CN106600434B

CN106600434B - Crop growth remote sensing monitoring method based on crop model and assimilation technology

Info

Publication number: CN106600434B
Application number: CN201610900984.8A
Authority: CN
Inventors: 王来刚; 郑国清; 郭燕; 贺佳; 程永政; 刘婷; 杨春英; 张彦; 杨秀忠; 张红利
Original assignee: Institute Of Agricultural Economics And Information Henan Academy Of Agricultural Sciences
Current assignee: Institute Of Agricultural Economics And Information Henan Academy Of Agricultural Sciences
Priority date: 2016-10-18
Filing date: 2016-10-18
Publication date: 2021-04-23
Anticipated expiration: 2036-10-18
Also published as: CN106600434A

Abstract

The invention relates to a crop growth monitoring and forecasting method, in particular to a crop growth remote sensing monitoring method based on a crop model and a assimilation technology, which takes GF-1 satellite data as a main information source and carries out inversion of growth indexes such as NDVI, EVI and LAI under the support of an application common key technology and an agricultural remote sensing basic product production subsystem; the regional WOFOST crop growth model is used as a core, the remote sensing growth index or crop yield is used as an assimilation quantity, an SCE-UA algorithm cost function is constructed, inversion assimilation aiming at GF-1 growth parameters is formed, and therefore crop growth can be effectively monitored and forecasted.

Description

Crop growth remote sensing monitoring method based on crop model and assimilation technology

Technical Field

The invention relates to a crop growth monitoring and forecasting method, in particular to a crop growth remote sensing monitoring method based on a crop model and a assimilation technology.

Background

Crop growth is a daily term for agriculture, and at present, no clear definition is provided for growth at home and abroad, and no standardized algorithm is provided. It is understood that growth is the growth and development of the crop, and the situation is interpreted as "the appearance" in the dictionary, i.e., the appearance of the growth and development of the crop. The trend is also a philosophy concept in Chinese, the potential and shape pairs, which indicates the trend determined by the specific structure and layout of things. It is appropriate that popbanger et al define crop growth as "conditions and trends in crop growth". From experience, even when agricultural technicians, agricultural managers, farmers and agricultural workers are used as the term for the growth of crops, the biomass of crops is concerned, which is the basis of the growth judgment, and the biomass variation tendency is the same, i.e., the biomass is better when the biomass is changed from weak to strong, and the biomass is worse when the biomass is changed from weak to weak.

For the growth of a certain area of crops, the growth is determined by two aspects of individual characteristics and population characteristics of the crops, the individual characteristics of the crops are mainly described by the characteristics of roots, stems, leaves, flowers, ears and the like, taking wheat as an example, parameters reflecting the individual characteristics are as follows: the length, number and layout of roots, the plant height and tillering number, the number, shape and color of leaves, the grain number per spike and the thousand-grain weight, and the like.

The population characteristics comprise density, layout and dynamics, wherein the density refers to the number of plants or tillers in a unit area, the density is determined by the number of basic seedlings, the number of tillers and the number of spikes by taking wheat as an example, the layout refers to the plant distribution condition, the layout refers to the uniform condition of plant distribution caused by seedling shortage by taking corn as an example, and the dynamics mainly refers to the growth period and the condition of environmental stress on crops. The individual parameters and the population parameters can be directly counted, and the other parameters need to be comprehensively calculated by combining the characteristics of the individuals and the population, including the leaf area index and the coverage.

The individual characteristics of the crops can be easily measured, but the comprehensive evaluation of the growth vigor of different individuals is difficult, the description of the group characteristics of the crops in a smaller range can be realized, and the judgment of the quality of the group characteristics is also empirical. The crops always grow and develop in specific regions and specific environments, particularly in dynamic climatic environments, and the characteristics of individual crops and groups are difficult to be integrated by using a comprehensive index so as to accurately reflect the growth situation of the crops. In reality, the main crop growth is always focused on a larger area, and various area difference factors make quantitative accurate description and evaluation of the crop growth in the larger area more difficult.

The research and establishment of a mature regional crop parameter inversion technology oriented to the crop growth monitoring and crop model assimilation system application requirements are necessary requirements and trends of digital agriculture. A large number of mature crop growth mechanism models are widely applied to the researches of single-point and regional crop growth process simulation prediction, field management, growth monitoring, yield estimation and the like. When data assimilation techniques are coupled to these models, it is necessary to fully consider the mechanistic nature of each model, the sensitivity and uncertainty of assimilation parameters, and the applicability to regional extension. These problems have not been solved well in the existing studies on the assimilation of crop models, and there is a lack of systematic and intensive research and discussion. At present, no relevant report of a large-area or global crop model data assimilation service operation system published at home and abroad exists. The intensive research and effective solution of the key problems can help to improve the crop monitoring and forecasting level of the agricultural condition remote sensing monitoring service.

The combination of remote sensing information and crop growth models is a typical example of interdisciplinary reinforcement. Although there have been some studies in recent years on assimilation methods in which remote sensing information is combined with crop growth models, there have been few studies on problems such as complexity of crop growth models themselves, assimilation algorithms, and the like. Some studies, while extending to the regional scale, coupled crop growth models are still relatively simple and many processes do not take into account, or make assumptions. The remote sensing information and a complete crop mechanization and chemistry growth model are really combined and applied to areas, and research with good effect is achieved. So far, most of remote sensing data used in such research is ground spectral measurement data and aerial images obtained by self-test, and only regional application research mainly uses AVHRR remote sensing data, as well as SPOT/HRV data and MODIS data.

The selection of the combination point of the remote sensing information and the crop growth model is mainly based on LAI and reflectivity, and other parameters such as surface temperature, soil humidity and the like are also used; the center of gravity of many studies is put on assimilating remote sensing observation data to correct the model, especially most studies take LAI concerned in the ecosystem as a connection variable with a crop growth model, and few assimilations work aiming at the LAI is carried out.

Similarly, the adopted optimization algorithm has different effects, for example, many researchers adopt the ensemble kalman filtering method, which can perform parallel computation, but the problem of filtering divergence often occurs in the practical application of the ensemble kalman filtering, which is shown in that the analysis value is closer to the background field with the increase of assimilation time, and finally, observation data is completely rejected. The assimilation strategy is also an existing problem, what cost function is adopted will affect the assimilation result, and the cost function based on the prior knowledge has a perfect theoretical system, so that not only the prior knowledge item is introduced, but also the uncertainty of each item in the cost function is considered (from the perspective of the nearest distance, the normalization of each item is represented); in addition, in the cost function based on the prior knowledge, a covariance matrix describing uncertainty of an observation item is a problem which always plagues inversion research, and few researches show how to describe the item.

