly divided into and there is point (turepositiveorpresence), corresponding b represents to be divided into by mistake exists point (falsepositiveorpresence), c represents known distribution point and is divided into by mistake and is absent from point (falsenegativeorabsence), and d represents correctly to be divided into and is absent from point (truenegativeorabsence).
Table 1 confusion matrix
In R language PresenceAbsence bag, have chosen the MaxSens+Spec (maximum sensitivity and specificity sum) the selection standard as optimal threshold, the change of its pop rate is insensitive. Herein by consultant expert and with reference to relevant document, the value in risky district is taked equidistant segmentation, here choose four classes: 0��25%, 25��50%, 50��75% and 75��100%, fire risk district is divided into corresponding four class risk class: low-risk, medium risk, high risk and excessive risk. To four kinds of models, calculating and obtain the global SOD intrusion risk result only adding LAI prediction with bioclimate variable prediction and bioclimate variable, as shown in Fig. 7 a to Fig. 7 h, the various optimal thresholds predicted the outcome are in Table 2. It is specifically intended that four kinds of Niche Models of contrast, Bio and Bio+LAI optimal threshold size is chosen and is based on two sets of data, it is impossible to it obscured.
Table 2 confusion matrix
Model prediction accuracy is evaluated
Model prediction evaluation of result index according to threshold value dependency relation, be divided into threshold value to be correlated with (threshold-dependent) and threshold value unrelated (threshold-independent) two class.
The index of the evaluation model that threshold value is relevant comprises overall success (OverallPredictionSuccess, OPS), sensitivity (Sensitivity), specificity (Specificity), Kappa coefficient, true skill statistics (TrueSkillStatistic, TSS) etc.
Another kind of evaluation index is unrelated with threshold value, ROC curve is the abbreviation of Receiver operating curve (ReceiverOperatingCharacteristicCurve), originate from statistical decision theory, be used for the relation illustrating between grader hit rate and rate of false alarm. ROC curve as possible judgement dividing value, thus calculates each value predicted the outcome and obtains corresponding sensitivity and specificity, be with sensitivity for vertical coordinate, and 1-specificity is the curve that abscissa is formed. The area that ROC curve and abscissa surround is AUC (theAreaUndertheROCCurve), and AUC is because not by the impact of threshold value, being presently the most one of conventional model-evaluation index. AUC means more greatly more big with random distribution spacing, and the dependency between the species distribution model of environmental variable and prediction is more big, namely illustrates that model prediction result is more good. It is generally acknowledged that AUC diagnoses unsuccessfully when being 0.5-0.6, during 0.7-0.8, diagnostic value is general, and during 0.8-0.9, diagnostic value is better, and during more than 0.9, diagnostic value is outstanding.
The each evaluation index result of calculation of gained is in Table 3. In the middle of eight models, OPS, TSS of SVM_Bio+LAI model and AUC are best, and namely it is affected by threshold value and is not subject to threshold value impact assessment index is all optimum, and the two of GLM model-evaluation index values are minimum. Meanwhile, to by threshold value impact assessment index, the introducing of LAI all improves the precision of prediction of each model; On the AUC not affected by threshold value, the introducing impact of LAI is little. Therefore, sometimes rely solely on AUC and can not think that model prediction accuracy is just higher unilaterally, it should consider many evaluation indexes comprehensively. It can thus be appreciated that, the introducing of LAI can be greatly promoted the precision of prediction of each model, utilize LAI and traditional meteorological data affecting forest disease and pest, than being used alone, meteorological data prediction forest disease and pest outburst risk precision is high, estimation range is more reasonable, can reflect mechanism that somewhere forest suffers its vegetation growth situation after pest and disease damage, be significant to satellite remote sensing date product is incorporated in the sector application of forest disease and pest simultaneously.
Table 3 model accuracy evaluation result
As can be seen from the above description, present invention achieves following technique effect: use LAI data to carry out forest disease and pest outburst risk profile, improve the precision of forest disease and pest outburst risk profile.
Obviously, those skilled in the art should be understood that, each module of the above-mentioned present invention or each step can realize with general calculation element, they can concentrate on single calculation element, or it is distributed on the network that multiple calculation element forms, alternatively, they can realize with the executable program code of calculation element, thus, can be stored in storage device is performed by calculation element, and in some cases, shown or described step can be performed with the order being different from herein, or they are fabricated to respectively each integrated circuit modules, or the multiple modules in them or step are fabricated to single integrated circuit module realize. so, the present invention is not restricted to the combination of any specific hardware and software.
The foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, for a person skilled in the art, the present invention can have various modifications and variations. All within the spirit and principles in the present invention, any amendment of making, equivalent replacement, improvement etc., should be included within protection scope of the present invention.
Claims (10)
1. the method for a forest disease and pest outburst risk profile, it is characterised in that including:
Obtain forest disease and pest outburst known distribution region;
Obtaining envirment factor variable, wherein, described envirment factor variable includes: bioclimate variable and leaf area index;
According to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk.
2. method according to claim 1, it is characterised in that
According to described known distribution region and described envirment factor variable, before utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk, also include: according to the spatial resolution of described leaf area index, described known distribution region is carried out distributed points duplicate removal process;
According to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: process, according to described envirment factor variable and duplicate removal, the distributed areas obtained, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
3. method according to claim 1, it is characterised in that obtain envirment factor variable, including:
Utilize the month highest temperature, lowest temperature and precipitation data to solve and obtain bioclimate variable, and obtain leaf area index;
Make the spatial resolution of described bioclimate variable consistent with the spatial resolution of described leaf area index by bilinear interpolation.
4. method according to claim 1, it is characterised in that obtain leaf area index, including:
Obtain the leaf area index in scheduled time sequence;
Screen, from the leaf area index obtained, the leaf area index that described damage by forest-insects is corresponding according to ground mulching categorical data;
The leaf area index that obtains of screening is averaged synthesis according to Preset Time granularity, obtains for predicting that described damage by forest-insects breaks out the leaf area index of risk.
5. the method according to claim 1,3 or 4, it is characterised in that
After obtaining envirment factor variable, also include: utilize PCA that described envirment factor variable is carried out dimension-reduction treatment;
According to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: the envirment factor variable obtained according to described known distribution region and dimension-reduction treatment, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
6. the device of a forest disease and pest outburst risk profile, it is characterised in that including:
First acquisition module, is used for obtaining forest disease and pest outburst known distribution region;
Second acquisition module, is used for obtaining envirment factor variable, and wherein, described envirment factor variable includes: bioclimate variable and leaf area index;
Prediction module, for according to described known distribution region and described envirment factor variable, utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk.
7. device according to claim 6, it is characterised in that
Described device also includes: deduplication module, for according to described known distribution region and described envirment factor variable, before utilizing the Niche Model pre-build to predict described damage by forest-insects outburst risk, according to the spatial resolution of described leaf area index, described known distribution region is carried out distributed points duplicate removal process;
Described prediction module, for processing, according to described envirment factor variable and duplicate removal, the distributed areas obtained, utilizes the Niche Model pre-build to predict described damage by forest-insects outburst risk.
8. device according to claim 6, it is characterised in that described second acquisition module, including:
First acquiring unit, is used for utilizing the month highest temperature, lowest temperature and precipitation data to solve and obtains bioclimate variable, and obtain leaf area index;
First processing unit, makes the spatial resolution of described bioclimate variable consistent with the spatial resolution of described leaf area index by bilinear interpolation.
9. device according to claim 6, it is characterised in that described second acquisition module, including:
Second acquisition unit, for obtaining the leaf area index in scheduled time sequence;
Screening unit, for screening, from the leaf area index obtained, the leaf area index that described damage by forest-insects is corresponding according to ground mulching categorical data;
Synthesis unit, for the leaf area index that obtains of screening being averaged synthesis according to Preset Time granularity, obtains for predicting that described damage by forest-insects breaks out the leaf area index of risk.
