CN109711102A - A method for rapid assessment of crop disaster losses - Google Patents

Abstract

The invention provides a rapid assessment method for crop disaster loss. The method is based on the crop model of the DSSAT system and the GLUE parameter estimation tool, utilizes the historical data of the crops in the disaster area, and obtains the leaf area index of the crops under various scenarios based on the crop model simulation. and yield, and then construct the regression equation. Then, based on the satellite image of Google Earth Engine (GEE), the leaf area index of each pixel on the satellite image was brought into the linear regression equation to obtain the yield per pixel. Finally, the relative yield loss is obtained by comparing the yield of the disaster year in the affected area with the yield of the previous year. The method of the invention does not require a large amount of ground observation data, but only needs a small amount of experimental data for calibration and calibration, which improves the evaluation efficiency, reduces the time cost of disaster loss evaluation, and provides a guarantee for disaster prevention and mitigation. The method of the invention realizes the quantitative assessment of disaster losses, can reliably estimate the loss and yield rate of different spatial scales, and improves the quality of disaster damage estimation.

Description

A method for rapid assessment of crop disaster losses

technical field

The invention relates to the technical field of agricultural remote sensing, in particular to a cross-scale, crop disaster loss quantitative assessment method based on a remote sensing data processing platform Google EarthEngine (GEE) and a crop model, in particular to a crop disaster loss rapid assessment method.

Background technique

Agriculture and weather are closely related. The whole process of crop growth is in the natural environment, and the formation of its yield is extremely vulnerable to the stress and interference of unfavorable factors such as meteorological disasters. At present, the assessment of agrometeorological disasters is mainly divided into two categories: risk assessment and impact assessment. Risk means the possibility of disaster occurrence and damage. The technical basis of assessment is risk analysis technology. The obtained risk intensity division cannot accurately quantify disaster losses, but can only describe disasters roughly. The previous impact assessment mainly includes the following three methods: 1) On the basis of the multi-year production data of the ground station, the linear moving average method and the orthogonal polynomial approximation method are used to obtain the trend yield, and the meteorological yield is obtained by decomposing the actual yield of the year minus the trend yield. , calculate relative meteorological yields to estimate disaster losses. This method is based on the assumption that all crop yield reductions are caused by chilling damage, and it is not effective in assessing losses from a single disaster and cannot be dynamically assessed. 2) Based on historical observation data, the regression equation of disaster indicators and yields is constructed, and the disaster losses are extrapolated. However, the disaster indicators are very regional, and it is difficult to extend and apply to other regions and cannot meet the loss assessment below the county level. 3) The crop model has been applied to yield forecasting, farmland management decision support, disaster index construction and loss quantification due to its advantages of being oriented to the crop growth process and having a strong mechanism. A continuum from maturity to maturity, reflecting how crop growth responds to different environmental and management factors. However, most crop models are for specific field experiments and can only perform single-point simulation. The regional scale application of the model can be achieved through regionalization of variety parameters and spatial interpolation of meteorological elements, but new errors are inevitably introduced. Although regional crop models can represent spatial differences, a large amount of driving data is required for construction and operation, and due to the large spatial heterogeneity of surface parameters, varieties and management methods, it is very difficult to determine its parameters, and large-scale research is still not easy. accomplish. Furthermore, the spatial resolution of regional models depends on meteorological or soil data, making fine-grained loss mapping difficult. Therefore, actively exploring reasonable and practical methods for more accurate and faster regional loss assessment can provide ideas for the operational operation of agricultural disaster early warning and agricultural insurance.