In summary, since the combined research of the remote sensing information and the crop growth model (including other models) is an emerging research subject internationally, various research results are dispersed, comparable and mature and uniform, and therefore, it is very necessary to develop various researches and attempts, including used remote sensing data, models, methods and the like.

Data assimilation originated in the middle of the 20 th century, and with the improvement of global earth observation capability and the demand for global environmental change research in the 80 th century, data assimilation has attracted increasing attention as a bridge to link observed data and models. Data assimilation algorithms have also made great research progress in the past decades as important components of data assimilation systems, many scholars are continuously engaged in introducing new research results in the mathematical field into the data assimilation field, and a series of classical data assimilation algorithms such as variational algorithms and ensemble kalman filters are proposed and widely applied. In the 21 st century, intelligent algorithms represented by particle filtering and Bayesian methods are successively introduced into the data assimilation field, and the data assimilation is promoted to develop to a new height. However, under the combined action of various factors such as a high-dimensional space, multiple scales, nonlinearity, non-gaussian, complex uncertainty and state space correlation, the shortcomings of the existing data assimilation algorithms are increasingly prominent, and new data assimilation algorithms aiming at the problems are urgently needed to be proposed, which is also the development direction of the data assimilation algorithms.

Disclosure of Invention

The invention aims to provide a crop growth remote sensing monitoring method based on a crop model and a assimilation technology, which takes GF-1 satellite data as a main information source and develops inversion of long-term potential indexes such as NDVI, EVI and LAI under the support of an application common key technology and an agricultural remote sensing basic product production subsystem; the regional WOFOST crop growth model is used as a core, the remote sensing growth index or crop yield is used as an assimilation quantity, an SCE-UA algorithm cost function is constructed, inversion assimilation aiming at GF-1 growth parameters is formed, and therefore crop growth can be effectively monitored and forecasted.

The technical scheme adopted by the invention is as follows: a crop growth remote sensing monitoring method based on a crop model and a assimilation technology is characterized in that: the method comprises the following steps:

step 1, collecting, preprocessing and evaluating remote sensing data

Firstly, collecting high-resolution remote sensing data, and then preprocessing the data, namely processing the high-resolution remote sensing image, wherein the preprocessing comprises the following 4 steps:

1) radiation correction: 2 steps of radiometric calibration and atmospheric correction are adopted, so that system errors carried by satellite sensors and the like and errors generated in the atmospheric radiation transmission process can be eliminated or weakened; the correction converts original pixel values (DN values) of all wave bands of the image into surface reflectivity or radiance which can quantitatively reflect the real attributes of the ground objects and is used for calculating vegetation indexes required by various growth monitoring;

2) and (3) geometric correction: the method comprises 2 steps of orthorectification and geometric registration, and the steps can eliminate the geometric distortion and the solid position deviation of the image and enable the image to have the orthoprojection property; the step enables all wave band images of the same multispectral image to be completely overlapped, enables images of different time phases and the same position of different sensors to be completely overlapped, and provides reliable geographical reference for monitoring the growth vigor of crops;

3) image fusion: the spatial resolution of the original image can be improved through the steps, and a high-resolution remote sensing base map is provided for monitoring the growth of crops; this step can be implemented or not according to the actual decision;

4) identification of crop species and extraction of planting area: the categories of various crops in the working area can be decoded through the steps, the planting area of the crops can be counted, various special results can be made, and basic attribute information can be provided for monitoring the growth of the crops;

step 2, remote sensing image data storage and query

Constructing a remote sensing image library by using the image library function of ERDAS IMAGINE software; meanwhile, by taking the remote sensing data query system function of the project group as a reference, a remote sensing data query system capable of running in service is constructed for using subsequent data;

step 3, extracting and inverting crop growth parameters

Adopting a Band math function in ENVI software to obtain indexes reflecting growth parameters of wheat, corn and rice, wherein the indexes mainly comprise parameters such as NDVI, EVI, LAI and the like;

(4) growth parameter inversion

Firstly, sorting acquired field observation data, and analyzing the parameter sensitivity of a WOFOST model; then calculating the equivalent multispectral reflectivity of the remote sensing data, and carrying out inversion and error analysis on the growth parameters;

(5) growth parameter assimilation based on WOFOST crop growth model

The obtained basic data and the actually measured data observed in the field are sorted to prepare for the operation of the model;

analyzing the sensitivity parameters of the WOFOST model, and determining the parameters to be optimized;

constructing a cost function and realizing an SCE-UA optimization algorithm;

and analyzing the assimilation precision of the model.

The invention has the beneficial effects that:

1. the method collects basic data (weather, rainfall and the like), collects ground data of a test area, remote sensing data and substitute data RapidEye and Landsat-8, and sorts and preprocesses the obtained data.

2. OO-1 and OO-4 data, RapidEye and Landsat-8 are used as main data sources, and inversion research of potential indexes such as NDVI, EVI and LAI is carried out under the support of an application commonality key technology and an agricultural remote sensing basic product production subsystem.

3. A regional WOFOST crop growth model is taken as a core, or an ACRM spectrum simulation model is coupled, a remote sensing growth index or crop yield is taken as an assimilation quantity, an SCE-UA algorithm cost function is constructed, an OO-1-OO-4 growth parameter inversion assimilation technology is formed, and the method is used for crop growth product production in Henan province.

4. The assimilation problem of a larger area can be solved by combining a crop model and an operation scheme for assimilating remote sensing information with a distribution calculation method and establishing a downscaling optimization algorithm.

5. The remote sensing information is combined with the crop single-point model in a assimilation mode, although the point-to-surface inversion of parameters in the crop model can be achieved, the space-time distribution and the change characteristics of key elements of crop growth can be obtained, and the possibility is provided for monitoring the large-range growth vigor of crops, the method has obvious defects. First, although current crop models are developed based on crop growth mechanisms, the mechanisms are simulated by using mathematical formulas, and a large number of empirical values are used as coefficients in the mathematical simulation, so that the models are limited in wide-range application. Secondly, most of the existing researches aim at relatively homogeneous areas, most of images matched with the homogeneous areas are SPOT and TM, however, the spectral resolution of the images is low, high-precision crop canopy information is difficult to invert, errors are generated when remote sensing data are applied to model parameter correction, and the parameters are subjected to model operation to generate error accumulation, so that the deviation between the operation result of the assimilated model and the actual result is generated. In the research, the assimilation problem of a larger area can be solved by adopting an SCE-UA global optimization algorithm, combining a distribution calculation method and establishing a downscaling optimization algorithm.