10. the device according to claim 6,8 or 9, it is characterised in that
Described device also includes: dimensionality reduction module, for, after obtaining envirment factor variable, utilizing PCA that described envirment factor variable is carried out dimension-reduction treatment;
Described prediction module, for according to described known distribution region and described envirment factor variable, the Niche Model pre-build is utilized to predict described damage by forest-insects outburst risk, including: the envirment factor variable obtained according to described known distribution region and dimension-reduction treatment, utilize the Niche Model pre-build to predict described damage by forest-insects outburst risk.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410602522.9A CN105631526A (en) | 2014-10-31 | 2014-10-31 | Forest pest disease outbreak risk prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410602522.9A CN105631526A (en) | 2014-10-31 | 2014-10-31 | Forest pest disease outbreak risk prediction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105631526A true CN105631526A (en) | 2016-06-01 |
Family
ID=56046434
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410602522.9A Withdrawn CN105631526A (en) | 2014-10-31 | 2014-10-31 | Forest pest disease outbreak risk prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105631526A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108062408A (en) * | 2017-12-28 | 2018-05-22 | 闻丞 | A kind of data processing method and device |
| CN109064460A (en) * | 2018-08-01 | 2018-12-21 | 中国科学院合肥物质科学研究院 | Wheat severe plant disease prevention method based on multiple timings property element depth characteristic |
| CN110378540A (en) * | 2019-08-02 | 2019-10-25 | 桂林理工大学 | A kind of Weighted Atmospheric Temperature Used calculation method suitable for Beibu Bay, guangxi area |
| CN110837911A (en) * | 2019-09-06 | 2020-02-25 | 沈阳农业大学 | Large-scale ground surface arthropod space distribution simulation method |
| CN111260148A (en) * | 2020-02-10 | 2020-06-09 | 长江大学 | Pine wood nematode invasion risk prediction method based on ecological niche factor model |
| CN111723328A (en) * | 2020-06-18 | 2020-09-29 | 中国科学院空天信息创新研究院 | Leaf area index time series reconstruction method, device, equipment and storage medium |
| CN113051314A (en) * | 2021-03-25 | 2021-06-29 | 东北林业大学 | Pine wood nematode propagation factor determination device and method based on data acquisition and analysis |
| CN114154428A (en) * | 2022-02-09 | 2022-03-08 | 中国科学院地理科学与资源研究所 | Target species potential intrusion risk assessment method and device |
| CN114707813A (en) * | 2022-03-15 | 2022-07-05 | 青海省畜牧兽医科学院 | Echinococcus multilocularis natural epidemic source classification prediction method based on ecological niche model |
| CN115796366A (en) * | 2022-12-02 | 2023-03-14 | 同济大学 | Forest pest prediction method and forest pest prediction map system |
| CN116681279A (en) * | 2023-05-25 | 2023-09-01 | 北京健卫病媒有害生物防控中心 | Forestry pest control risk assessment system based on data analysis |
| CN116863410A (en) * | 2023-09-05 | 2023-10-10 | 深圳市中大合顺生物科技有限公司 | Data acquisition and processing method and system based on pest control |
| CN117253145A (en) * | 2023-09-15 | 2023-12-19 | 沈阳农业大学 | Method and system for determining growth situation of broom corn millet and mung bean in different intercropping modes |
-
2014
- 2014-10-31 CN CN201410602522.9A patent/CN105631526A/en not_active Withdrawn
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108062408A (en) * | 2017-12-28 | 2018-05-22 | 闻丞 | A kind of data processing method and device |
| CN109064460A (en) * | 2018-08-01 | 2018-12-21 | 中国科学院合肥物质科学研究院 | Wheat severe plant disease prevention method based on multiple timings property element depth characteristic |
| CN109064460B (en) * | 2018-08-01 | 2021-09-28 | 中国科学院合肥物质科学研究院 | Wheat severe disease prediction method based on multiple time sequence attribute element depth characteristics |
| CN110378540B (en) * | 2019-08-02 | 2023-05-09 | 桂林理工大学 | Atmospheric weighted average temperature calculation method suitable for northern Guangxi Bay region |
| CN110378540A (en) * | 2019-08-02 | 2019-10-25 | 桂林理工大学 | A kind of Weighted Atmospheric Temperature Used calculation method suitable for Beibu Bay, guangxi area |
| CN110837911B (en) * | 2019-09-06 | 2021-02-05 | 沈阳农业大学 | Large-scale ground surface arthropod space distribution simulation method |
| CN110837911A (en) * | 2019-09-06 | 2020-02-25 | 沈阳农业大学 | Large-scale ground surface arthropod space distribution simulation method |
| CN111260148A (en) * | 2020-02-10 | 2020-06-09 | 长江大学 | Pine wood nematode invasion risk prediction method based on ecological niche factor model |
| CN111723328A (en) * | 2020-06-18 | 2020-09-29 | 中国科学院空天信息创新研究院 | Leaf area index time series reconstruction method, device, equipment and storage medium |
| CN111723328B (en) * | 2020-06-18 | 2024-02-13 | 中国科学院空天信息创新研究院 | Leaf area index time sequence reconstruction method, device, equipment and storage medium |
