CN103955606A - Remote sensing technology-based grassland locust plague progressive prediction method - Google Patents

Abstract

The invention discloses a remote sensing technology-based grassland locust plague progressive prediction method. According to the method, the distribution of key habitat elements which influencing the development of a grassland locust population is obtained by the means of quantitative remote sensing inversion, meteorological station observation and the like, wherein the key habitat elements subjected to remote sensing inversion comprises land surface temperature, vegetation coverage and soil moisture; the spawning suitability, the incubation suitability and the growth suitability of locusts are analyzed quantitatively by establishing an evaluation model, and a locust plague risk early prediction model is established; a locust plague risk level prediction result is corrected by utilizing the remote sensing observation of an incubation period and a third period and the locust density data measured in the field according to the incubation and development time axes of the locusts, and the situations of a grassland locust plague are predicted progressively. According to the technical scheme provided by the invention, the progressive update of sensitive habitat elements is obtained by performing quantitative inversion on remote sensing data with higher time resolution, so that the prediction precision of a locust plague monitoring and prediction model is improved.

Description

A kind of gradual Forecasting Methodology of the grassland plague of locusts based on remote sensing technology

Technical field

The present invention relates to damage control and earth observation and field of navigation technology, be specifically related to one and utilize RS and GIS systems technology, set up the method for the gradual prediction of pastoral area-Nong district's plague of locusts according to the stage of development of grassland grasshopper, for disaster surveillance and the forecast of the grassland plague of locusts provide the technical scheme with higher precision of prediction.

Background technology

For a long time, the work of locust detecting and reporting pest information is considered to one of main task of meadow administrative authority, plant protection department, and carrying out timely and effectively locust diaster prevention and control is a very urgent task.But China's grassland area is very vast, is mainly distributed in northwest, North China and southern mountain, traffic is inconvenience very, control personnel and control equipment wretched insufficiency again, so that bear the character of much blindness and passivity in plague of locusts control.Although country and local annual input are prevented and treated fund in a large number, produce effects not remarkable, Chang Yin affects adversely and prevents and treats best period and cause heavy economic losses.

At present, Grasshopper Population grows with plague of locusts disaster surveillance not have what especially effectively way, and it is also few that domestic and foreign literature is consulted finding report, especially lacks the disaster surveillance forecasting procedure for prairie soil locust.Integrate and see, in prior art, mainly rely on research station, the grassland fixed point on ground to carry out the observation of population density, according to the variation of different period Grasshopper Population density, the method that dependence people is principal commander's micro-judgment is carried out the prediction of plague of locusts the condition of a disaster.Wherein, the prediction that Grasshopper Population is grown is crucial, mainly carry out based on biological method at present: (1), according to the observation of locust insect life habit and experiment, determines the growth and development environment factor range (as humiture) that Grasshopper Population is suitable; (2) grow the field observation in " length of time " by locust and analyze population development situation, thereby judge whether the plague of locusts can occur.Such as, plague of locusts research station, general grassland can enter the region collection pedotheque of selecting to occur last year the plague of locusts after the winter, go back to laboratory and analyze locust worm's ovum density, in conjunction with meteorological observation, infer the ratio of surviving the winter of worm's ovum, then Second Year at the end of spring and the beginning of summer when, then sampling observation egg hatch situation, and the development condition in different " length of time ", thereby judge that possibility (risk) and the order of severity thereof of the plague of locusts occur in certain region " qualitative ".

Existing biological method is counted as the means of mainly observing and predicting of current administrative authority, but the concept of geographical space is often out in the cold.The object that locust is observed and predicted, not only will be quantitatively with time of origin on, and on geographic position, grasp plague of locusts generation development, need when, where to answer, may occur the plague of locusts of which kind of degree, administrative authority just can take effective damage control means, prevents and reduces natural disasters and the disaster relief.Ignored spatial information, only relied on very some position observation (monitoring station and meteorological site) result of limitation, carried out observing and predicting of plague of locusts the condition of a disaster, both lacked accuracy, also deficiency thinks that administrative authority provides support.In fact, the information overwhelming majority that locust relates in observing and predicting is relevant with locus, comprising the enquiry data of locust itself as the distribution of locust and population structure, locust density and Schistosome eggs distribution, and the information of all kinds of habitats key element.Can think, observe and predict in work locust, the concept in space permeate in from data acquisition, data analysis to the overall process of result output, process the information just remote sensing relevant with locus, the speciality of Geographic Information System.Developing rapidly of remote sensing technology, increasing sensor in orbit, and formed star, sky, ground integrated multi-platform remote sense monitoring system, provide strong instrument for continuously, dynamically monitoring and extract various environmental informations on a large scale, and technical day by day ripe.Utilize the ecological environmental condition that remote sensing technology can be depended on for existence according to locust to carry out space surveillance, determine the possible position that the plague of locusts occurs, the risk that the prediction plague of locusts occurs, for plant protection department provides the effectively scientific basis of control of the plague of locusts.Meanwhile, Geographic Information System can be used for tissue, management, analysis and the modeling of spatial data, for plague of locusts monitoring provides spatial analysis and modeling tool with prediction.Therefore utilize method that remote sensing technology combines with Geographic Information System to carry out locust monitoring and become new trend.

From Present Domestic, use RS and GIS system in the present situation of locust disaster monitoring and forecast, one, mainly concentrate on and utilize remotely-sensed data to provide habitat parameter in a big way to portray, but mainly the method based on Classification in Remote Sensing Image is distinguished different soil cover types, then grow with the substantial connection of green vegetation and forecast the possibility that the plague of locusts occurs according to Grasshopper Population; Its two, only use GIS to carry out growing the functions such as the management of relevant factor of the habitat (as temperature, precipitation), visual and cartography export with locust.In these methods or lack description to the Grasshopper Population stage of development, just utilized Geographic Information System that the functions such as data management, visual or cartography export are provided for plague of locusts risk forecast, to plague of locusts risk, forecast is difficult to provide the scheme of decision support technique truly.

The Grasshopper Population that causes calamity roughly can be divided into the migratory locusts (as Asiatic migrotory locust, Central Asia migratory locusts etc.) with the transfer ability that flies more by force, and the weak main home environment that relies on of transfer ability is grown the grassland " native locust " developing.The kind of prairie soil locust is more, as common in Xinjiang just have more than 10 plant more than, and the life habit of this two classes locust and to cause the process difference of calamity larger.Because Grasshopper Population is grown and is subject to the impact of the many factors such as meteorology, soil, vegetation, thereby whether the Grasshopper Population in a certain year somewhere can amount reproduction, so that the last plague of locusts that forms, and has larger time and space uncertainty.And because data acquisition difficulty only inputs to build plague of locusts risk model according to locust habit and disposable habitat key element, tending to exist predicated error, this is also difficult point and the problem of current plague of locusts risk profile.

Summary of the invention

For above-mentioned the deficiencies in the prior art, overcome while being difficult to obtain each factor of the habitat empty in the current locust forecast based on website observation and biological method and change, predict the outcome and lack enough space orientation, and precision of prediction is not high-leveled and difficult thinks the problems such as administrative authority's application, the invention provides a kind of gradual, there is the Forecasting Methodology of self-control, mainly cause mainly for China's grassland grasshopper disaster that calamity person---prairie soil locust carries out plague of locusts the condition of a disaster Risk Monitoring and forecast: the spatial and temporal distributions that first makes full use of quantitative remote sensing inversion technique and set up the crucial habitat key element that affects grassland grasshopper population development, the up-to-date observation data of obtaining in conjunction with remote sensing in Grasshopper Population growth course again and meteorological site data, set up a kind of gradual grassland plague of locusts forecast and the condition of a disaster risk evaluating method based on the Grasshopper Population stage of development.Gradual main finger, the prediction of the method is not to provide once net result, but the different critical stages of growing along with the variation of habitat conditions and Grasshopper Population, model can be revised adjustment.This invention can obtain by the remote-sensing inversion of high time resolution the gradual renewal of responsive habitat key element, thereby improves the precision that predicts the outcome of plague of locusts monitoring and prediction model.

