Multi-source remote sensing data quantitative inversion regional crop water consumption demand data fusion method

Abstract

The invention relates to a multi-source remote sensing data quantitative inversion regional crop water consumption demand data fusion method. The method comprises the steps of data acquisition, data assimilation, data fusion and formation of a multi-source heterogeneous database. A distributed multi-source heterogeneous database is built by using advanced technologies such as comprehensive grid computing and cloud computing, a high-performance storage retrieval multi-source heterogeneous database and a data product oriented to a complex irrigation data system are built, agricultural information is collected in real time, a database is formed, and a database management system is established. And finally, efficient data management and sharing are realized. Related agricultural irrigation processes are systematically monitored, including monitoring of water consumption of crops, monitoring and forecasting of agricultural irrigation and the like. An agricultural production management system and an agricultural expert system are developed and are mainly used for diagnosing water shortage, fertilizer shortage and pest and disease damage of crops. The data acquisition forms are diversified, the parameter optimization quantity is large, the acquisition of large-range farmland parameters is realized, and a complete system is formed with remote sensing irrigation monitoring.

Description

Technical Field

[0001] The invention belongs to the field of intelligent crop irrigation, and in particular relates to a method for fusing regional crop water consumption data by quantitatively inverting multi-source remote sensing data. Background Art

[0002] The acquisition, management, processing and analysis of complete agricultural irrigation source data, decision support and information sharing provide data support for the development of high-water-efficiency agriculture. With the advancement of remote sensor technology, agricultural irrigation-related data sources are becoming increasingly abundant, from the initial multispectral data to the currently widely used hyperspectral data, radar data, hyperspectral data, multi-angle data, etc. Remote sensing data can be divided into high, medium and low spatial resolution remote sensing data based on spatial resolution. At present, medium and high resolution remote sensing data are used to implement the area monitoring of major crop production areas. With the development of computer technology and artificial intelligence technology, the development of intelligent crop area, water consumption and growth characteristics monitoring methods based on machine learning, deep learning and other algorithms can reduce human influence and improve monitoring efficiency. At the same time, with the continuous increase in the amount of remote sensing data, the corresponding agricultural big data analysis technology, combined with the historical data of remote sensing estimation, site monitoring data, etc., can also improve the efficiency and accuracy of crop area, water consumption and growth characteristics monitoring, and provide technical support for the periodic, long-term and quantitative high-water-efficiency agricultural irrigation database and data product standardization production.

[0003] The construction of agricultural information database in my country has achieved initial success, but the following problems exist in the construction of related databases: First, the distribution of crops in my country is wide and scattered, the planting structure is complex, and similar information is more difficult to obtain than in foreign countries. Database construction is limited to a small range and a small number of parameter optimizations, especially the acquisition technology of large-scale farmland parameters is still imperfect, and related technologies have not formed a complete system with remote sensing irrigation monitoring; secondly, there is no distributed multi-source heterogeneous database built with advanced technologies such as comprehensive grid computing and cloud computing, and there is no high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems.

[0004] In this regard, the present invention provides a method for fusion of regional crop water consumption data for quantitative inversion of multi-source remote sensing data, which combines the construction of my country's irrigation test station network, relevant data spatialization models and algorithms, and spatial data fusion algorithms, and utilizes relevant interpolation and fusion algorithms to significantly improve the relevant technologies of agricultural remote sensing monitoring systems, and integrates the algorithm system into the development of high-water-efficiency agricultural irrigation databases and data products. This method can better serve the development of my country's high-water-efficiency irrigation agriculture, and has broad market prospects. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method for fusion of regional crop water consumption data for quantitative inversion of multi-source remote sensing data that combines the construction of my country's irrigation test station network, relevant data spatialization models and algorithms, and spatial data fusion algorithms, and utilizes relevant interpolation and fusion algorithms to significantly improve the relevant technologies of agricultural remote sensing monitoring systems, integrate the algorithm system into the development of high-water-efficiency agricultural irrigation databases and data products, and better serve the development of my country's high-water-efficiency irrigation agriculture, so as to overcome the defects in the prior art.

[0006] The technical solution of the present invention is achieved as follows: a method for fusion of multi-source remote sensing data quantitatively inverting regional crop water consumption data, the method comprising the following steps:

[0007] S1, data collection;

[0008] S2, data assimilation;

[0009] S3, data fusion;

[0010] S4. Form a multi-source heterogeneous database.

[0011] The data collected in S1 includes fixed-point collection data, mobile collection data, drone remote sensing data and satellite remote sensing data.

[0012] In the data assimilation process in S2, a crop water consumption model is constructed by inputting in-situ detection data and crop driving data as well as observed field sampling data and ground verification data, and the crop morphology data and vegetation index data of satellite remote sensing are assimilated to invert the regional crop water consumption information.