Remote sensing technology can monitor the growth and development of crops in a large-scale, dynamic and real-time manner, and is widely used in various fields such as planting area, disease and insect pest monitoring, disaster monitoring and precision agriculture. Google Earth Engine (GEE) is a platform provided by Google for online visualization, calculation, analysis and processing of a large amount of global-scale earth science data (especially satellite data), providing global Sentinel, MODIS, Landsat TM/OLI and other multi-source, multi-scale remote sensing data , is a massive remote sensing data processing, archiving and analysis platform that supports parallel cloud computing, which solves the difficulties of traditional remote sensing image collection, large storage capacity and low processing efficiency. The combination of remote sensing and crop models can achieve continuous yield simulation at different spatial resolutions and solve the problem of regional scale application of site crop models, and has been widely used, but there are the following two key problems: 1) Although the statistical model established using remote sensing indices Simple and easy to use, but statistical models based on single-phase vegetation index and measured yield are limited to a specific time, place and year. 2) Although some achievements have been made in the estimation of the damage of remote sensing data based on crop model assimilation, the model is not easy to generalize due to its large amount of input data and complex operation process, and the data collection is very complicated and the operation efficiency is low.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide a rapid assessment method for crop disaster losses, so as to reduce or avoid the aforementioned problems.

In order to solve the above-mentioned technical problems, the present invention proposes a method for rapid assessment of crop disaster losses, comprising the following steps:

Step S1: Input the soil data, meteorological data and agricultural production data of crops in the area to be assessed into the DSSAT system to generate respectively the soil file S, the weather file W and the crop file A that can be called; through the GLUE parameter estimation tool, Call and calculate the previously generated files S, W and A, and calculate and obtain the calibration data file C containing the calibration values of the crop variety parameters in the area; call the aforementioned soil file S, meteorological file W, crop file A through the crop model of the DSSAT system And the calibration data file C obtained carries out simulation calculation, obtains the LAI file corresponding to the daily LAI and the output file Y corresponding to the output;

Step S2: Set the annual accumulated temperature anomaly in the area as the disaster indicator H;

Step S3: Taking the peak growth period of the crop as the center, two time windows are selected for each 30 days before and after, and the daily LAI corresponding to the two windows is obtained from the LAI file; then the yield yield of each scenario category is obtained from the yield file Y as the cause. Variables, LAI _d ¹ of a day in the previous window, LAI _d ² of a day in the latter window, and disaster indicator H are independent variables, and the following linear regression equation is established, and then the coefficients of the linear regression equation and the corresponding front and rear windows are calculated. The date of the LAI is saved:

Yield=β _0,d +β _1,d *LAI _d ¹ +β _2,d *LAI _d ² +β ₃ *H

In the formula: Yield is the yield under each scenario category, LAI _d ¹ and LAI _d ² are the LAI on the d day in the front window and rear window under each scenario category, respectively, and H is the disaster indicator;

Step S4: Based on the satellite image of the area obtained by the GEE platform, extract the maximum dynamic vegetation index WDRVI of each pixel in the two time windows and the corresponding date; and convert the WDRVI into LAI;

Step S5: According to the dates corresponding to the LAI of the front and rear windows, the coefficients of the linear regression equation obtained in step S3, and then calculate the output pixel by pixel based on the LAI of the two windows obtained from the GEE satellite image and the disaster indicator H obtained from the actual meteorological data, Finally, the change in yield relative to the previous year is calculated to obtain the relative yield loss of this disaster.

Preferably, the formula for calculating the relative yield loss in the step S5 is:

In the formula: _LR is the relative yield loss, Y _l is the yield of the previous year (year without disaster), and Y _n is the yield of the year in which the disaster occurred; so far, the rapid assessment process of crop disaster loss is completed.

Preferably, the step S1 further includes the following steps:

Set the scenario categories of multiple disaster scenarios and management scenarios, call and calculate the meteorological file W, crop file A, soil file S and calibration data file C corresponding to each scenario category through the crop model in DSSAT, and obtain the same as the above. Multiple LAI files and yield files Y with the same number of scenario categories.

Preferably, the specific calculation formula for converting WDRVI into LAI in the step S4 is:

WDRVI=-0.681+1.437(1-e- ^0.351LAI )

where WDRVI is the maximum dynamic vegetation index, LAI is the leaf area index, ρ _NIR is the near-infrared band reflectance measured by the sensor load, and ρ ^RED is the red light band reflectance.

Preferably, the step S5 further includes the following steps: after obtaining the dates corresponding to the front and rear windows LAI through the GEE satellite image, calculating the regression coefficient of the corresponding combination date according to the dates corresponding to the front and rear windows LAI, calculating the output pixel by pixel, and finally obtaining Yield mapping in disaster years.