Technical scheme of the present invention is:

A kind of gradual Forecasting Methodology of the grassland plague of locusts based on remote sensing technology, the method obtains the spatial and temporal distributions of the crucial habitat key element that affects grassland grasshopper population development by quantitative remote sensing inversion technique and meteorological site observation data, obtain locust disaster risk index by building plague of locusts risk forecast model, again locust disaster risk index is revised, thereby obtain the gradual prediction to grassland plague of locusts the condition of a disaster, comprise the steps:

1) a complete Grasshopper Population is divided into the stage of laying eggs growth cycle, hatching stage and growth phase, for the Grasshopper Population three phases of growth cycle, obtain by quantitative remote sensing inverting the crucial habitat key element that Grasshopper Population is grown, comprise land surface temperature, vegetation coverage and soil moisture, by calculating suitable index (the Oviposition Suitability Index that obtains respectively laying eggs, OSI), hatch suitable index (Incubation Suitability Index, and growth suitable index (Development Suitability Index ISI), DSI),

2) by the suitable index of laying eggs, hatch grow suitable index of suitable exponential sum and calculate and obtain locust disaster risk index, as plague of locusts risk profile result; Gradual prediction provided by the invention comprises that initial predicted, incubation period are revised and revise three length of times:

2.1) in the time that locust egg not yet starts to hatch, carry out initial plague of locusts risk profile, by the suitable index of laying eggs in this Grasshopper Population growth cycle, hatch the suitable index of growth in a cycle in suitable exponential sum and calculate, obtain initial locust disaster risk index, as initial plague of locusts risk profile result;

2.2) in the time that Locust ovum hatching finishes, carry out incubation period correction, obtain the suitable index of new hatching by these stylish remote sensing monitoring data, more initial plague of locusts risk index is revised, thereby obtain incubation period revised plague of locusts risk profile result;

2.3) in three length of times of locust, carry out correction in three length of times, by actual remote sensing monitoring data and the meteorological rainfall spatial interpolation data to three length of times after Locust ovum hatching, obtain the suitable index of new growth, again further to step 2.2) plague of locusts risk index that obtains revises, as finally predicting the outcome.

In above-mentioned Forecasting Methodology, step 1) method that obtains soil moisture by quantitative remote sensing inverting is the segmentation inversion method that the present invention proposes.First the method is divided into three types by earth's surface covering state, is respectively the soil under soil and the covering of airtight vegetation under bare soil, sparse vegetation covering; Then using pixel as elementary cell, divide to distinguish above-mentioned three kinds of vegetation cover types by the threshold value of remote sensing normalized differential vegetation index, thereby complicated earth surface is covered and simplified; For three kinds of vegetation cover types, then carry out soil moisture retrieval and optimization by different inverse models, be specially: for bare soil, obtain soil moisture by the inverting of thermal inertia model method; Soil under covering for airtight vegetation, obtains soil moisture by the inverting of temperature vegetation drought index; Soil under covering for sparse vegetation, obtains soil moisture by heat of mixing inertia and the inverting of temperature vegetation drought index model.

Step 1 of the present invention) obtain vegetation coverage by quantitative remote sensing inverting method for based on improved pixel two sub-model inversion methods, the method by by one from different places the vegetation coverage data of the fieldwork sample prescription of soil cover type carry out statistical regression with the normalized differential vegetation index that calculates from remotely-sensed data, obtain the normalized differential vegetation index value of theoretic pure vegetation and pure soil pixel by least square method, then the two sub-model invertings of substitution pixel obtain the vegetation coverage of this area.

In above-mentioned Forecasting Methodology, step 1) to obtain the suitable index of laying eggs be according to locust lay eggs vegetation coverage, the soil texture and the soil moisture situation in stage, the structure suitable index of laying eggs, for representing inhibition or the suitable situation that habitat key element is laid eggs to locust, characterize the locust successful possibility of laying eggs, detailed process comprises:

C1) obtain sandy loam index by calculating the amassing of soil content index and clay content index, and be normalized, obtain the soil types factor (Soil Type Factor, STF);

C2) temperature vegetation drought index (the Temperature Vegetation Dryness Index that utilizes actual measurement soil humidity data and obtained by remotely-sensed data inverting, TVDI) carry out statistical study, set up the remote-sensing inversion model of soil moisture, and be used for calculating Soil moisture factor (Soil Moisture Factor, SMF);

C3) utilize remotely-sensed data inverting to obtain the vegetation coverage in egg-laying season, build egg-laying season vegetation blanketing fctor (Vegetation Factor for OSI, VFO);

C4) the soil types factor, Soil moisture factor and egg-laying season vegetation blanketing fctor are first normalized, then calculate and obtain the suitable index of laying eggs by linear weighted function method.

Above-mentioned acquisition is laid eggs in the linear weighted function method of suitable index, in one embodiment of the invention, to the weight of the soil types factor, Soil moisture factor and egg-laying season vegetation blanketing fctor respectively value be 0.3,0.5 and 0.2, for representing the difference of three factor significance levels in region.

In above-mentioned Forecasting Methodology, step 1) obtain hatching suitable index be the soil moisture factor and the humidity factor that first obtains Locust ovum overwintering survival rate, incubation period, obtain the suitable index of hatching by calculating again, the impact of locust egg successfully being hatched for characterizing severe winter and incubation period habitat conditions, detailed process comprises:

F1) obtain Locust ovum overwintering survival rate according to the time that cold wave arrives for the first time by empirical statistics estimation;

F2) close on the land surface temperature calculating of period by the locust incubation period in this cycle and obtain temperature factor, the inverting of recycling remotely-sensed data obtains temperature vegetation drought index (TVDI), calculates the humidity factor before the incubation period;

F3) temperature factor to Locust ovum overwintering survival rate, incubation period and humidity factor are first normalized, and are then calculated and are obtained the suitable index of hatching by product method.

In above-mentioned Forecasting Methodology, step 1) obtain growth suitable index be the geographical height above sea level factor (the Elevation Factor that first obtains estimation range, EF), locust vegetation pattern in the growth period factor (Vegetation Type Factor, and vegetation blanketing fctor (Vegetation Cover Factor2, VCF VTF) ₂), then obtain the suitable index of growth (DSI) of locust growth and development stage by calculating, and from hatching successfully, ovum is subject to the inhibiting effect of habitat key element around for quantitative description locust, and detailed process comprises:

S1) according to the height above sea level scope of geographical sea level elevation and the suitable life of locust, calculate the height above sea level factor by elevation segmentation;

S2) utilize remotely-sensed data inverting to obtain the vegetation coverage of locust growth phase, according to the vegetation coverage scope that is applicable to locust and grows, calculate the vegetation blanketing fctor of locust growth phase;

S3) obtain the vegetation pattern factor of locust growth phase according to vegetation pattern data;

S4) the vegetation blanketing fctor to the height above sea level factor, locust growth phase and the vegetation pattern factor are normalized, then calculate the suitable index of growth by linear weighted function method.