[0013] In the data fusion process in S3, the MODIS main growth period high temporal resolution vegetation index and the Sentinel-2 high spatial resolution vegetation index are analyzed by fusion algorithm through spatiotemporal data to obtain the high spatial resolution vegetation index of the regional crop growth stage. The information after data assimilation analysis is used as a prediction variable, and regression analysis is performed in combination with a neural network model to construct a neural network regression model.

[0014] In the process of forming a multi-source heterogeneous database in S4, the fused data is applied at the regional scale to obtain the regional crop water consumption data after assimilation analysis of each pixel, and a long-sequence multimodal data processing and integration method is established to achieve regional scale daily high spatial resolution crop water consumption data, build an irrigation big data storage structure and indexing method, and form a high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems.

[0015] The fixed-point data collection is completed using soil moisture sensors and meteorological sensors.

[0016] The mobile collected data includes phenotypic data and physiological data. The phenotypic data includes leaf area index data, plant height data and root depth data, and the physiological data includes plant water content data, transpiration rate data and canopy temperature data.

[0017] The UAV remote sensing data includes multispectral data and hyperspectral data. The multispectral data includes canopy characteristic parameters, morphological characteristic parameters and vegetation index extraction parameters. The hyperspectral data includes sensitive band data, spectral index data and stress index inversion data.

[0018] The satellite remote sensing data includes optical remote sensing data and microwave remote sensing data. The optical remote sensing data includes spectrum characteristic parameters, texture characteristic parameters and vegetation index extraction parameters. The microwave remote sensing data includes temperature characteristic parameters, surface characteristic parameters and drought characteristic parameters.

[0019] The present invention has the following positive effects:

[0020] The present invention utilizes a distributed multi-source heterogeneous database built with advanced technologies such as integrated grid computing and cloud computing to construct a high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems, realize real-time collection of agricultural information, form a database and establish a database management system, and ultimately achieve efficient data management and sharing. Systematic monitoring of agricultural irrigation-related processes, including monitoring of crop water consumption, agricultural irrigation monitoring and forecasting, etc. Develop agricultural production management systems and agricultural expert systems, mainly for the diagnosis of crop water shortage, fertilizer shortage, and pests and diseases. The data collection forms are diverse, and the parameter optimization amount is large, which realizes the acquisition of a wide range of farmland parameters and forms a complete system with remote sensing irrigation monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the structure of the method steps of the present invention.

[0022] Figure 2 It is a schematic diagram of the specific framework structure of the method of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In the following description of the invention, it should be noted that the terms "upper", "lower", "left", "right", "inner", "outer", etc., indicating directions or positional relationships, are based on directions or positional relationships shown in the drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific direction or be constructed and operated in a specific direction. The term "connection" only indicates the connection between devices and has no special meaning.

[0025] Specific embodiment: Figure 1 , 2 As shown, a method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption requirements is provided, and the method comprises the following steps:

[0026] S1. Data collection: The collected data include fixed-point collection data, mobile collection data, UAV remote sensing data and satellite remote sensing data. The fixed-point collection data is completed using soil moisture sensors and meteorological sensors. The mobile collection data includes phenotypic data and physiological data. The phenotypic data includes leaf area index data, plant height data and root depth data. The physiological data includes plant water content data, transpiration rate data and canopy temperature data. The UAV remote sensing data includes multispectral data and hyperspectral data. The multispectral data includes canopy characteristic parameters, morphological characteristic parameters and vegetation index extraction parameters. The hyperspectral data includes sensitive band data, spectral index data and stress index inversion data. The satellite remote sensing data includes optical remote sensing data and microwave remote sensing data. The optical remote sensing data includes spectral characteristic parameters, texture characteristic parameters and vegetation index extraction parameters. The microwave remote sensing data includes temperature characteristic parameters, surface characteristic parameters and drought characteristic parameters.

[0027] S2. Data assimilation: A crop water consumption model is constructed by inputting in-situ detection data and crop driving data as well as observed field sampling data and ground verification data, and the crop morphology data and vegetation index data from satellite remote sensing are assimilated to invert the regional crop water consumption information.

[0028] S3. Data fusion: The high temporal resolution vegetation index of the main growth period of MODIS and the high spatial resolution vegetation index of Sentinel-2 are analyzed through spatiotemporal data by fusion algorithm to obtain the high spatial resolution vegetation index of the regional crop growth stage. The information after data assimilation analysis is used as the prediction variable, and regression analysis is performed in combination with the neural network model to construct a neural network regression model.

[0029] S4. Form a multi-source heterogeneous database: Apply the fused data at the regional scale, obtain the regional crop water consumption data after assimilation analysis of each pixel, establish a long-sequence multimodal data processing and integration method, realize the daily high-spatial resolution crop water consumption data at the regional scale, build an irrigation big data storage structure and indexing method, and form a high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems.

[0030] In the process of fusing MODIS and Sentinel-2 remote sensing images, the main goal is to combine the high temporal resolution of MODIS and the high spatial resolution of Sentinel-2 to generate vegetation indices (such as NDVI, EVI, etc.) with both high temporal resolution and high spatial resolution. Using the spatiotemporal data fusion model, the temporal and spatial information are processed separately, and the fusion of remote sensing images is achieved using the following formula:

[0031]

[0032] in:

[0033] F(x,y,t) is the fused vegetation index with high spatial and temporal resolution.