Preferably, in the step S5, in the production calculation process of the previous year, the chilling injury scenario is excluded from the simulation scenarios, and only different management scenarios are simulated to obtain a regression coefficient matrix, and a production map of a disaster-free year is obtained.

The rapid assessment method for crop disaster loss of the present invention does not require a large amount of ground observation data, but only needs a small amount of experimental data for calibration and calibration. In addition, processing a large amount of remote sensing data based on GEE also greatly saves the cost of data processing, improves the evaluation efficiency, reduces the time cost of disaster loss assessment, and provides a guarantee for disaster prevention and mitigation. In addition, the method of the invention realizes quantitative assessment of disaster losses, can reliably estimate yield reduction rates at different spatial scales, improves the quality of disaster damage assessment, and can provide a quantitative basis for agricultural insurance.

Description of drawings

The following drawings are only intended to illustrate and explain the present invention schematically, and do not limit the scope of the present invention. in,

1 shows a schematic flowchart of a method for rapid assessment of crop disaster losses according to a specific embodiment of the present invention;

Figure 2 shows a schematic diagram of historical low temperature chilling damage loss according to another specific embodiment of the present invention;

FIG. 3 shows a schematic diagram of regional cooling damage loss according to another specific embodiment of the present invention.

Detailed ways

In order to have a clearer understanding of the technical features, objects and effects of the present invention, the specific embodiments of the present invention will now be described with reference to the accompanying drawings. Wherein, the same parts use the same reference numerals.

Based on the aforementioned problems in the prior art, the present invention combines remote sensing observation data, meteorological data and crop models to create a rapid assessment method for disaster losses that extends from point to area. This method is not limited by the ground measured data, can be dynamically evaluated, is easy to operate, and has strong generalization ability. It can not only conduct large-scale research, but also quantify the loss of the county level and even the field, aiming to realize the fine mapping of crop yield loss. , in order to provide reference for the operational operation of disaster assessment and provide guarantee for disaster prevention and mitigation.

Specifically, the present invention provides a method for rapid assessment of crop disaster losses. The method is based on the crop model of the DSSAT system and the GLUE parameter estimation tool, and uses soil, meteorological and crop production data in the disaster area to simulate and obtain the leaf area of the crop. index and yield, and then construct the regression equation. Then, based on the satellite image of Google Earth Engine (GEE), the leaf area index of each pixel on the satellite image was brought into the linear regression equation to obtain the yield per pixel. Finally, the relative yield loss is obtained by comparing the yield of the area to be assessed with the yield of the previous year.

Among them, DSSAT (Decision Support System for Agro technology Transfer) is one of the most widely used model systems. DSSAT is a comprehensive computer system developed and developed by the University of Hawaii authorized by the United States Agency for International Development under the sponsorship and guidance of IBSNAT (International Benchmark Sites Network for Agro technology Transfer) International Benchmark Network for Agricultural Technology Transfer. DSSAT is not a general model, it has developed different models for different crops. DSSAT currently consists of 26 different crop simulation models, including CERES (Crop Environment REsource Synthesis) series models, CROPGRO bean crop model, SUBSTORpotato potato model, CROPSIMcascava cassava model, OILCROP sunflower model and other crop models. The GLUE parameter estimation tool is a functional module of DSSAT. Google EarthEngine (GEE) is a tool that can batch process satellite image data and belongs to a series of tools of Google Earth. Compared with traditional image processing tools such as ENVI, GEE can process massive images quickly and in batches, and can quickly calculate vegetation indices such as NDVI, predict crop-related yields, and monitor changes in drought conditions.

The above are all existing well-known technologies, which, as the calculation tool used in the research and development of the present invention, are not the content claimed in the present invention. Those skilled in the art should understand that the above-mentioned computer software systems in the prior art may have different software versions and improvements. Based on the technical concept disclosed in the present invention, those skilled in the art can choose any existing computer software system to re-engineer. Now, as long as it conforms to the concept of the method of the present invention.