In one embodiment of the invention, above-mentioned steps S4) the linear weighted function method that obtains the suitable index of growth be vegetation blanketing fctor to the height above sea level factor, locust growth phase and the vegetation pattern factor weight respectively value be 0.2,0.5 and 0.3, for representing the difference of three factor significance levels in region.

In Forecasting Methodology provided by the invention, step 2.3) carry out, in three length of time makeover process, need to recalculating the suitable index of growth, and having added the rainfall factor (Rainfall Factor, RF), the impact that reflection rainfall is grown on locust.Detailed process is as follows:

X1) according to the height above sea level scope of geographical sea level elevation and the suitable life of locust, calculate the height above sea level factor by elevation segmentation;

X2) utilize remotely-sensed data inverting to obtain the vegetation coverage of locust growth phase, according to the vegetation coverage scope that is applicable to locust and grows, calculate the vegetation blanketing fctor of locust growth phase;

X3) obtain the vegetation pattern factor of locust growth phase according to vegetation pattern data;

X4) observe the rainfall amount data that obtain carry out after space interpolation according to meteorological site, calculate the rainfall factor (RF);

X5) the vegetation blanketing fctor to the height above sea level factor, locust growth phase, the vegetation pattern factor and the rainfall factor are normalized, then are recalculated and obtained revising the suitable index of growth later by linear weighted function method.

Wherein, X1), X2) and X3) the height above sea level factor, the vegetation blanketing fctor of locust growth phase and the computing method of the vegetation pattern factor in step with step S1), S2) and S3).Step X4) in the rainfall factor calculate according to maximum in region, minimum rainfall amount.In one embodiment of the invention, the linear weighted function method of the suitable index of growth that above-mentioned acquisition is revised for three length of times be vegetation blanketing fctor, the vegetation pattern factor and the rainfall factor to the height above sea level factor, locust growth phase weight respectively value be 0.15,0.40 and 0.2 and 0.25, for representing the difference of four factor significance levels in region.The rainfall factor is only in step 5) three length of times revise just and introduce because at this moment rainfall is by limiting the activity of locust, locust is caused to calamity and plays certain inhibiting effect.In step 3) initial plague of locusts risk profile and step 4) incubation period revise, all without introducing the rainfall factor.

Further, step S1) and step X1) the intermediate altitude factor is that the elevation upper limit and the lower limit of growing according to the most applicable locust in a region maximum elevation value, minimum height value and this region calculate acquisition, the elevation upper limit and the lower limit needs of the most applicable described locust growth are studied acquisition by field study.In one embodiment of the invention, the elevation upper and lower limit that the applicable grassland grasshopper in Xinjiang region grows is respectively 2300 meters and 600 meters.

In one embodiment of the invention, the time that the above-mentioned gradual Forecasting Methodology of the grassland plague of locusts based on remote sensing technology is carried out initial plague of locusts risk profile is the annual last ten-days period in April; The time of carrying out incubation period correction is the last ten-days period in May; The time of carrying out correction in three length of times is the last ten-days period in June.

Beneficial effect of the present invention:

The present invention makes full use of advanced earth observation technology, by the spatial and temporal distributions information of quantitative remote sensing inverting means quick obtaining locust habitat key element, three critical stages of growing according to locust again, a kind of gradual plague of locusts Risk Forecast Method is proposed, its result precision relies on higher than tradition predicting the outcome that biological method and meteorological site observation data obtain, and the prevention and control that can be locust disaster provide early warning support.Concrete advantage is as follows:

(1) utilize satellite remote sensing date and the quantitative inversion algorithm can the rapid extraction space distribution information of locust habitat key element on a large scale, by setting up quantitative evaluation model, can realize the suitability degree evaluation that grow locust in certain habitat;

(2) three critical periods of locust being grown: the crucial period that egg-laying season, incubation period and growth period change as prediction Grasshopper Population, the coupling relation of each ecological factor and locust life habit be can portray preferably, plague of locusts risk evaluation model and method built;

(3) due in incubation period and growth period, the variation of habitat conditions can affect the variation of Grasshopper Population density, or be explosive growth, or development round about, therefore, the present invention proposes a kind of gradual strategy and predicts locust hatching, upgrowth situation, predicts more accurately possibility and risk that the plague of locusts occurs, avoid relying on the error of once inputting data prediction result, can bring into play the remote sensing technology energy quick obtaining every day of the advantage of habitat key element on a large scale.

Brief description of the drawings

Fig. 1 is embodiment of the present invention technical scheme process flow diagram (in figure, the definite of three late Aprils critical period, late May and late Junes is that different regions and different year may there are differences because of the coupling of habitat key element according to the Xinjiang region investigation acquisition of recent years).

Fig. 2 is that Acridoid From Xinjiang in 2011 is grown suitability exponential distribution figure (A: the suitability index of laying eggs; B: hatching suitability index; C: growth suitability index).

Fig. 3 is grassland, Xinjiang region in 2011 the plague of locusts occurrence risk figure (A: the predicting the outcome of the middle ten days and the last ten days in April that predicts the outcome; B: the last ten-days period in May revised result).

Embodiment

Below in conjunction with accompanying drawing, by embodiment, further set forth the present invention, but the scope not limiting the present invention in any way.

The present embodiment, taking the frequent Xinjiang region occurring of the grassland plague of locusts as example, adopts gradual plague of locusts Risk Forecast Method provided by the invention, and locust the condition of a disaster is carried out to risk profile.

As shown in Figure 1, according to the interaction relationship of grassland grasshopper population development and habitat key element, the present invention grows Grasshopper Population to be divided into three phases: the stage of laying eggs, hatching stage and growth phase, the basis building as locust the condition of a disaster risk forecast model.This three phases is to determine whether Grasshopper Population growth finally can form the key of disaster.Use RS and GIS systems technology to build the suitable index of laying eggs, hatch suitable index and the suitable index of growing based on this three phases, observe and predict the field inspection at station and report the pattern gathering by each department locust again, and the renewal of remote sensing monitoring data, initial forecast result (P in Fig. 1) is carried out to the modified twice (M in Fig. 1 ₁with M ₂) set up " gradual " Forecasting Methodology of locust disaster risk, make to predict the outcome more accurately and reliably.Overall technological scheme flow process of the present invention as shown in Figure 1, P, M in figure ₁, M ₂need to determine according to the particular geographic location of grassland grasshopper disaster the time division in incubation period, three length of times.In figure, the definite of three late Aprils critical period, late May and late Junes is that different regions and different year may there are differences because of the coupling of habitat key element according to the Xinjiang region investigation acquisition of recent years.Xinjiang region in this example, hatch suitable index according to the suitable exponential sum of laying eggs of this cycle, start the risk of the plague of locusts to carry out initial predicted (P) from annual late April, adopt up-to-date LST and the TVDI data (this is because Locust ovum hatching also can be subject to the impact of the condition such as temperature, moisture that starts hatching a period of time before) of before the incubation period, closing on the remotely-sensed data inverting of incubation period, again in conjunction with temperature record every day in winter in this cycle, build plague of locusts risk index, carry out initial prediction (P in Fig. 1).In the locust late May in hatching later stage, the up-to-date land surface temperature obtaining according to the remotely-sensed data inverting of incubation period and TVDI data, recalculate the suitable exponential sum plague of locusts risk index of hatching, initial predicted result is revised, improve the precision of plague of locusts risk profile, this (sees the M in Fig. 1 for incubation period correction ₁).Show according to the condition survey result in Xinjiang of China area, late June is the period finishing three length of times of grassland grasshopper, it is the critical period of locust growth and the locust of going out, according to the vegetation coverage data of the space interpolation data of the rainfall amount of the meteorological site in June in this cycle, remote-sensing inversion, recalculate the suitable exponential sum plague of locusts risk index of growth, carry out the correction again of plague of locusts risk, this is to revise (the M in Fig. 1 three length of times ₂).