People illustrated in attached drawings

FIG. 1 is a flow chart of the method of the present invention.

FIG. 2 is a flow chart of the method verification of the present invention.

FIG. 3 is a flow chart of the remote sensing image preprocessing of the present invention.

Fig. 4 is a diagram showing the arrangement of sampling points in a sampling area according to the present invention.

Detailed Description

Remote sensing image preprocessing

(ii) preprocessed content

Systematic errors carried by satellite sensors and the like and errors generated during atmospheric radiation transmission are eliminated or attenuated by radiation correction (including radiometric calibration and atmospheric correction). The step converts original pixel values (DN values) of all wave bands of the image into surface reflectivity or radiance which can quantitatively reflect the real attributes of the ground objects and is used for calculating vegetation indexes required by various growth monitoring.

Geometric distortion and solid position deviation of the image are eliminated through geometric correction (including orthorectification and geometric registration), and the orthographic projection property is achieved. The step enables all wave band images of the same multispectral image to be completely overlapped, enables images of different time phases and the same position of different sensors to be completely overlapped, and provides reliable geographical reference for monitoring the growth vigor of crops.

The spatial resolution of the original image is improved through image fusion, and a high-resolution remote sensing base map is provided for monitoring the growth of crops. This step may be performed based on actual decisions.

The method comprises the steps of extracting the planting area of crops, interpreting the categories of various crops in a working area, counting the planting area, making various special results and providing basic attribute information for monitoring the growth of the crops.

(II) selection of satellite images

And selecting a remote sensing image with proper ground resolution according to the requirement of monitoring result precision. The remote sensing monitoring of the crop growth with the precision of 1:250000, 1:100000, 1:50000 and 1:10000 scale is carried out, and remote sensing images with the spatial resolution of not less than 30m, 10m, 5m and 2.5m are preferably selected.

According to task requirements, a time phase is selected according to the time resolution of the image so as to meet the convention of investigation periods, and the remote sensing image with land utilization, vegetation coverage, crop species and crop planting structure change characteristics is correspondingly and easily distinguished in the image.

The task requirements and the operation conditions are comprehensively considered, and a proper spectrum range is selected. The spectral range adopted by the satellite remote sensing image is generally visible light, near infrared, thermal infrared, microwave and the like.

(III) radiation correction

Mainly including radiometric calibration and atmospheric corrections. Radiometric calibration is to eliminate systematic errors caused by factors such as sensors in image data, and usually a calibration formula is used to calculate the radiance or reflectance of an original image after calibration.

The formula for converting the image DN value into the radiance image using the absolute scaling factor is:

in the formula: in the formulaL_eThe equivalent radiance of the satellite camera wave band after conversion is in W.m^-2·sr^-1·μm^-1DN is a satellite camera wave band output count value; gain is the on-orbit radiation responsivity of the camera and has the unit of DN (W.m)^-2·sr^-1·μm^-1) Offset is the dark noise offset.

The purpose of atmospheric correction is to eliminate the influence of factors such as atmosphere and illumination on the reflection of the ground objects and obtain real physical model parameters such as the reflectivity, radiance and surface temperature of the ground objects, including eliminating the images of water vapor, oxygen, carbon dioxide, methane, ozone and the like reflected by the ground objects in the atmosphere; eliminating the influence of atmospheric molecule and aerosol scattering.

Atmospheric correction is generally based on statistical or physical models, and the correction method is chosen to comply with the following criteria: if the processing speed has higher requirement, simple and efficient methods such as a simplified dark pixel method and the like can be selected; if the requirement on the correction accuracy is high, a method based on an atmospheric radiation transmission model such as MODTRAN4 or 6S is required.

(IV) geometric correction

The geometric correction mainly comprises orthorectification and geometric registration. The ortho-image is an image in which the displacement of an image point due to the relief and the sensor error is corrected. In places with large relief, such as mountain areas, the orthographic correction is used for solving errors caused by the large relief, and the DEM is needed for the orthographic correction.

The orthoimage is generally produced by selecting some ground control points on the photo, and performing tilt correction and projective aberration correction on the image simultaneously by using the originally acquired Digital Elevation Model (DEM) data within the photo range, and resampling the image into an orthoimage, which is equivalent to rough correction on the geographical position of the original image.

The resampling method should be chosen to comply with the following criteria: if more spectral information of the original image needs to be reserved, selecting a nearest resampling method; if the image is required to be smoother and the visual effect is better, a more complex method such as bilinear interpolation is required to be selected.

Geometric registration is a necessary step when multi-temporal multi-sensor images are applied, in order to make multiple different images completely superimposed at the same position.

In the correction, the homonymous image points among different images need to be selected, and the selection of the homonymous image points should preferentially consider the obvious ground object points which are not easy to change along with time, such as buildings, ridge line junctions, river junctions or road junctions and the like. The image points of the same name should be distributed as uniformly as possible in the whole image range, and the number of the points should meet the requirement of the selected correction method.

Through automatic image matching, a large number of high-quality homonymous image points can be quickly distinguished and generated under the condition of no or little manual intervention. The method is mainly based on image matching of gray scale, and combines with image matching based on features, and uses a least square method to control the number and quality of the homonymous image points.

The correction method should be chosen to comply with the following criteria: the plain area with flat terrain adopts a polynomial method (second time or third time), and the processing speed is high; in areas with large relief, such as mountainous areas and hilly areas, a projection transformation method with DEM is adopted, so that the correction precision is high; other situations where effective calibration is not possible may choose linear/non-linear rubber stretch.

(V) high resolution image fusion

Image fusion can provide high-resolution fused multispectral imagery for a job with only high-resolution panchromatic imagery and low-resolution multispectral imagery.

The following criteria should be followed in the selection of the image fusion method: if the spectral characteristics of the multispectral image need to be better kept, Gram-Schmidt Pan Sharpening can be selected; if the texture features of the panchromatic image need to be highlighted, selecting the HCS; if the ratio of the Resolution of the multispectral image to the panchromatic image is 4:1, then the dedicated algorithm reactive Resolution Merge can be selected.