The method for rapid assessment of crop disaster losses of the present invention is further described in detail below by way of specific examples, as shown in Figure 1, which shows a schematic flowchart of a method for rapid assessment of crop disaster losses according to a specific embodiment of the present invention, As shown in the figure, the method for rapid assessment of crop disaster loss of the present invention comprises the following steps:

Step S1: Input the soil data, meteorological data and crop agricultural production data of the area to be assessed into the DSSAT system, and generate respectively a soil file S, a weather file W and a crop file A that can be called.

For example, the maize in Oroqen Autonomous Banner of Inner Mongolia can be used as the research object, and the soil, meteorology and agricultural production (planting date, planting density, irrigation, nitrogen application) data of maize in this area between 2013 and 2017 can be input To the DSSAT system, callable files S, W and A are generated respectively. Among them, the meteorological data, soil data and agricultural production data of corn in the Oroqen Autonomous Banner of Inner Mongolia can be collected from the yearbook of the local agricultural department and the observation data of the meteorological department. These data are all public data and only need to be extracted.

Then, through the GLUE parameter estimation tool, call and calculate the previously generated files S, W and A, and obtain the calibration data file C containing the calibration values of the parameters of the crop varieties in this area.

The GLUE parameter estimation tool is a functional module built into the DSSAT system. This module can read various data files of the DSSAT system, and iterate repeatedly through the "trial and error method" to obtain specific characteristic parameters related to the crop. calibration values, and can automatically generate a file containing these calibration values.

For example, taking the corn of Oroqen Autonomous Banner in Inner Mongolia as an example, files S, W and A were generated by inputting the historical data of soil, meteorology and agricultural production from 2013 to 2017. Then call these files through GLUE. After the operation, a calibration data file C containing calibration values of many characteristic parameters can be generated. The calibration values in the calibration data file C are unchanged for the corn of the Oroqen Autonomous Banner in Inner Mongolia. Therefore, this Calibration data file C can also be recalled in subsequent steps and used for other research work on maize in the region.

For example, in a specific embodiment of the present invention, the following specific characteristic parameters are more relevant to the research on corn:

It should be understood that the calibration values of the characteristic parameters obtained by the GLUE calculation are obtained by iterative calculation of the data files S, W and A generated from the historical data of the crops in the area, and can be identified as the local characteristics of the crops in this area. Calibration data, these local calibration data usually do not change for the same crop in the same area, so in the subsequent steps of the present invention, the calibration values of these specific characteristic parameters can always be used to simulate the LAI (Leafarea) of the crop. index, leaf area index) and yield. For example, the calibration values of the six characteristic parameters in the above list represent the flowering date, harvest date, number of leaves, etc. of the corn in Oroqen Autonomous Banner, Inner Mongolia. For corn in different regions, these characteristic parameters are different. Of course, the characteristic parameters required by different crops are also different. For example, the characteristic parameters of potatoes do not need to be G2 or G3, because the characteristic parameters have nothing to do with potatoes.

In another specific embodiment of the present invention, for example, the historical data of maize in Oroqen Autonomous Banner, Inner Mongolia from 2013 to 2017 can be split, and the data from 2013 to 2015 can be used to calculate the above six characteristic parameters through GLUE. For the calibration value, the standard Root Mean Square Error (NRMSE) was calculated with the data from 2016-2017 to carry out a reliability check on the calibration values of the above six calculated characteristic parameters to control the precision and accuracy. Further, for example, after obtaining the calibration values of the six specific corn-related characteristic parameters P1, P2, P5, G2, G3 and PHINT in the aforementioned list through GLUE, GLUE will automatically record the calibration values of these six characteristic parameters. In the generated calibration data file C, that is, the values of the six characteristic parameters of each piece of data in the calibration data file C are the calibration values after iterative calculation, so the calibration data file C obtained through transformation forms the Inner Mongolia Oroqen A localized data file of the corn of the Autonomous Banner, in the subsequent steps, it can be determined that the values of the six characteristic parameters of the corn of the Oroqen Autonomous Banner of Inner Mongolia are all equal to the calibration value after the iterative calculation.