Concrete implementation step is as follows:

1) build the suitable index (OSI) of laying eggs: according to upper one year the lay eggs vegetation in stage of locust cover, the situation such as the soil texture and soil moisture, to the soil types factor (Soil Type Factor, STF), Soil moisture factor (Soil Moisture Factor, and egg-laying season vegetation blanketing fctor (Vegetation Factor for OSI SMF), VFO) three factors are normalized, then adopt linear weighted model to obtain to lay eggs suitable index (Oviposition Suitability Index, OSI), in order to the inhibition or the suitable situation that represent to lay eggs to locust in habitat, characterize the locust successful possibility of laying eggs.Specifically comprise following process:

1.1) obtain the soil types factor (STF) by calculating: locust lay eggs with sandy loam optimum, sandy loam i.e. soil between sandy soil and clay.Clay content is in 40% left and right, and silt content is in 60% left and right, and optimum locust lays eggs, and utilizes the amassing of silt content index (SI) and clay content index (CI) to obtain sandy loam index, and is normalized.Computing formula is as follows:

$\mathrm{SI}=\left\{\begin{array}{cc}(S_{\max }-S)/(S_{\max }-60)& (60\&le;S)\\ (S-S_{\min })/(60-S_{\min })& (S<60)\end{array}\right.---\left(1\right)$

$\mathrm{CI}=\left\{\begin{array}{cc}(C_{\max }-C)/(C_{\max }-40)& (40\&le;C)\\ (C-C_{\min })/(40-C_{\min })& (C<40)\end{array}---\left(2\right)\right.$

STF＝SI*CI (3)

In formula, SI is silt content index, and CI is clay content index, and STF is the soil types factor; S and C represent respectively silt content data and clay content data, are dimensionless, in percentage; S _max, S _minbe respectively maximal value and the minimum value of silt content in study area; C _max, C _minbe respectively maximal value and the minimum value of clay content.

1.2) obtain Soil moisture factor (SMF) by inverting: the growth of locust egg need to absorb moisture, the too high or too low growth that all can affect locust egg of soil moisture.The temperature vegetation drought index (TVDI) being obtained by remotely-sensed data inverting is inversely proportional to soil moisture, be used for representing the variation of soil moisture herein, by the statistical study of actual measurement soil humidity data and TVDI, the inversion formula of setting up soil moisture is as follows.Wherein, when soil moisture optimum locust between 10%～20% lays eggs, therefore corresponding TVDI scope is 0.71～0.86.

$\mathrm{SMF}=\left\{\begin{array}{cc}({\mathrm{TVDI}}_{\max }-\mathrm{TVDI})/({\mathrm{TVDI}}_{\max }-0.86)& (\mathrm{TVDI}>0.86)\\ 1& (0.71\&le;\mathrm{TVDI}<0.86)\\ (\mathrm{TVDI}-{\mathrm{TVDI}}_{\min })/(0.71-{\mathrm{TVDI}}_{\min })& (\mathrm{TVDI}<0.71)\end{array}\right.---\left(4\right)$

In formula, SMF is Soil moisture factor, and TVDI is temperature vegetation drought index, is dimensionless, and value is between 0～1; TVDI _max, TVDI _minbe respectively maximum and minimum TVDI value that in survey region, remote-sensing inversion obtains.

1.3) obtain vegetation blanketing fctor (VFO) by calculating: locust generally covers lower place in vegetation and lays eggs, and therefore, vegetation coverage is more greatly more not suitable for locust and lays eggs.The FVC1 is here locust while the laying eggs vegetation coverage in (7～August).

$\mathrm{VFO}=\frac{{\mathrm{FVC}1}_{\max }-\mathrm{FVC}1}{{\mathrm{FVC}1}_{\max }-{\mathrm{FVC}1}_{\min }}---\left(5\right)$

In formula, VFO is the vegetation blanketing fctor in egg-laying season; FVC1 is the vegetation coverage data of utilizing the egg-laying season that remote sensing images inverting obtains, is dimensionless variable, and value is between 0～1; FVC1 _max, FVC1 _minrespectively vegetation coverage maximal value and minimum value on egg-laying season remote sensing images.

1.4) calculate and obtain the suitable index (OSI) of laying eggs by linear weighted function method: because the impact of laying eggs of different factor pair locusts is different, by linear weighted function and, can calculate the locust suitable index of laying eggs.

OSI＝b ₁×nSTF+b ₂×nSMF+b ₃×nVFO (6)

In formula, b ₁, b ₂, b ₃being respectively is weight number, and in this example, value is 0.3,0.5 and 0.2; N represents index to carry out the normalization of 1～10 interval.

2) build the suitable index of hatching (ISI): build the suitable index of hatching (Incubation Suitability Index by the temperature and humidity factor of Locust ovum overwintering survival rate, incubation period, ISI), characterize the impact that severe winter and incubation period habitat conditions are successfully hatched locust egg; Method is, first three indexes is normalized, and then takes linear weighted model to build ISI; Specifically comprise following process:

2.1) calculate Locust ovum overwintering survival rate: the calculating Main Basis of the Locust ovum overwintering survival rate time that cold wave arrives is for the first time estimated, concrete calculation procedure is as follows: Step1: according to every mean daily temperature on September, upper 1 to Second Year March and every day maximum temperature data, select mean daily temperature to reduce by 10 DEG C of above, day maximum temperatures and be less than 5 DEG C, and date is the earliest as time of cold wave first.Step2: computing time X value: first 5 days of early September was 1, and latter 5 days is 2, and first 5 days of the middle ten days was 3, and latter 5 days is 4, the like Step3: be calculated to be motility rate, the relation of cold wave time and survival rate is suc as formula (7) first,

K＝-0.06X ²+4.2X+7 (7)

In formula, K is survival rate, and X is time value, and computing method are shown in above-mentioned Step2.

2.2) calculate temperature factor (TF) and humidity factor (MF):

$\mathrm{TF}=\left\{\begin{array}{c}\frac{{\mathrm{LST}}_{\max }-\mathrm{LST}}{{\mathrm{LST}}_{\max }-15}(\mathrm{LST}\&GreaterEqual;15)\\ \frac{\mathrm{LST}-{\mathrm{LST}}_{\min }}{15-{\mathrm{LST}}_{\min }}(\mathrm{LST}<15)\end{array}\right.---\left(8\right)$

In formula, TF and MF are respectively the soil moisture factor and Soil moisture factor; LST is the land surface temperature being obtained by remote-sensing inversion, and unit is degree Celsius, LST _max, LST _minbe respectively the maximal and minmal value of LST.

Before incubation period then closes on (late April) obtain the last remote sensor MODIS land surface temperature (LST) data and TVDI data, be used for building the soil moisture factor (TF).Soil moisture factor (MF) directly represents by TVDI value.

2.3) calculate and obtain the suitable index of Locust ovum hatching (ISI): after temperature factor and humidity factor are normalized, press as shown in the formula

Son calculates ISI:

ISI＝K*n(TF*MF) (9)

In formula, K expression (7) result of calculation, n represents the product of variable TF and MF to carry out the normalization in 1～10 interval.