| CN113051314A (en) * | 2021-03-25 | 2021-06-29 | 东北林业大学 | Pine wood nematode propagation factor determination device and method based on data acquisition and analysis |
| CN114154428A (en) * | 2022-02-09 | 2022-03-08 | 中国科学院地理科学与资源研究所 | Target species potential intrusion risk assessment method and device |
| CN114154428B (en) * | 2022-02-09 | 2022-09-23 | 中国科学院地理科学与资源研究所 | Target species potential intrusion risk assessment method and device |
| CN114707813A (en) * | 2022-03-15 | 2022-07-05 | 青海省畜牧兽医科学院 | Echinococcus multilocularis natural epidemic source classification prediction method based on ecological niche model |
| CN115796366A (en) * | 2022-12-02 | 2023-03-14 | 同济大学 | Forest pest prediction method and forest pest prediction map system |
| CN116681279A (en) * | 2023-05-25 | 2023-09-01 | 北京健卫病媒有害生物防控中心 | Forestry pest control risk assessment system based on data analysis |
| CN116681279B (en) * | 2023-05-25 | 2024-04-19 | 北京健卫病媒有害生物防控中心 | Forestry pest control risk assessment system based on data analysis |
| CN116863410A (en) * | 2023-09-05 | 2023-10-10 | 深圳市中大合顺生物科技有限公司 | Data acquisition and processing method and system based on pest control |
| CN116863410B (en) * | 2023-09-05 | 2023-12-22 | 深圳市中大合顺生物科技有限公司 | Data acquisition and processing method and system based on pest control |
| CN117253145A (en) * | 2023-09-15 | 2023-12-19 | 沈阳农业大学 | Method and system for determining growth situation of broom corn millet and mung bean in different intercropping modes |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105631526A (en) | Forest pest disease outbreak risk prediction method and device | |
| Bey et al. | Mapping smallholder and large-scale cropland dynamics with a flexible classification system and pixel-based composites in an emerging frontier of Mozambique | |
| Fernandes et al. | Assessment of land use and land cover changes and valuation of carbon stocks in the Sergipe semiarid region, Brazil: 1992–2030 | |
| Long et al. | Object-oriented crop classification using multitemporal ETM+ SLC-off imagery and random forest | |
| Zhang et al. | Multi-scale habitat selection by two declining East Asian waterfowl species at their core spring stopover area | |
| Do et al. | Monitoring landscape fragmentation and aboveground biomass estimation in Can Gio Mangrove Biosphere Reserve over the past 20 years | |
| Zhou et al. | Object-oriented classification of sugarcane using time-series middle-resolution remote sensing data based on AdaBoost | |
| Ayalew et al. | Land use change and climate dynamics in the Rift Valley Lake Basin, Ethiopia | |
| Farrow et al. | Assessing the risk of root rots in common beans in East Africa using simulated, estimated and observed daily rainfall data | |
| LI et al. | Spatio-temporal changes in rice area at the northern limits of the rice cropping system in China from 1984 to 2013 | |
| Melati | The use of remote sensing data to monitor land use systems and forest variables of the tropical rainforest landscape under transformation in Jambi Province, Sumatra, Indonesia | |
| Kumari et al. | Object-based machine learning approach for soybean mapping using temporal sentinel-1/sentinel-2 data | |
| Xiao et al. | Regional prediction of Fusarium head blight occurrence in wheat with remote sensing based Susceptible-Exposed-Infectious-Removed model | |
| Zhu et al. | Identification of soybean based on Sentinel-1/2 SAR and MSI imagery under a complex planting structure | |
| Alberto et al. | Disease risk map of anthracnose-twister of onion based on previous disease locations as a future predictors | |
| Song et al. | Potential global distribution of the guava root-knot nematode Meloidogyne enterolobii under different climate change scenarios using MaxEnt ecological niche modeling | |
| Tang et al. | Forecasting wheat aphid with remote sensing based on relevance vector machine | |
| Neta et al. | Seasonal forecasting of pest population dynamics based on downscaled SEAS5 forecasts | |
| Sitokonstantinou et al. | A Sentinel based agriculture monitoring scheme for the control of the CAP and food security | |
| Akkermans et al. | Quantifying successional land cover after clearing of tropical rainforest along forest frontiers in the Congo Basin | |
| Graw et al. | Bush Encroachment Mapping for Africa: Multi-scale analysis with remote sensing and GIS | |
| Janiola et al. | Land use and land cover (LULC) change detection using multitemporal landsat imagery: A cas e study in Allah Valley Landscape in Southern, Philippines | |
| Kamenova et al. | Crop Type Mapping and Winter Wheat Yield Prediction Utilizing Sentinel-2: A Case Study from Upper Thracian Lowland, Bulgaria | |
| Mayer et al. | Employing the agricultural classification and estimation service (ACES) for mapping smallholder rice farms in Bhutan | |
| Chaiyana et al. | Mapping and predicting cassava mosaic disease outbreaks using earth observation and meteorological data-driven approaches |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20160601 |
|
| WW01 | Invention patent application withdrawn after publication |