After that, the aforementioned soil file S, meteorological file W, crop file A and the obtained calibration data file C are called through the crop model of the DSSAT system for simulation calculation, and the LAI file corresponding to the daily LAI (Leaf area index, leaf area index) is obtained. And the yield file Y corresponding to the yield.

Among them, the LAI file records the daily LAI of the crop within one year, that is, the LAI data and the corresponding date of the crop that records no more than 365 days in the LAI file. Of course, since the growth date of the crop generally does not exceed one year, so , usually there are less than 365 valid LAI data. In addition, the yield file Y corresponding to the yield records the yield of the crop corresponding to the date, but because the crop has a long growing period, the yield is often 0, and the yield on different dates is also different. For crops, the yield on the last harvest date The maximum output value is the actual output value that needs to be used in the output file Y. That is to say, after calling and calculating the crop model of the DSSAT system, two files will be automatically generated, one is the LAI file and the other is the yield file Y, in which the LAI file records the daily LAI value corresponding to the date, and the yield A maximum yield value is recorded in the file Y, and this maximum yield value is the yield yield to be used subsequently in the present invention.

For example, taking the corn in Oroqen Autonomous Banner, Inner Mongolia as the research object, the corn files S, W and A and the calibration data file C in this area are called through the corn-related CERES-Maize model in the DSSAT system, and the CERES-Maize model is run. After the Maize model, a LAI file and a yield file Y can be generated.

Among them, the CERES-Maize model is a model related to maize crops in the DSSAT system. Of course, if the research is related to bean crops, the DSSAT-CROPGRO model can be used, and so on, those skilled in the art can extend to other existing DSSAT system crop models. It should be noted that the CERES-Maize model is a relatively comprehensive model system related to corn, which requires input of three types of data items, including meteorological data, soil data and agricultural production data, each data item includes multiple small items. , for example, meteorological data includes: daily maximum temperature, minimum temperature, rainfall and radiation; soil data includes: soil type, profile characteristics, soil physical and chemical properties, that is, soil name, soil color, soil water retention performance and soil texture of each layer, organic carbon, Total nitrogen, PH, cation exchange capacity; agricultural production data include: sowing date, flowering date, maturity date, unit yield, and management data such as irrigation and fertilization.

Of course, due to the historical data in the regional yearbook, there is only one production figure per year. For example, for the Oroqen Autonomous Banner of Inner Mongolia, it may only record the total annual corn production of the entire county, and it will not be broken down to each village. Therefore, in the Y file obtained after the crop model calculation, there is only one available yield Yield, and corresponding to this yield Yield, there is also only one LAI file, which also records the LAI of no more than 365 crops (considering the growth cycle of the crops. , the amount of data available in the LAI file will be less), and the number of samples is relatively small, which is not conducive to the subsequent component regression equation. Therefore, in order to improve the calculation accuracy, the number of samples needs to be expanded.

Therefore, in yet another specific embodiment of the present invention, a solution for increasing the number of samples is further provided. That is, according to the actual local cultivation and management of crops, scenario categories of various disaster scenarios and management scenarios can be set. In one specific embodiment, for example, seven disaster scenario categories and sixty-four management scenario categories may be set for corn. For example, the seven types of disaster scenarios can be respectively: three types of delayed chilling damage of maize under the condition that the temperature during the maize growth period decreases by 1, 2, and 3°C, respectively; Four types of maize obstacle-type chilling injury were randomly set under the lower limit of growth temperature in the four growth stages of maturation. The sixty-four categories of management scenarios can be, eight planting dates and eight planting densities, resulting in 8*8=64 categories of management scenarios. Combining seven disaster scenario categories and sixty-four management scenarios, a total of 7*8*8=448 scenario categories can be obtained. Of course, according to the actual local cultivation and management situation, more scene categories can be set, for example, different categories can be set according to varieties, irrigation and fertilization.