3) build the suitable index of growth (DSI): according to the geographical height above sea level factor (the Elevation Factor of estimation range, EF), locust vegetation pattern in the growth period factor (Vegetation Type Factor, and vegetation blanketing fctor (Vegetation Cover Factor2, VCF VTF) ₂), first three indexes are normalized, then take linear weighted model to build the suitable index (Development Suitability Index, DSI) of locust growth and development stage.This index quantitative description locust from ovum is hatched successfully, be subject to the inhibiting effect of habitat key element around.Specifically comprise following process:

3.1) by calculating the height above sea level factor (EF): geographical height above sea level is the key factor that affects locust growth habitat conditions, according to the region characteristic in Xinjiang, gets sea level elevation 2300m and 600m as separation.If be applied to other survey regions, the value of this separation will be adjusted definite according to the locust life habit of study area.

$\mathrm{EF}=\left\{\begin{array}{cc}\frac{{\mathrm{DEM}}_{\max }-\mathrm{DEM}}{({\mathrm{DEM}}_{\max }-H_1)}& (\mathrm{DEM}>H_1)\\ 1& (H_2\&le;\mathrm{DEM}\&le;H_1)\\ \frac{\mathrm{DEM}-{\mathrm{DEM}}_{\min }}{H_2-{\mathrm{DEM}}_{\min }}& (\mathrm{DEM}<H_2)\end{array}\right.---\left(10\right)$

Formula, EF is the height above sea level factor, and DEM is Law of DEM Data, and unit is m, DEM _max, DEM _minbe respectively survey region maximum and minimum height value, unit is m; H ₁with H ₂represent respectively the elevation upper limit and lower limit that the most applicable locust in study area grows, need to determine span according to concrete region and locust life habit.

3.2) by calculating the vegetation blanketing fctor (VFD) of locust growth phase, concrete grammar is as follows:

$\mathrm{VFD}=\left\{\begin{array}{c}\frac{{\mathrm{FVC}2}_{\max }-\mathrm{FVC}2}{{\mathrm{FVC}2}_{\max }-0.5}(\mathrm{VFD}>0.5)\\ 1(0.25\&le;\mathrm{VFD}\&le;0.5)\\ \frac{\mathrm{FVC}2-{\mathrm{FVC}2}_{\min }}{0.25-{\mathrm{FVC}2}_{\min }}(\mathrm{VFD}<0.25)\end{array}\right.---\left(11\right)$

In formula, VFD is the vegetation coverage factor of locust growth phase, and FVC2 is according to the vegetation coverage of the locust growth phase of remotely-sensed data inverting, is dimensionless, value 0～1, FVC2 _max, FVC2 _minrepresent respectively the minimum and maximum value of each remote sensing pixel vegetation coverage in locust growth phase study area.Coefficient 0.50 and 0.25 represents to be respectively applicable to the minimum and maximum threshold value of the vegetation coverage that grows of locust, and vegetation coverage is too high can restriction locust movable and obtain enough illumination, and too lowly can not provide sufficient food.

3.3) determine the assignment of each vegetation pattern according to vegetation pattern data, obtain the vegetation pattern factor (VTF) of locust growth phase; .

3.4) above-mentioned three indexes are normalized, calculate the suitable index of growth (DSI) by linear weighted model:

DSI＝d ₁×nEF+d ₂×nVFD+d ₃×nVTF (12)

In formula, d ₁, d ₂, d ₃be respectively the weighting coefficient of three factors after by 1～10 normalization, be respectively 0.2,0.5 and 0.3.

4) carry out initial plague of locusts risk profile: structure lay eggs suitable index, hatch suitable exponential sum and grow on the basis of suitable index, the method multiplying each other of taking to count builds locust disaster risk index (Locust Risk Indicator, LRI), the risk size of quantitative forecast grassland plague of locusts outburst.Utilize historical data to the empirical relationship between LRI and risk class, set up region plague of locusts risk class and divide foundation, make plague of locusts risk rating scheme.Due to predicting the outcome of this step be historical data based on upper one year as input, be therefore called " initial plague of locusts risk profile ", see shown in the P stage in Fig. 1;

5) revise and obtain plague of locusts risk profile result (M of revised beginning for the first time by the incubation period ₁): it is the basis actual remote sensing monitoring data of locust incubation period then that the incubation period is revised, mainly comprise land surface temperature, TVDI data, recalculate the suitable exponential sum plague of locusts risk index of hatching, the locust density data of combined ground investigation again, adjustment risk class is divided, and initial risks is predicted the outcome and revised;

6) by revising the net result (M that obtains plague of locusts risk profile three length of times ₂): revise three length of times is actual habitat key element (rainfall amount, the vegetation coverage) monitoring result of utilizing for three length of times, recalculate the suitable exponential sum plague of locusts risk index of growth as input data, combined ground locust density observation again, revises plague of locusts risk class.Through three the length of time revised Output rusults as the net result of this method plague of locusts risk profile, can be used as the foundation of the department's decision-making of preventing and reducing natural disasters.Growth and development stage after three length of times, can further rely on the variable density of ground observation locust, as the foundation of " closing on " early warning.

Step 5) and step 6) in, it is respectively the correction to predicting the outcome, concrete grammar is to utilize step 1)～step 4) method, replace the data of previous stage as input using actual then remote-sensing inversion and ground observation data, recalculate hatching the suitable exponential sum suitable index of growing respectively.In addition, in step 6) three length of times revise, when calculating the growth suitability index of validation period, need to add the rainfall amount factor (RF), its computing method are as follows:

$\mathrm{RF}=\left(\frac{P_{\max }-P}{P_{\max }-P_{\min }}\right)$

In formula, P is rainfall amount, and unit is mm; P _max, P _minbe respectively maximal value and the minimum value of rainfall amount in region.

Then, above-mentioned steps 3.4) in formula (12) just become:

DSI＝d ₁×nEF+d ₂×nVFD+d ₃×nVTF+d ₄×nRF

In formula, d ₁, d ₂, d ₃, d ₄the vegetation blanketing fctor (VFD), the vegetation pattern factor (VTF) that is respectively the height above sea level factor (EF), three length of times and four factors of the rainfall factor (RF) weighting coefficient after by 1～10 normalization, be respectively 0.15,0.4 and 0.2 and 0.25.

Due to fluctuation and the variation of annual each habitat key element, plague of locusts risk profile result precision is backward naturally higher, but gives the time of the department's emergency response of preventing and reducing natural disasters also short, therefore, the present invention has adopted the prediction mode of a kind of " gradual ", meets the demand of plague of locusts defence.

In the above-mentioned grassland plague of locusts Risk Forecast Method based on remote sensing technology, relate to the remote-sensing inversion of the Grasshopper Population factor of the habitat of multiple keys, periodically obtain the quantitative data of large-scale locust habitat key element by remote sensing technology.Wherein, China's terrestrial climate data earning in a day data set (medial temperature, maximum temperature, minimum temperature) relies on surface weather observation to obtain, can share website from National Meteorological Bureau downloads, administrative division, dem data obtain from local Mapping departments, and the soil texture, silt content data can be obtained from agricultural sector.In addition, in the present invention, rely on the crucial habitat key element that remote-sensing inversion obtains to be: land surface temperature, vegetation coverage, soil moisture, these three factor change in time and space are obvious, to locust development impact maximum, and are difficult to obtain by conventional means.The inversion algorithm of land surface temperature is comparatively ripe at present, and inversion accuracy is also enough, and more difficult is the remote-sensing inversion of vegetation coverage and soil moisture.The inverting extracting method of these two key factors of the vegetation coverage factor and Soil moisture factor is as follows:

A) the segmentation inversion method of the soil moisture based on remote sensing:

Vegetation cover situation difference, the remote sensor of selecting are surveyed spectral coverage difference, and the information chain between soil moisture and the sensor information of foundation is also just different.In general, the research of Remote Sensing of Soil Moisture inverse model is all that ground table status is defined as to two kinds of perfect conditions, and one is bare area, and another one is under vegetation coverage condition.Thermal inertia ratio juris is the heat interchange based on earth's surface and air, by the theoretical model that road radiation transmission process is derived and obtained, covers and can produce very large error due to the interference of vegetation information compared with high earth's surface in vegetation; And surface temperature based on energy-balance equation is in conjunction with the method for vegetation index, as: temperature vegetation drought index (Temperature Vegetation Dryness Index, TVDI), vegetation coverage is less approach bare area in the situation that also can because of Soil Background, on vegetation information, impact significantly increases, thereby make the true water cut of scholar's earth obtain wrong estimation.Therefore, the present invention proposes a kind of method of segmentation inversion soil moisture.First the method is divided into three types by earth's surface covering state: the soil under the soil under bare soil, sparse vegetation cover and airtight vegetation cover; Then using pixel as elementary cell, distinguish above-mentioned three kinds of vegetation cover types by vegetation index threshold value, thereby complicated earth surface is covered to (irrelevant with growth season) to be simplified, select again suitable Optimization inversion model, to improve topsoil remote-sensing humidity inversion accuracy and the automatization level in whole region.Specific as follows:

A1) in bare soil situation: build thermal inertia model.Soil moisture content is higher, and the thermal inertia of soil is also just larger, and soil moisture luffing is just less; Otherwise, antecedent soil moisture lack of water, Soil Thermal inertia is just little, and soil moisture luffing is just larger.Utilize thermal infrared remote sensing to survey the soil moisture and change, obtain Soil Thermal inertia, reach the object of estimation soil moisture content.Thermal infrared sensor inverting surface temperature on daytime, synthesizes it by maximal value synthetic method.For the soil moisture content inverting in large region, can adopt synthetic product of a couple of days, to eliminate to greatest extent the impact of cloud, ensure the continuity of data space simultaneously.Synthetic method is formula (13):

LST _MAX＝MAX(LST _DAY1,LST _DAY2,LST _DAY3……,LST _DAYk) (13)

Wherein MAX () gets max function, LST for wave band computing _mAXfor the synthetic rear image of maximal value that function returns, LST _dAY1, LST _dAY2, LST _dAY3, LST _dAYkbe respectively the surface temperature on daytime of every day in k days, unit is degree Celsius.K be for the synthesis of the number of days of temperature value, utilize formula (14), calculate all band albedo α (being that different sensors has difference with MODIS data instance) here,

α＝0.160ρ ₁+0.291ρ ₂+0.243ρ ₃+0.116ρ ₄+0.112ρ ₅+0.081ρ ₇-0.0015 (14)

In formula, α is all band albedo, dimensionless; ρ _ifor the reflectivity of 1st～7 wave bands of MODIS sensor, dimensionless, value 0～1.

Formula in recycling formula (15) calculates apparent thermal inertia ATI, can obtain the statistical relationship of ATI and soil moisture under the support of ground measured data, predicts the soil humidity information of unknown pixel with this.

ATI＝(1-α)/△T (15)

In formula, ATI is apparent thermal inertia, dimensionless; α is all band albedo, dimensionless; △ T is maximum temperature difference round the clock, and unit is Kelvin.

A2) under airtight vegetation coverage condition: build TVDI model.The Ts-NDVI feature space that utilizes land surface temperature and vegetation index to form, can derive index a---TVDI who represents water stress.At the feature space of simplifying, limit (Ts will wet _min) be treated to the straight line parallel with NDVI axle, do limit (Ts _max) linear with NDVI, TVDI=1 on dry limit, TVDI=0 on wet limit.TVDI is larger, and soil moisture is lower, and TVDI is less, and soil moisture is higher.Vegetation cover situation replaces with NDVI, and in the two-dimensional space of surface temperature Ts and NDVI, TVDI is expressed as formula (16):

$\mathrm{TVDI}=\frac{\mathrm{Ts}-{\mathrm{Ts}}_{\min }}{{\mathrm{Ts}}_{\max }-{\mathrm{Ts}}_{\min }}=\frac{\mathrm{Ts}-(a_2+b_2\mathrm{NDVI})}{a_1+b_1\mathrm{NDVI}-(a_2+b_2\mathrm{NDVI})}$

In formula, TVDI is temperature vegetation drought index, dimensionless, and value 0～1, the surface temperature of the given pixel of Ts-, unit is Kelvin, the normalization difference vegetation index of the given pixel of NDVI-, dimensionless; Ts _maxand Ts _minthe temperature value that is respectively He Shi limit, dry limit in Ts-NDVI feature space, unit is Kelvin.TVDI asks at last taking Ts-NDVI feature space as basis, taking the effective water cut of study area upper soll layer between wilting point and field capacity as qualifications, the TVDI minimum on wet limit, is 0, soil moisture content approaches field capacity; The TVDI maximum on dry limit, is 1, and soil moisture content approaches wilting coefficient.A ₁, a ₂, b ₁, b ₂respectively to dry limit (Ts according to ground measured data _max) and wet limit (Ts _min) correction coefficient.

A) under sparse vegetation covers: build hybrid mean value model.For sparse vegetation areal coverage soil moisture retrieval, should consider the impact of vegetation, using normalized differential vegetation index as a simple vegetation state evaluating, be introduced into regression model, in low vegetation-covered area, utilize the mean value of two kinds of model inversion results, inverting topsoil humidity value, like this, the error that can avoid simple thermal inertia or TVDI method to bring.

Three kinds of different situations of soil under soil under covering according to bare soil, sparse vegetation and airtight vegetation cover, adopt different models, carry out soil moisture retrieval.Three kinds of different land covers type design Remote Sensing of Soil Moisture inversion methods of soil under soil under covering for bare soil, sparse vegetation and airtight vegetation cover, select by the optimization to inverse model and segmentation, finally integrate and set up the segmentation inversion model that is applicable to different cover situations.Segmentation composite model is in operation, and the selection of submodel is to determine according to the statistical relationship of the NDVI value of survey region, finds out the flex point that thermal inertia model and TVDI model accuracy change with NDVI, is the diacritical point of three kinds of cover types.Like this, the operation of composite model does not need artificial intervention, only need to analyze survey region, finds out the threshold limit value of the NDVI of suitable model segmentation selection.

B) the vegetation coverage inversion method based on improved pixel two sub-models:

Pixel two sub-models are that a pixel is regarded as to the mixed pixel being made up of vegetation and soil two parts, suppose vegetation cover area ratio the vegetation coverage of this pixel be f _c, the area ratio that soil covers is 1-f _c.If the sensor information that the pure pixel being covered by vegetation obtains is S _veg, the information S that in mixed pixel, vegetation part is contributed _vcan be expressed as S _vegand f _cproduct.Describe the vegetation state of each remote sensing image picture element with normalized differential vegetation index (Normalized Difference Vegetation Index, NDVI), can obtain the inversion algorithm of vegetation coverage, suc as formula (17):

fc＝(NDVI-NDVI _soil)/(NDVI _veg-NDVI _soil) (17)

Wherein, f _cfor the pixel vegetation coverage of remote-sensing inversion, dimensionless, value is 0～1; NDVI is the NDVI value of this pixel, and dimensionless, between value-1～+ 1; NDVI _vegfor the NDVI value of pure vegetation pixel, NDVI _soilfor the NDVI value of pure soil pixel.

In actual applications, conventional NDVI maximal value and minimum value replace the vegetation index of pure vegetation and pure soil.But in remote sensing image, be difficult to guarantee that the pixel finding is pure pixel.Therefore in invention, propose to utilize the NDVI value of carrying out statistical regression and ask for pure vegetation pixel and pure soil pixel by fieldwork sample prescription data, the anti-NDVI that pushes away _vegand NDVI _soilvalue.