In image fusion, the ratio of the resolution of the multispectral image to the resolution of the panchromatic image should not be greater than 8:1, and the fusion effect is greatly reduced if the resolution difference is too large.

3.1.1.1 ground data acquisition

Basic information

Including survey time, survey location, GPS location, geographical location, administrative area location, crop type, crop variety, with sample area major crops and variety types see table 32.

TABLE 32 major crop variety types and maturity

(II) division of the growth period of the main crops

(1) Winter wheat growth stage division

The winter wheat growth period included: sowing, seedling emergence, tillering, overwintering, turning green, rising, jointing, booting, heading, flowering, maturity and ripening. The respective growth periods are defined in table 33.

TABLE 33 Standard for Observation of winter wheat growth period

(2) Summer corn growth period division

The summer maize has a growth period including: sowing, seedling emergence, three leaves, seven leaves, stem extension, large bell mouth, tasseling, spinning, milk maturity, complete maturity and harvesting. The respective growth periods are defined in table 34.

TABLE 34 summer maize growing period observation standard

(3) Division of rice growth period

The rice growth period includes: seedling, transplanting, turning green, tillering, booting, jointing, heading, flowering, grouting, yellow maturing and harvesting. The respective growth periods are defined in table 35.

TABLE 35 summer maize growing period observation standard

(3) Grading standard of main crop growth vigor

Grading standard for winter wheat growth

On one type of seedling: the winter wheat plant has good growth condition. The plant is strong, the density is uniform, the height is normal, the leaf color is normal, the inflorescence is good in development, the ears are large, and the seed is full; and no or only slight plant diseases and insect pests and meteorological disasters exist, the growth of the winter wheat is hardly influenced, and the yield can be expected to reach the level of a high-yield annual scene.

Under one kind of seedlings: the winter wheat plant has good growth condition. The plant is strong, the density is uniform, the height is normal, the leaf color is normal, the inflorescence is good in development, the ears are large, and the seed is full; slight plant diseases and insect pests and meteorological disasters have certain influence on the growth of winter wheat, and the yield is between a high-yield annual view and a normal annual view.

On the second kind of seedlings: the growth condition of winter wheat plants is moderate. The density of the plants is not uniform, and the phenomena of few seedlings and ridge breaking exist; the growth height is less than regular, and the ear is slightly smaller; the yield can be expected to reach the level of average yield when the system is subjected to insect damage or meteorological disasters.

Under the second kind of seedlings: the growth condition of winter wheat plants is moderate. The plant density is not uniform, and obvious seedling shortage and ridge breaking phenomena occur; the growth height is obviously irregular, and the ear is small; plants suffer from moderate insect damage or meteorological disasters, and the yield cannot be expected to reach the level of annual average yield.

Three types of seedlings: winter wheat plants grow poorly or poorly. The plant density is not uniform, the plant is short and small, the height is irregular, and the seedling and ridge are seriously lacked; the small grain size of the ears is less, the weeds are more, the diseases and the pests or the meteorological disasters seriously damage winter wheat, the expected yield is much lower, and the yield level is the yield level of the annual view of reduced production.

Second summer maize growth standard

On one type of seedling: the summer maize plants have good growth conditions. The plant is strong, the density is uniform, the height is normal, the leaf color is normal, the inflorescence is good in development, the ears are large, and the seed is full; no or only slight plant diseases and insect pests and meteorological disasters exist, the growth of summer corns is hardly influenced, and the yield can be expected to reach the level of a high-yield annual scene.

Under one kind of seedlings: summer maize plants grow well. The plant is strong, the density is uniform, the height is normal, the leaf color is normal, the inflorescence is good in development, the ears are large, and the seed is full; slight plant diseases and insect pests and meteorological disasters have certain influence on the growth of summer corns, and the yield is between a high-yield annual view and a normal annual view.

On the second kind of seedlings: summer maize plants grow moderately. The density of the plants is not uniform, and the phenomena of few seedlings and ridge breaking exist; the growth height is less than regular, and the ear is slightly smaller; the yield can be expected to reach the level of average yield when the system is subjected to insect damage or meteorological disasters.

Under the second kind of seedlings: summer maize plants grow moderately. The plant density is not uniform, and obvious seedling shortage and ridge breaking phenomena occur; the growth height is obviously irregular, and the ear is small; plants suffer from moderate insect damage or meteorological disasters, and the yield cannot be expected to reach the level of annual average yield.

Three types of seedlings: summer maize plants grow poorly or poorly. The plant density is not uniform, the plant is short and small, the height is irregular, and the seedling and ridge are seriously lacked; the small grain size of the ears is less, the weeds are more, the disease and pest damage or meteorological disasters seriously damage summer corns, the expected yield is much lower, and the yield level is the yield level of the annual view of production reduction.

③ Rice growth grading Standard

On one type of seedling: the rice plant has good growth condition. The plant is strong, the density is uniform, the height is normal, the leaf color is normal, the inflorescence is good in development, the ears are large, and the seed is full; no or only slight plant diseases and insect pests and meteorological disasters exist, the growth of the rice is hardly influenced, and the yield can be expected to reach the level of a high-yield annual scene.

Under one kind of seedlings: the growth condition of rice plants is good. The plant is strong, the density is uniform, the height is normal, the leaf color is normal, the inflorescence is good in development, the ears are large, and the seed is full; slight plant diseases and insect pests and meteorological disasters have certain influence on the growth of rice, and the yield is between a high-yield annual landscape and a normal annual landscape.

On the second kind of seedlings: the growth condition of rice plants is moderate. The density of the plants is not uniform, and the phenomena of few seedlings and ridge breaking exist; the growth height is less than regular, and the ear is slightly smaller; the yield can be expected to reach the level of average yield when the system is subjected to insect damage or meteorological disasters.

Under the second kind of seedlings: the growth condition of rice plants is moderate. The plant density is not uniform, and obvious seedling shortage and ridge breaking phenomena occur; the growth height is obviously irregular, and the ear is small; plants suffer from moderate insect damage or meteorological disasters, and the yield cannot be expected to reach the level of annual average yield.

Three types of seedlings: the growth conditions of rice plants are poor or bad. The plant density is not uniform, the plant is short and small, the height is irregular, and the seedling and ridge are seriously lacked; small grain size, a lot of weeds, serious damage to rice caused by plant diseases and insect pests or meteorological disasters, and low expected yield, which is the yield level of the annual view of reduced production.