In another specific embodiment, the obtained 448 scenarios are divided, and each scenario class corresponds to a different weather file W and a crop file A corresponding to the scenario class, so 448 weather files W, Crop file A, soil file S (same as the original, no change) and calibration data file C (this is common, no change), after the 448 kinds of combined files are calculated by calling CERES-Maize, for each scenario A yield file Y related to yield and a LAI file recording the daily LAI of the corn growth period under this scenario can be simulated. Finally, 448 LAI files and corresponding 448 yields can be obtained every year through the 448 scenarios. File Y, relative to the fact that there is only one yield per year (1 LAI file and 1 yield file Y), this undoubtedly greatly expands the sample size.

Step S2: Set the local multi-year accumulated temperature anomaly as the disaster index H to describe the impact of disasters on crops.

For example: for low temperature and chilling damage, the multi-year accumulated temperature anomalies from May to September at each simulated site can be used as the disaster indicator H, and the calculation formula is:

In the formula, H is the daily average temperature ≥10℃ accumulated temperature anomaly in the study period (℃ d), T _i is the daily average temperature ≥10℃ on the i-th day (℃), n is the number of days in the calculation period, It is the annual average of active accumulated temperature with daily average temperature ≥10℃.

Or, in another specific embodiment, for example, for corn, the 30-year anomaly value of the accumulated temperature from May to September for corn can be used as the disaster indicator H to describe the effect of low temperature and chilling damage on corn growth. influences.

Step S3: Take the peak growth period of the crop as the center, select two time windows for each 30 days before and after, and extract the daily LAI corresponding to the two windows from the LAI file; then take the yield of each scenario category in the yield file Y as The dependent variable, the LAI of one day in the previous window (LAI _d ¹ ), the LAI of one day in the latter window (LAI _d ² ), and the disaster indicator H are independent variables, and the following linear regression equation is established, and then the obtained linear regression equation is The coefficient of and the date of the corresponding front and rear window LAI are saved:

Yield=β _0,d +β _1,d *LAI _d ¹ +β _2,d *LAI _d ² +β ₃ *H

In the formula: Yield is the yield under each scenario category, LAI _d ¹ and LAI _d ² are the LAI on the d day in the front window and rear window under each scenario category, respectively, and H is the disaster indicator.

For each LAI file and yield file Y, by combining the LAIs of the two windows, a total of 30*30=900 regression relationships can be obtained for the first 30 days and the last 30 days. For 448 LAI files and yield file Y , a total of 448 sets of regression relationships of the same number as mentioned above can be obtained. Afterwards, an algorithm such as the least squares method can be used to solve the calculation, and the obtained coefficients of the regression equation and the dates of the corresponding front and rear windows LAI are saved.

For example, in a specific embodiment, two time windows can be selected 30 days before and after the silking date of corn as the center to obtain the daily LAI of the two windows; then the output of the 448 scenario categories output by the crop model is dependent variable. Specifically, for each LAI file and yield file Y, the period from June 18 to July 17 is set as the front window, and the period from July 18 to August 17 is set as the back window. If you choose a day in the front window, then the back window can have 30 days to choose arbitrarily, and so on, forming 30*30=900 choices. Since 448 scenario categories are set, for 448 LAI files and yield files Y, each LAI file and yield file Y can have 900 choices, which combine to form a very large sample for regression calculation. , the coefficients of the equation obtained by least squares regression will also be more accurate.

Step S4: Based on the satellite image of the area obtained by GEE, obtain the maximum dynamic vegetation index WDRVI of each pixel in the two time windows and the corresponding date DOY (Day Of Year, the day of the year) image. ; then convert the maximum dynamic vegetation index (WDRVI) to LAI.

For example, in a specific embodiment, all Landsat-8s from June 18 to July 17 (front window) and July 18 to August 17 (back window) of each year can be obtained from the GEE platform and Sentinel-2 image, de-cloud it, and then extract the maximum dynamic vegetation index (WDRVI) and its date of the two time windows before and after each pixel each year and convert WDRVI to LAI, and convert WDRVI to LAI. The calculation formula is:

WDRVI=-0.681+1.437(1-e- ^0.351LAI )

where WDRVI is the maximum dynamic vegetation index, LAI is the leaf area index, ρ _NIR is the near-infrared band reflectance measured by the sensor load, and ρ _RED is the red light band reflectance.