For each pixel on remote sensing images, there is following relation:

NDVI _veg·fc+NDVI _soil·(1-fc)＝NDVI (18)

$\left[\begin{array}{cc}{\mathrm{fc}}_1& 1-{\mathrm{fc}}_1\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ {\mathrm{fc}}_n& 1-{\mathrm{fc}}_n\end{array}\right]\left[\begin{array}{c}{\mathrm{NDVI}}_{\mathrm{veg}}\\ {\mathrm{EDVI}}_{\mathrm{soil}}\end{array}\right]=\left[\begin{array}{c}{\mathrm{NDVI}}_1\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\mathrm{NDVI}}_n\end{array}\right]---\left(19\right)$

In formula, fc _isharp for to survey by ground sample the vegetation coverage value obtaining, NDVI _ifor the remote sensing vegetation index value of this pixel, NDVI _soilwith NDVI _vegbeing the NDVI value of pure soil pixel and pure vegetation pixel, is all dimensionless.Utilize least square method, can solve its least square solution NDVI _veg, NDVI _soil.For every kind of land use pattern, calculate respectively NDVI _vegand NDVI _soilvalue.Then just set up the vegetation coverage inverse model for specific region.The method has solved a definite difficult problem for the NDVI value of pure vegetation and pure soil pixel in pixel two sub-models, take into full account the priori of specific region in research, can improve the precision of utilizing Multi-spectral Remote Sensing Data inverting region vegetation coverage, avoid thinking in existing method and determined the disadvantage of this two parameter.

In the present embodiment, taking the frequent Xinjiang region occurring of the grassland plague of locusts as example, in crucial habitat key element vegetation coverage and the remote-sensing inversion of soil moisture and the step of plague of locusts risk profile wherein, preferably, determining of some key parameters is as follows:

First, utilize on the TERRA of NASA (NASA) transmitting and AQUA satellite and carry Moderate Imaging Spectroradiomete (MODIS) sensing data, 8 days land surface temperatures (LST), vegetation coverage and the soil moisture data in 2010 6, region, inverting Xinjiang, July and 2011 3, April.Wherein the statistical model of soil moisture and TVDI is as shown in the formula y=-66.662x+69.939, x, and y is respectively TVDI and the soil moisture value of certain pixel.According to correlative study, lay eggs when soil moisture optimum locust between 10%～20% in region, Xinjiang.The inversion method of vegetation coverage has adopted formula (17)～(19) to carry out inverting estimation.The product that land surface temperature data have directly adopted US Geological Survey to utilize the thermal infrared wave band of MODIS remotely-sensed data to produce.

Then, calculate the suitability index of laying eggs, hatching suitability index and growth suitability index.Concrete steps as shown in Figure 1, are grown Grasshopper Population to be divided into three phases: the stage of laying eggs, hatching stage and growth phase, first by calculating the suitable index of laying eggs, hatch the suitable exponential sum suitable index of growing, result of calculation is as shown in Figure 2.Take Xinjiang as example, in the time that calculating formula is laid eggs suitability index, b1, the b2 in above-mentioned formula (6), the value of b3 coefficient are respectively 0.3,0.5,0.2, represent the weight difference of three factors; Coefficient H in the formula (10) of calculating growth suitability index ₁and H ₂according to field study and locust habit result of study respectively value be 2300m and 600m, the coefficient d in formula (12) ₁, d ₂, d ₃get respectively 0.2,0.5,0.3, the assignment of the vegetation pattern factor is in table 1.

Table 1 vegetation pattern factor value and corresponding vegetation pattern

Then, obtain plague of locusts risk index.Lay eggs after suitability index, hatching suitability index and growth suitability index having obtained locust, just can take the mode that each index multiplies each other to build plague of locusts risk index, divide plague of locusts risk class and can obtain predicting the outcome first of plague of locusts outburst risk then.Take Xinjiang as example, plague of locusts risk index is carried out to risk stratification according to index height: high (>200), medium higher (150～200), medium (100～150) on the low side, low (<100) four grades, see Fig. 3, A predicts the outcome that (this year April, humiture rose very fast mid or late April, a little partially early, the MODIS8 of employing days generated data is across the middle ten days and the last ten days for Locust ovum hatching); B is revised result of late May.This result shows that the risk of which region generation plague of locusts is high, and which regional risk is lower, can be the plant protection department prevention and control plague of locusts early warning support is provided.

A situation arises for table 2 Xinjiang plague of locusts reality in 2011

Finally, carry out gradual correction prediction.The sociales of grassland in Xinjiang locust mainly contain Italian locust, red shin halberd line locust, secret note dolly locust, Siberia locust etc., the normal time the earliest the incubation period appear at the first tenday period of a month in May, late May completes hatching substantially.Now can observe and predict station by each locust and start to carry out incubation period field study, investigation content mainly contains geographic position, locust kind, locust density, aerial temperature and humidity, upper soll layer humiture.Elapsed time is mainly at locust incubation period (the first tenday period of a month in May are to the last ten-days period), three length of times (mid or late June), adult stage (by the end of June) three locust developmental stages.Fig. 3 B is depicted as according to the habitat key element of in late May, 2011 remotely-sensed data inverting, the plague of locusts risk index obtaining after mode input is adjusted.Table 2 be the actual plague of locusts in 2011 a situation arises, can find out from accompanying drawing 3B, utilize the result of method provided by the invention to grassland, Xinjiang region plague of locusts forecast and the last actual plague of locusts in 2011 a situation arises very identical, prove the validity of the method.Simultaneously visible, in the result (Fig. 3 A) of in mid or late April, 2011 prediction and table 2, the statistical of actual conditions, differs greatly, and plague of locusts occurrence risk is lower generally, and especially high risk region is on the low side; But the result (Fig. 3 B) of in late May, 2011 prediction, with actual a situation arises coincide better, illustrate that the gradual Forecasting Methodology of the present invention's proposition can greatly improve model prediction precision.

Claims (12)

1. the gradual Forecasting Methodology of the grassland plague of locusts based on remote sensing technology, it is characterized in that, described method obtains the spatial and temporal distributions of the crucial habitat key element that affects grassland grasshopper population development by quantitative remote sensing inversion technique and meteorological site observation data, obtain locust disaster risk index by building plague of locusts risk forecast model, again locust disaster risk index is revised, thereby obtain the gradual prediction to grassland plague of locusts the condition of a disaster, comprise the steps:

1) a complete Grasshopper Population is divided into the stage of laying eggs, hatching stage and growth phase growth cycle, for the Grasshopper Population three phases of growth cycle, obtain by quantitative remote sensing inverting the crucial habitat key element that Grasshopper Population is grown, comprise land surface temperature, vegetation coverage and soil moisture, by calculating the suitable index that obtains respectively laying eggs, hatching the suitable exponential sum suitable index of growing;

2) by the suitable index of laying eggs, hatch grow suitable index of suitable exponential sum and calculate and obtain locust disaster risk index, as plague of locusts risk profile result; Described gradual prediction comprises that initial predicted, incubation period are revised and revise three length of times:

2.1) in the time that locust egg not yet starts to hatch, carry out initial plague of locusts risk profile, by the suitable index of laying eggs in this Grasshopper Population growth cycle, hatch the suitable index of growth in a cycle in suitable exponential sum and calculate, obtain initial locust disaster risk index, as initial plague of locusts risk profile result;

2.2) in the time that Locust ovum hatching finishes, carry out incubation period correction, obtain the suitable index of new hatching by these stylish remote sensing monitoring data, more initial plague of locusts risk index is revised, thereby obtain incubation period revised plague of locusts risk profile result;

2.3) in three length of times of locust, carry out correction in three length of times, by actual remote sensing monitoring data and the meteorological rainfall spatial interpolation data to three length of times after Locust ovum hatching, obtain the suitable index of new growth, again further to step 2.2) plague of locusts risk index that obtains revises, as finally predicting the outcome.