(4) Primary crop soil information

Information of soil type

The method comprises the steps of soil type, national standard code, soil size fraction composition, soil organic matter content and soil saturated water holding capacity.

Information on soil moisture

The soil moisture measurement depth is 0-20 cm, and the soil moisture content is 20-40 cm.

(5) Sampling point layout method

And (3) laying a sample square: 10-20 sample parties are selected in the monitored area, and the size of the sample parties is 1km multiplied by 1 km. The distribution of the squares is relatively uniform and requires a certain representativeness of the planting structure in the squares.

Sample survey time: early in the growth monitoring.

Survey content of a sample party: planting structure and growth condition in the sample.

The method for surveying the sample comprises the following steps: and measuring the planting structure chart in the sample by using a GPS.

Principle of selecting sample area

The research area of the field sample point of the project is selected from the city of Chang (winter wheat-summer corn) and the city of Xinyang (rice). According to the needs of division and experiments, representative 5 sample point counties are selected in the Schanchang city, 3-5 sample areas are distributed in each county, and 22 sample points are distributed in total; similarly, a representative 5 sample points county are selected in Xinyang city, and 20 sample points are arranged in total.

Principle of sampling area in spot county:

a. the sample area is the main production area of the county, and the area of the sample area is not less than 1km multiplied by 1 km.

b. The sample area is far away from villages as far as possible, and relatively flat and regular cultivated land is selected as far as possible.

c. The planting system of crops in the sample area is relatively stable.

Second principle of arrangement of spots

3-5 sampling points are selected in each sampling area (1km multiplied by 1km), the arrangement of the sampling points should be uniformly distributed in the sampling area, and the coverage range of the observation points can be uniformly distributed in a square shape or in a concentric circle mode. And selecting specific points with the observation range of about 9 square meters from each observation sample point for observation, as shown by the range in a green frame in fig. 4.

(6) Crop growth information collection

Ground data survey

The investigation content mainly comprises the collection of 7 types of data such as crop planting area ground sample prescription, phenological period, field management, soil parameters, physical parameters, physiological parameters, yield and the like.

② investigation of ground sample prescription of crop planting area

Uniformly distributing 10 ground sample parties, wherein each sample party is not less than 1km multiplied by 1km, and in 2013-2014, respectively carrying out one-time measurement on winter wheat, summer corn and rice every year by adopting a differential GPS.

Observation of the phenological period of crops

The observation was performed 1 time every 5 days, and the phenological period of the crop was recorded.

Investigation of crop field management

Recording the field management measures of the crops during the growth and development period, including soil preparation, sowing, pesticide spraying, fertilization, intertillage, weeding, pest and disease damage, irrigation, harvest and the like, and observing for 1 time every 15 days.

Investigation of crop soil parameters

Before crop seeding, indexes such as soil basic fertility are investigated, and the indexes comprise: soil texture, soil moisture, soil organic matter, soil total nitrogen, soil alkaline hydrolysis nitrogen, soil available phosphorus and soil available potassium.

Investigating physical parameters of crops

The method comprises the contents of crop growth grade, height, density, coverage, colony tillering and the like, and the plant height and dry matter are measured after 1 observation every 15 days.

Seventhly, investigating physiological parameters of crops

Including Leaf Area Index (LAI), chlorophyll content, spectrum, Photosynthetically Active Radiation (PAR), etc., were observed 1 time every 15 days.

Investigation of crop yield

Actual yield: in the mature period, 2 square sampling points with the side length of 2m are selected from various parties, and the seeds are independently harvested and threshed, wherein the process is repeated for 3 times, the yield is measured by weighing, and the water content of the seeds is calculated to be 13%.

Theoretical yield: the number of wheat ears, the number of grains per ear and the thousand-grain weight are investigated in the mature period, and the theoretical yield is calculated.

Theoretical yield (kg/mu) is equal to ear number per mu x ear number per thousand seed weight (g)/1000 x 1000

(7) Ground sampling and analyzing method

Firstly, soil sample collection

Before crop seeding, collecting soil layer samples of 0-40 cm at various points in the field by using a soil drill, wherein each 20cm is one layer, and the total number of the soil layer samples is 2; loading a part of the sampled soil into an aluminum box, weighing the fresh soil, drying the fresh soil at 110 ℃ to constant weight, and calculating the water content of the soil; the other part is put into a self-sealing bag and taken back to a laboratory for natural air drying, thereby being convenient for measuring the soil foundation fertility.

Measurement of Dry matter

The sampling period of the test is that sampling is carried out once every 15 days, 10 representative plants are selected by each sample prescription and brought back to the room, and the samples are divided according to leaves, stems, leaf sheaths, glumes, spike stalks and seeds. The samples were de-enzymed in an oven at 105 deg.C for 15min, and then dried at 80 deg.C to constant weight.

Third, measuring physiological indexes

Measuring the chlorophyll content by using a SPAD-502 chlorophyll measuring instrument. Selecting sunny and calm weather, measuring for 10: 00-14: 00, and measuring the spectrum by using a spectrometer; LAI and canopy photosynthetic active radiation were measured using a SunScan canopy analyzer.

(8) Crop sampling method

Density sampling method

And randomly selecting crop plant sample segments of six rows per meter at each sample point in the field, measuring the length and the row spacing, and calculating the mu density. The plant height of the crops is measured in the range, and the standard for measuring the plant height is as follows: only the distance from the ground to the height of the crop canopy (natural plant height) is measured.

② sampling method

Randomly selecting 30 plants at various points in the field, taking the plants back to a laboratory to measure the leaf area, and adopting a length-width coefficient method for the leaf area. The organs (leaves, stem sheaths, spike glumes and seeds) are subjected to sample separation, then the water is removed for 30min in an oven at 105 ℃, the mixture is dried to constant weight at 80 ℃ and weighed. To calculate leaf area, the area of the sampled plants (length and width) was recorded for each spot.

3. Monitoring of crop growth

(1) Means of monitoring

The remote sensing data multispectral wave bands are used for carrying out linear and nonlinear combination to generate vegetation indexes (such as NDVI, EVI and the like), the agricultural parameters (such as LAI and biomass) observed on the ground of large crops such as winter wheat, corn, rice and the like are combined to establish the correlation between the vegetation indexes and the agricultural parameters, and the agricultural parameters of large-area remote sensing inversion are generated by utilizing the quantitative relation, so that the growth vigor of the crops is indicated.