Among them, Landsat-8 is a remote sensing image obtained by the 8th satellite of a series of earth observations used by the United States to detect earth resources and the environment. The spatial resolution of bands 1-7 and 9-11 is 30m, and band 8 is 15m resolution Panchromatic band, revisit period 16 days, spatial resolution 30m, temporal resolution. Sentinel-2 is the second satellite of the "Global Environment and Safety Monitoring" program, launched on June 23, 2015, with a spatial resolution of 10m and a revisit period of 10 days. In addition, it is not complicated to extract the maximum dynamic vegetation index WDRVI and its date through GEE's satellite images. The GEE platform provides a programming interface, which can be directly obtained from the GEE platform by entering a code.

The following is the sample code for extracting WDRVI through the GEE platform:

Step S5: According to the dates corresponding to the LAI of the front and rear windows, the coefficients of the linear regression equation obtained in step S3, and then calculate the output pixel by pixel based on the LAI of the two windows obtained from the GEE satellite image and the disaster indicator H obtained from the actual meteorological data, Finally, calculate the yield change relative to the previous year to get the relative yield loss of this disaster:

In the formula: _LR is the relative yield loss, Y _l is the yield of the previous year (year without disaster), and Y _n is the yield of the year in which the disaster occurred; so far, the rapid assessment process of crop disaster loss is completed.

In particular, the coefficients of the linear equation obtained by regression will also change slightly due to the difference in historical data at different times. Therefore, in a specific embodiment, after the dates corresponding to the front and rear windows LAI can be obtained from the GEE satellite images, the regression coefficient of the corresponding combined date can be calculated according to the dates corresponding to the front and rear windows LAI, the output can be calculated pixel by pixel, and finally the disaster year can be obtained. production map. Similarly, in the production calculation process of the previous year (disaster-free year), the chilling injury scenario can also be excluded from the simulation scenarios, and only different management scenarios can be simulated to obtain the regression coefficient matrix, and the production map of the disaster-free year can be obtained, which can then be quantified. Relative yield loss calculations for chilling damage.

In another specific embodiment of the present invention, the yield loss in years with severe low temperature and chilling damage in the Oroqen Autonomous Banner, Inner Mongolia since 1980 was evaluated, and it was integrated to the county level and compared with statistical data, as shown in Figure 2 It shows a schematic diagram of the historical low-temperature chilling damage loss chart according to another specific embodiment of the present invention. It can be seen from the figure that the measured losses are all within one time variance of the estimated loss, indicating that this method can reduce the damage caused by low-temperature chilling damage. production reductions can be reliably quantified.

In addition, in another specific embodiment of the present invention, as shown in the schematic diagram of regional cooling damage loss as shown in Figure 3, the present invention is based on the Sentinel-2 image at a spatial resolution of 10m for the barrier-type low-temperature chilling damage that occurred on September 8, 2018. The evaluation carried out has obtained the yield loss of 43 villages in Oroqen, the results are very close to the actual statistical results, and can be refined to a smaller regional level.

To sum up, the method for rapid assessment of crop disaster loss of the present invention can perform cross-scale crop disaster loss assessment based on Google Earth Engine (GEE) and a crop model, and can improve the accuracy and timeliness of regional crop damage assessment.

Compared with the traditional damage estimation method, the rapid assessment method of the present invention does not require a large amount of ground observation data, but only needs a small amount of experimental data for calibration and calibration; and the present invention can strip the influence of a single disaster species, and can be used for large-scale or county or even fields. The loss between blocks can be reliably estimated quantitatively; the method of the present invention has few limitations, strong spatial generalization ability, high universality, and can be transplanted to other regions, crops and disasters. Remote sensing data reduces computing costs and greatly improves loss assessment efficiency; in addition, the method of the present invention can be dynamically assessed, has strong timeliness and is easy to operate, is not limited by ground measured data, and is conducive to business popularization and application.

Those skilled in the art should understand that although the present invention is described in terms of multiple embodiments, not each embodiment only includes an independent technical solution. This description in the description is only for the sake of clarity, and those skilled in the art should understand the description as a whole, and regard the technical solutions involved in each embodiment as being able to be combined into different embodiments to understand the present invention scope of protection.