2. Forecasting Methodology as claimed in claim 1, it is characterized in that, step 1) the described method that obtains soil moisture by quantitative remote sensing inversion method is segmentation inversion method, first described method is divided into three types by earth's surface covering state, is respectively the soil under soil and the covering of airtight vegetation under bare soil, sparse vegetation covering; Then using pixel as elementary cell, divide to distinguish above-mentioned three kinds of vegetation cover types by the threshold value of the normalized differential vegetation index that calculated by remote sensing, thereby complicated earth surface is covered and simplified; For three kinds of vegetation cover types, carry out soil moisture retrieval and optimization by different inverse models again, be specially: for bare soil, obtain soil moisture by the inverting of thermal inertia model method; Soil under covering for airtight vegetation, obtains soil moisture by the inverting of temperature vegetation drought index; Soil under covering for sparse vegetation, obtains soil moisture by heat of mixing inertia and the inverting of temperature vegetation drought index model.

3. Forecasting Methodology as claimed in claim 1, it is characterized in that, step 1) described obtain vegetation coverage for based on improved pixel two sub-model inversion methods by quantitative remote sensing inversion method, described method is by carrying out statistical regression by the vegetation coverage data of the fieldwork sample prescription in an area and the normalized differential vegetation index calculating from remotely-sensed data, obtain the normalized differential vegetation index value of theoretic pure vegetation and pure soil pixel by least square method, then the two sub-model invertings of substitution pixel obtain the vegetation coverage of this area.

4. Forecasting Methodology as claimed in claim 1, it is characterized in that, step 1) to obtain the suitable index of laying eggs be to calculate according to lay eggs vegetation coverage, the soil texture and the soil moisture situation in stage of locust, for representing inhibition or the suitable situation that habitat key element is laid eggs to locust, characterize the locust successful possibility of laying eggs, detailed process comprises:

C1) obtain sandy loam index by calculating the amassing of soil content index and clay content index, and be normalized, obtain the soil types factor;

C2) utilize the temperature vegetation drought index of surveying soil humidity data and obtained by remotely-sensed data inverting to carry out statistical study, set up the remote-sensing inversion model of soil moisture, be used for calculating Soil moisture factor;

C3) utilize remotely-sensed data inverting to obtain the vegetation coverage in egg-laying season, obtain thus egg-laying season vegetation blanketing fctor;

C4) the soil types factor, Soil moisture factor and egg-laying season vegetation blanketing fctor are first normalized, then calculate and obtain the suitable index of laying eggs by linear weighted function method.

5. Forecasting Methodology as claimed in claim 4, it is characterized in that, step C4) described acquisition lays eggs in the linear weighted function method of suitable index, to the weight of the soil types factor, Soil moisture factor and egg-laying season vegetation blanketing fctor respectively value be 0.3,0.5 and 0.2, for representing the difference of three factor significance levels in region.

6. Forecasting Methodology as claimed in claim 1, it is characterized in that, step 1) obtain hatching suitable index be to calculate according to the soil moisture factor and the humidity factor of Locust ovum overwintering survival rate, incubation period, the impact of locust egg successfully being hatched for characterizing severe winter and incubation period habitat conditions, detailed process comprises:

F1) obtain Locust ovum overwintering survival rate according to the time that cold wave arrives for the first time by empirical statistics estimation;

F2) calculated and obtained temperature factor by the land surface temperature before the locust hatching of this cycle, the inverting of recycling remotely-sensed data obtains temperature vegetation drought index, calculates the humidity factor before the incubation period;

F3) temperature factor to Locust ovum overwintering survival rate, incubation period and humidity factor are first normalized, and are then calculated and are obtained the suitable index of hatching by product method.

7. Forecasting Methodology as claimed in claim 1, it is characterized in that, step 1) obtain growth suitable index be to calculate according to the geographical height above sea level factor of estimation range, locust vegetation pattern in the growth period factor and vegetation blanketing fctor, from ovum is hatched successfully, be subject to the inhibiting effect of habitat key element around for quantitative description locust, detailed process comprises:

S1) according to the height above sea level scope of geographical sea level elevation and the suitable life of locust, calculate the height above sea level factor by elevation segmentation;

S2) utilize remotely-sensed data inverting to obtain the vegetation coverage of locust growth phase, according to the vegetation coverage scope that is applicable to locust and grows, calculate the vegetation blanketing fctor of locust growth phase;

S3) obtain the vegetation pattern factor of locust growth phase according to vegetation pattern data;

S4) the vegetation blanketing fctor to the height above sea level factor, locust growth phase and the vegetation pattern factor are normalized, then calculate the suitable index of growth by linear weighted function method.

8. Forecasting Methodology as claimed in claim 7, it is characterized in that, step S4) the linear weighted function method that obtains the suitable index of growth be vegetation blanketing fctor to the height above sea level factor, locust growth phase and the vegetation pattern factor weight respectively value be 0.2,0.5 and 0.3, for representing the difference of three factor significance levels in region.

9. Forecasting Methodology as claimed in claim 1, it is characterized in that, step 2.3) described three the length of times revise obtain the new suitable index of growth be to obtain according to the estimation range geographical height above sea level factor in three length of times, locust vegetation pattern in the growth period factor, vegetation blanketing fctor and the rainfall factor, be subject to the inhibiting effect of habitat key element around growth period for describing, detailed process comprises:

X1) according to the absolute elevation scope of geographical sea level elevation and the suitable life of locust, calculate the height above sea level factor by elevation segmentation;

X2) utilize remotely-sensed data inverting to obtain the vegetation coverage of locust growth phase, the vegetation coverage scope of growing according to applicable locust, calculates the locust vegetation blanketing fctor in three length of times;

X3) obtain the vegetation pattern factor of locust growth phase according to vegetation pattern data;

X4) calculate the locust rainfall factor in three length of times according to the rainfall amount space interpolation data of meteorological site observation;

X5) the vegetation blanketing fctor to the height above sea level factor, locust growth phase, the vegetation pattern factor and the rainfall factor are normalized, then calculate the suitable index of growth by linear weighted function method.

10. Forecasting Methodology as claimed in claim 9, it is characterized in that step X5) obtain growth suitable index linear weighted function method be vegetation blanketing fctor, the vegetation pattern factor and the rainfall factor to the height above sea level factor, locust growth phase weight respectively value be 0.15,0.4,0.2 and 0.25.

11. as arbitrary in claim 7 or 9 as described in the gradual Forecasting Methodology of the grassland plague of locusts based on remote sensing technology, it is characterized in that, the described height above sea level factor is that the elevation upper limit and the lower limit of growing according to the applicable locust in the maximum elevation value in a region, minimum height value and described region calculate acquisition, and the elevation upper limit and lower limit needs that described applicable locust grows are studied acquisition by field study.

12. Forecasting Methodologies as claimed in claim 1, is characterized in that step 2.1) time of carrying out initial plague of locusts risk profile is the annual last ten-days period in April; Step 2.2) time of carrying out incubation period correction is the last ten-days period in May; Step 2.3) time of carrying out correction in three length of times is the last ten-days period in June.