(2) Principle of monitoring

The remote sensing satellite sensor records the spectral reflectivity of the ground object, and when the environmental factors of the images shot by the satellite are basically the same, the differences of the reflectivities of the similar crops can be regarded as the differences of the crops in the aspects of the growth process, the growth vigor, the physiological condition and the like. By combining the growth characteristics of crops, the remote sensing image is correspondingly processed and analyzed, so that indexes closely related to the growth condition of the crops can be constructed, and the growth vigor of the crops can be monitored.

Quantitative measurement of the Vegetation Index (VI) can not only indicate vegetation vigor, but also the vegetation index has better sensitivity to detect biomass over a single band. The vegetation index is a dimensionless number and is a unique spectral signal extracted by using optical parameters of plant leaf crowns. The vegetation index is linearly related to the crop distribution density, not only is the optimal indicator factor of the crop growth state and the crop spatial distribution density, but also the size of the vegetation index is closely related to elements such as the coverage (horizontal density) and the leaf area index (vertical density) of the crop, so that the leaf area index, the coverage and other crop growth indicators of the crop can be estimated by a remote sensing method.

(3) Monitoring content

Firstly, designing a quantitative remote sensing monitoring and evaluating index system for growth of bulk crops

The optimum vegetation indexes for monitoring the growth of crops in different growth periods in the same region are different, the optimum vegetation indexes for monitoring the growth of corns in the same growth period in different regions are also different, and the effectiveness and the suitability for monitoring the growth periods of different growth periods of bulk crops are researched by starting from two vegetation indexes, namely the NDVI and the EVI, so that a main growth quantitative remote sensing monitoring and evaluation index system for different growth periods in different regions of bulk crops is designed.

② investigation of quantitative remote sensing monitoring and evaluation technology for growth vigor of bulk crops

The traditional growth qualitative analysis method is combined to research the quantitative growth remote sensing monitoring and evaluation technology. The leaf area index is the most direct index reflecting the growth and development conditions of crops, and a quantitative remote sensing monitoring technology for evaluating the growth vigor of bulk crops by utilizing the vegetation index is discussed by establishing a quantitative relation model between the vegetation index of the crops and biophysical parameters such as the leaf area index and the like.

(4) Monitoring method

Selection of remote sensing parameters

According to the spectral response characteristics of crops, the reflectivity of the crops to visible light and near infrared light is greatly different at different growth and development stages, so that the indicating effect of the vegetation index constructed based on the spectral reflectivity on the growth vigor of the crops at different growth stages is correspondingly changed. Meanwhile, the vegetation indexes are constructed by different mechanisms, and differences exist in the correction of the background of the growing environmental conditions of crops, such as climate, soil and the like. Therefore, the adaptability of the growth indicators of the crops at different growth stages needs to be verified. The process is realized by establishing the correlation between the vegetation index in the whole growth stage of the crop and the agronomic parameter LAI, analyzing the correlation coefficient and using the vegetation index with the highest correlation to invert the growth vigor of the crop in different growth stages.

Inverse model of agronomic parameters

The leaf area index data used to fit the model parameters can be obtained by both surface measurements and model simulations. The earth surface measurement is used as a true value of the leaf area index of the vegetation, and an empirical model between the earth surface measurement and the true value can be established for inversion of the leaf area index by fitting the relationship between the true value and the remote sensing vegetation index. In addition, the leaf area index can also be obtained by adopting physical model simulation representing the canopy radiation transmission process, and the problem that the earth surface measurement lacks representativeness in the aspects of observation angle, vegetation type, background condition and the like can be solved by establishing a relation model by simulating the leaf area index and the vegetation index through the model.

The function form between the leaf area index and the vegetation index is different along with the difference of the vegetation index and the vegetation type, and the optimal function form and parameters are respectively fitted and selected for different areas and vegetation types, wherein the function form is linear, exponential, logarithmic, polynomial and power function.

Linear relation: y ═ ax ═ b

The exponential relationship: aeb^π

Logarithmic relation: y ═ aln (x) + b

The power function relation: y is ax^b

Second degree polynomial relation: y is ax²+bx+c

And (3) selecting a function with fitting coefficient of more than 0.7, significance of less than 0.05 and crop growth mechanism to perform inversion of the leaf area index. This process is performed at the statistical analysis function of the SPSS software.

(5) Monitoring result verification

Ground on-site observation and verification

The method can be used for verification by adopting a sampling investigation method, the number of the verification samples is checked on the spot according to the fact that the growth monitoring results of various types are not less than 5% of the total area, and the verification samples are uniformly distributed in space. Accurately positioning a sample space by adopting a GPS (global positioning system), and if a field growth judgment result within the geometric positioning accuracy of a remote sensing pixel is consistent with a remote sensing monitoring growth result, determining that the monitoring result is correct; otherwise, the monitoring result is wrong. And (4) filling a verification record table for field verification results, and continuously perfecting the growth inversion model. The field verification comprises information extraction result verification, difficult and difficult point in growth monitoring, sample verification needing to be supplemented, and monitoring result verification with larger difference compared with the existing data.

Verification of other satellite data

For regional verification of the growth monitoring result, the same or similar observation data of other satellites are used for detecting the result monitoring precision by adopting the same generation method of the product.

(6) Monitoring product manufacture

The monitoring product represents the crop growth monitoring result in the forms of characters, thematic maps, statistical tables and the like. The text information refers to the related information describing the satellite remote sensing growth monitoring result: including time, range, satellite, and sensor, etc.; the growth monitoring thematic map comprises a map name, a legend, a scale and geographic information of an administrative region, and the growth monitoring result space is distributed by coloring a first-class seedling, a second-class seedling and a third-class seedling in a layering mode so as to highlight the growth change of crops. For the character description in the daily, monthly and seasonal monitoring products and the statistics of the growth rate in each administrative region in the statistical table, the statistics of the growth rate in a specific daily day is based on the image of the single scene product on the same day. The statistics of the growth grades in each administrative region in the monthly and seasonal growth monitoring products are respectively the sum of the values of the pixels in the products in each day in the month and the season.

Product quality evaluation standardization process

1. Remote sensing image preprocessing quality evaluation

(1) Geometric correction

By taking a base map with an accurate geographic position as a reference, the picture surface error of the remote sensing image after geometric correction is not more than 0.5mm, and the maximum value is not more than 1 mm. When the image is corrected by adopting the image, the error of the image after the image is registered with the image should not be more than 0.5 pixel, and the maximum area with large topographic relief in the mountainous area should not exceed 1 pixel.