The above descriptions are only exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention. Any equivalent changes, modifications and combinations made by any person skilled in the art without departing from the concept and principles of the present invention shall fall within the protection scope of the present invention.

Claims (6)

1. A method for rapidly evaluating crop disaster damage comprises the following steps:

step S1: inputting soil data, meteorological data and agricultural production data of crops of a region to be evaluated into a DSSAT system, and respectively generating a soil file S, a meteorological file W and a crop file A which can be called; calling and calculating the previously generated files S, W and A through a GLUE parameter estimation tool, and calculating to obtain a calibration data file C containing calibration values of the regional crop variety parameters; calling the soil file S, the meteorological file W, the crop file A and the obtained calibration data file C through a crop model of a DSSAT system to perform simulation calculation to obtain an LAI file corresponding to the day-by-day LAI and a yield file Y corresponding to the yield;

step S2: setting a multi-year accumulated temperature range flat value of the area as a disaster index H;

step S3: selecting two time windows in each 3 days before and after the peak growth period of the crop as a center, and acquiring daily LAI corresponding to the two windows from an LAI file; then obtaining the Yield of each scene type from the Yield file Y as a dependent variable, and obtaining the LAI of one day in the previous window_d ¹LAI of one day in the rear window_d ²And establishing the following linear regression equation by taking the disaster index H as an independent variable, and then saving the obtained coefficient of the linear regression equation and the corresponding dates of the front window LAI and the rear window LAI:

Yield＝β_0,d+β_1,d*LAI_d ¹+β_2,d*LAI_d ²+β₃*H

in the formula: yield under each scene category, LAI_d ¹And LAI_d ²Respectively representing the LAI of the day d in the front window and the back window under each scene category, wherein H is a disaster index;

step S4: respectively acquiring the maximum dynamic vegetation index WDRVI of each pixel in two time windows and an image of a corresponding date based on the satellite image of the region acquired by the GEE; and converting WDRVI into LAI;

step S5: according to the dates corresponding to the LAI of the front window and the rear window, the coefficient of the linear regression equation obtained in the step S3, the output is calculated pixel by pixel based on the LAI of the two windows obtained by the GEE satellite image and the disaster index H obtained by actual meteorological data, and finally the output change relative to the previous year is calculated to obtain the relative output loss of the disaster.

2. The method of claim 1, wherein the formula for calculating the relative yield loss in step S5 is:

in the formula: l is_RFor relative yield loss, Y_lThe yield of the last year (disaster-free year), Y_nYield for the year in which the disaster occurred; so far, the crop disaster damage rapid assessment process is completed.

3. The method according to claim 1, wherein the step S1 further comprises the steps of:

setting a plurality of disaster situations and situation types of management situations, calling and calculating a meteorological file W, a crop file A, a soil file S and a calibration data file C corresponding to each situation type through a crop model in DSSAT, and obtaining a plurality of LAI files and yield files Y with the same number as the situation types.

4. The method of claim 1, wherein the specific calculation formula for converting WDRVI into LAI in step S4 is:

WDRVI＝-0.681+1.437(1-e^-0.351LAI)

in the formula: WDRVI is the maximum dynamic vegetation index, LAI is the leaf area index, p_NIRNear infrared band reflectivity, p, measured for sensor load_REDThe reflectivity is in the red light wave band.

5. The method according to claim 1, wherein the step S5 further comprises the steps of: after the dates corresponding to the LAI of the front window and the rear window are obtained through the GEE satellite images, the regression coefficient corresponding to the combined date is calculated according to the dates corresponding to the LAI of the front window and the rear window, the yield is calculated pixel by pixel, and finally the yield chart of the disaster year is obtained.

6. The method as claimed in claim 5, wherein in step S5, in the production calculation process of the previous year, the cold damage scenario is eliminated from the simulation scenarios, and only different management scenarios are simulated to obtain the regression coefficient matrix, so as to obtain the production chart of the disaster-free year.