(2) Atmospheric correction

The spectrum reflection characteristics of the vegetation area of the remote sensing image corrected by the atmosphere accord with the basic characteristics of healthy vegetation, and particularly, the reflectivity of the high reflection of a blue wave band is greatly reduced after the atmospheric correction.

2. Evaluation of product quality

(1) Inverse model evaluation

Based on a statistical regression model established by remote sensing parameters and sampling LAI, a function with fitting coefficient larger than 0.7, significance smaller than 0.05 and expression of crop growth mechanism is selected to carry out inversion of leaf area index. If the model precision shows better precision, the index model is preferentially selected to carry out the inversion of the leaf area index, if the precision is not proper, the representativeness of the ground sampling data is verified again, and a regression model is established until the precision meets the requirement.

(2) Quality evaluation of growing products

The verification of the growing products is mainly ground verification, crop LAI is collected from sampling points, consistency verification is carried out on the crop LAI and remote sensing inversion LAI, and Root Mean Square Error (RMSE) and Relative Error (RE) are adopted for analysis and evaluation. The final product was divided into 5 grades in total, as follows:

a level: and comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.1, the quality grade of the product is determined to be A grade.

B stage: and comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.2, the quality grade of the product is determined to be B grade.

C level: and comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.3, the quality grade of the product is determined to be C grade.

D stage: and comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.4, the quality grade of the product is determined to be D grade.

E, grade: and comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is more than 0.4, the quality grade of the product is determined to be grade E.

Satellite on-orbit operation post-product verification and improvement scheme

Product verification scheme

And (3) verifying by adopting a sampling investigation method, wherein the number of the verification samples is verified to be checked on the spot according to the fact that the growth monitoring results of various types are not less than 5% of the total area, and the verification samples are uniformly distributed in space.

Accurately positioning a sample space by adopting a GPS (global positioning system), and if a field growth judgment result within the geometric positioning accuracy of a remote sensing pixel is consistent with a remote sensing monitoring growth result, determining that the monitoring result is correct; otherwise, the monitoring result is wrong.

And (4) filling a verification record table for field verification results, and continuously perfecting the growth inversion model. The field verification comprises information extraction result verification, difficult and difficult point in growth monitoring, sample verification needing to be supplemented, and monitoring result verification with larger difference compared with the existing data.

And collecting crop LAI from the sampling point, carrying out consistency verification on the crop LAI and remote sensing inversion LAI, and analyzing and evaluating by adopting Root Mean Square Error (RMSE) and Relative Error (RE).

Product validation criteria are given in the following table.

Product quality verification standard table

Quality grade	Description of the invention	Validation standards
			Class A	Cloud-free image and growth grading precision>90％	Ground verification, error<10％
Class B	Cloud amount of image<10% growth grading precision>80％	Ground verification, error<20％
			Class C	Cloud amount of image is 10-20%, and grading precision of growth vigor>70％	Ground verification, error<30％
Class D	20-30% of image cloud amount and growth grading precision>60％	Ground verification, error<40％
			Class E	Cloud amount of image>30% of growth grading precision<60％	Ground verification, error>40％

Product verification implementation step

1. Arrangement of ground sampling points

Selecting 5 representative sample point counties in the xuchang city in the research area, wherein 3-5 sample areas are distributed in each county, and 22 sample points are distributed in total; similarly, a representative 5 sample points county are selected in Xinyang city, and 20 sample points are arranged in total.

Principle of sampling area in spot county:

(1) the sample area is the main production area of the county, and the area of the sample area is not less than 1km multiplied by 1 km.

(2) The sample area is far away from villages as far as possible, and relatively flat and regular cultivated land is selected as far as possible.

(3) The planting system of crops in the sample area is relatively stable.

Sample survey time: early in the growth monitoring.

3-5 sampling points are selected in each sampling area (1km multiplied by 1km), the arrangement of the sampling points should be uniformly distributed in the sampling area, and the coverage range of the observation points can be uniformly distributed in a square shape or in a concentric circle mode. And selecting a specific point with an observation range of about 9 square meters from each observation sample point for observation.

2. Crop growth information collection

The ground survey content mainly comprises the collection of 7 types of data such as crop planting area ground sample, phenological period, field management, soil parameters, physical parameters, physiological parameters, yield and the like. Uniformly distributing 10 ground sample parties, wherein each sample party is not less than 1km multiplied by 1km, and in 2013-2014, respectively carrying out one-time measurement on winter wheat, summer corn and rice every year by adopting a differential GPS. The observation was performed 1 time every 5 days, and the phenological period of the crop was recorded. Recording the field management measures of the crops during the growth and development period, including soil preparation, sowing, pesticide spraying, fertilization, intertillage, weeding, pest and disease damage, irrigation, harvest and the like, and observing for 1 time every 15 days. Before crop seeding, indexes such as soil basic fertility are investigated, and the indexes comprise: soil texture, soil moisture, soil organic matter, soil total nitrogen, soil alkaline hydrolysis nitrogen, soil available phosphorus and soil available potassium. The method comprises the contents of crop growth grade, height, density, coverage, colony tillering and the like, and the plant height and dry matter are measured after 1 observation every 15 days. Including Leaf Area Index (LAI), chlorophyll content, spectrum, Photosynthetically Active Radiation (PAR), etc., were observed 1 time every 15 days. Actual yield: in the mature period, 2 rectangular sampling points with the side length of 2m are selected from various parties, and the seeds are independently harvested and threshed, wherein the harvesting and the threshing are repeated for 3 times, the yield is measured by weighing, and the water content of the seeds is calculated to be 13%.

Crop sampling method

3. Acquisition of other remote verification data

And selecting a remote sensing image with the ground resolution not lower than 0.5m according to the requirement of the monitoring result precision. According to the growth characteristics of the research object, images in the crop growth period are selected for monitoring growth vigor, and the characteristics of land utilization, vegetation coverage, crop types and crop planting structure change can be distinguished.

The quality of the satellite image should meet the following requirements:

(1) and selecting panchromatic or multispectral images with smaller inclination angles and covering the working area. The images are correspondingly covered as far as possible in the key growth period of the crops, and at least one image is covered in each key growth period. The image is required to have rich and clear layers, uniform tone, moderate contrast and no noise or strip loss.

(2) The overlap between adjacent scene images should be not less than 4% of the image width, and in special cases the overlap may be less than the above-mentioned index.

(3) The cloud coverage in the image should be less than 3% and should not cover important terrain in the crop growth monitoring area. The total area of the dispersed cloud layers should not exceed 8% of the area of the operation area.

4. Product accuracy verification

(1) Grade A product identification standard

And comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.1, the quality grade of the product is determined to be A grade.

(2) Class B product identification standard

And comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.2, the quality grade of the product is determined to be B grade.

(3) Grade C product identification standard

And comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.3, the quality grade of the product is determined to be C grade.

(4) Grade D product identification standard

And comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is less than 0.4, the quality grade of the product is determined to be D grade.

(5) Class E product identification standards

And comparing and analyzing the data of the sampling points investigated by the ground test with the result obtained by inversion. Taking the schchang research area as an example, 22 sample data of ground survey are compared with the inversion result, and if RE is more than 0.4, the quality grade of the product is determined to be grade E.

Claims

1. A crop growth remote sensing monitoring method based on a crop model and a assimilation technology is characterized in that: the method comprises the following steps:

step 1, collecting, preprocessing and evaluating remote sensing data

1) radiation correction: the method comprises two steps of radiometric calibration and atmospheric correction, and is used for eliminating or weakening system errors carried by a satellite sensor and errors generated in the atmospheric radiation transmission process; the correction converts the original pixel values of each wave band of the image into the earth surface reflectivity or radiance which can quantitatively reflect the real attributes of the ground objects and is used for calculating the vegetation index required by various growth monitoring;

in radiometric calibration, the formula for converting the DN value of an image into a radiance image using an absolute calibration coefficient is:

in the formula: in the formula L_eThe equivalent radiance of the satellite camera wave band after conversion is in W.m^-2·sr^-1·μm^-1DN is a satellite camera wave band output counting value, and the satellite camera wave band output counting value is an original pixel value of each wave band of the image; gain is the on-orbit radiation responsivity of the camera and has the unit of DN/(W.m)^-2·sr^-1·μm^-1) Offset is the dark noise offset;

2) and (3) geometric correction: the method comprises two steps of orthorectification and geometric registration, wherein the geometric distortion and the solid position deviation of an image are eliminated through the two steps, and the image has an orthographic projection property; the step is to completely superpose the images of all wave bands of the same multispectral image, completely superpose the images of the same position of different sensors at different time phases, and provide a geographical reference for monitoring the growth of crops;

3) image fusion: the spatial resolution of the original image is improved through the step, and a high-resolution remote sensing base map is provided for monitoring the growth of crops; the step is to determine whether to implement according to the actual situation;

4) identification of crop species and extraction of planting area: the categories of various crops in the working area are decoded through the step, the planting area of the crops is counted, various special results are made, and basic attribute information is provided for monitoring the growth of the crops;

step 2, remote sensing image data storage and query

Constructing a remote sensing image library by using the image library function of ERDAS IMAGINE software; meanwhile, by taking the remote sensing data query system function of the project group as a reference, a remote sensing data query system for service operation is constructed for use of subsequent data;

step 3, extracting and inverting crop growth parameters

Adopting a Band math function in ENVI software to obtain indexes reflecting growth parameters of wheat, corn and rice, wherein the indexes comprise NDVI (normalized difference index), EVI (error variance index) and LAI (label average index); by establishing a correlation between the vegetation index and the agronomic parameter LAI in the whole growth stage of the crops, analyzing the correlation coefficient, and using the vegetation index with the highest correlation to invert the growth vigor of the crops in different growth stages;

(4) growth parameter inversion

Firstly, sorting acquired field observation data, and analyzing the parameter sensitivity of a WOFOST model; then calculating the multispectral reflectivity of the remote sensing data effect, and carrying out inversion and error analysis on the growth parameters;

simulating by adopting a physical model representing the canopy radiation transmission process to obtain a leaf area index, and establishing a relation model by simulating the leaf area index and the vegetation index through the model;

respectively fitting and selecting an optimal function form and parameters for different regions and vegetation types, wherein the function form is linear, exponential, logarithmic, polynomial and power function;

linear relation: y ═ ax + b

The exponential relationship: ae^bx

Logarithmic relation: y ═ aln (x) + b

The power function relation: y is ax^b

Second degree polynomial relation: y is ax²+bx+c

Selecting a function with fitting coefficient more than 0.7, significance less than 0.05 and crop growth mechanism to carry out inversion of leaf area index;

(5) growth parameter assimilation based on WOFOST crop growth model

constructing a cost function and realizing an SCE-UA optimization algorithm;

analyzing the assimilation precision of the model;

the remote sensing data multispectral wave bands are used for carrying out linear and nonlinear combination to generate vegetation indexes, quantitative relations between the vegetation indexes and agricultural parameters observed on the ground of bulk crops are established, and the agricultural parameters of large-area remote sensing inversion are generated by the quantitative relations, so that the growth vigor of the crops is indicated; the remote sensing satellite sensor records the spectral reflectivity of the ground object, when the environmental factors of the images shot by the remote sensing satellite are the same, the reflectivity difference of the same kind of crops is the difference of the growth process, the growth vigor and the physiological condition of the crops, and the remote sensing images are processed and analyzed by combining the growth characteristics of the crops, so that the indexes related to the growth condition of the crops are constructed, and the growth vigor of the crops is monitored;

the crop growth remote sensing monitoring method based on the crop model and the assimilation technology uses GF-1 satellite data as an information source, and carries out inversion of NDVI (normalized difference of arrival), EVI (evolution index) and LAI (evolution index) under the support of an application common key technology and an agricultural remote sensing basic product production subsystem; constructing an SCE-UA algorithm cost function by taking a regional WOFOST crop growth model as a core and taking a remote sensing growth index or crop yield as an assimilation quantity to form inversion assimilation aiming at GF-1 growth parameters, thereby effectively monitoring and forecasting the growth of crops; the method adopts an SCE-UA global optimization algorithm, combines a distribution calculation method and establishes a downscaling optimization algorithm, the assimilation process comprises direct assimilation and indirect assimilation, the direct assimilation obtains an intersatellite comparable qualitative growth parameter, and the indirect assimilation obtains a yield related quantitative growth parameter, so that the assimilation problem of a larger area is solved.