[0034] S2(x,y,t0) is the high-resolution image of Sentinel-2 at the initial time t0.

[0035] M(x,y,t) and M(x,y,t0) are MODIS images (high temporal resolution) at time t and t0 respectively.

[0036] ΔM(t) / ΔS2(t) is the rate of change in time and is used to capture the changes in MODIS and Sentinel-2 at time t.

[0037] By utilizing the frequent temporal changes of MODIS information and combining it with the high-resolution spatial data of Sentinel-2, a complementary fusion of time and space is achieved.

[0038] The present invention constructs a crop water consumption model through in-situ monitoring of the entire growth period of the main crops, assimilates drone and satellite remote sensing data, and inverts regional crop water consumption information. The ensemble Kalman filter method is used for assimilation analysis, and based on the spatiotemporal complementary advantages of multi-source remote sensing data, a data fusion method for quantitatively inverting regional crop water consumption from multi-source remote sensing data is obtained. The MODIS main growth period high temporal resolution vegetation index and the Sentinel-2 high spatial resolution vegetation index are analyzed through a spatiotemporal data fusion algorithm to obtain a data fusion product of high spatial resolution vegetation index and crop water consumption in the growth stage of regional crops. A long sequence multimodal data processing and integration method is established to realize regional scale daily high spatial resolution crop water consumption data, and by constructing an irrigation big data storage structure and indexing method based on a domestic data engine, a high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems are formed.

[0039] During the specific operation, the crop water consumption model is constructed by means of in-situ monitoring of the main crops throughout their growth period, and satellite remote sensing data is assimilated to invert the regional crop water consumption information. Through the interaction of crop growth model and crop water consumption model parameters, plant height, leaf area index, dry matter accumulation, canopy temperature and crop water consumption information are assimilated, remote sensing vegetation index information is embedded, and regional crop water consumption data is assimilated. Based on remote sensing observations and model simulation results, weighted calculations are performed and model parameters are updated to achieve posterior probability analysis of model simulation; based on the updated state parameters, the coupled model continues to run until the end of the growth period. The ensemble Kalman filter method is used for assimilation analysis, and the initial parameters are optimized and updated for different environmental conditions such as climate, soil, and vegetation. The model continues to run based on the updated parameters, and the accuracy of the model in estimating crop water consumption is verified.

[0040] In specific operation, based on the spatiotemporal complementary advantages of multi-source remote sensing data, the high temporal resolution vegetation index of the main growth period of MODIS and the high spatial resolution vegetation index of Sentinel-2 are analyzed through the spatiotemporal data fusion algorithm to obtain the high spatial resolution vegetation index of the regional crop growth stage. The soil moisture, leaf area index, dry matter mass, and plant height after dual parameter assimilation analysis are used as output variables, and remote sensing observation data such as vegetation index and backscatter coefficient and longitude and latitude information are used as prediction variables. Regression analysis is performed in combination with the CNN model to construct a neural network regression model. Combined with the ground data verification of monitoring points in different ecological zones, a data fusion method for quantitative inversion of regional crop water consumption from multi-source remote sensing data is formed.

[0041] During specific operation, based on the CNN model, regional-scale multi-source remote sensing observation data with spatiotemporal data fusion are applied at the regional scale to obtain the regional crop water consumption data after assimilation analysis of each pixel, and a long-sequence multimodal data processing and integration method is established to achieve daily high-spatial-resolution crop water consumption data at the regional scale. By constructing an irrigation big data storage structure and indexing method based on a domestic data engine, a high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems are formed.

[0042] During specific operation, 51 soil moisture monitoring points were set up in five major ecological zones, including Northeast China, North China, Northwest China, and Southwest China, providing 15,000 pieces of 0-100cm soil moisture data, 12,000 pieces of crop water requirement information, and 7,500 pieces of water consumption information as data support. Ground data verification observations during the growing season of winter wheat, summer corn, peanuts, soybeans, and cotton were carried out at the Xinxiang Comprehensive Experimental Base of the Institute of Farmland Irrigation, Chinese Academy of Agricultural Sciences. At the same time, the present invention carried out synchronous observations at the Hengshui Dryland Water-Saving Agriculture Experimental Station of the Hebei Academy of Agricultural Sciences. First, ground station observations were carried out in Xinxiang and Hengshui, and the Tianqi meteorological monitoring system and soil moisture sensors such as TDR and FDR were installed. Long-term ground real observation data are obtained to provide data support for the operation of crop growth models and water consumption models, result verification, and accuracy evaluation.

[0043] With the soil moisture and crop transpiration monitoring system deployment location as the core area, satellite-ground synchronous observations were carried out many times during the growing season of major crops such as winter wheat, summer corn, peanuts, soybeans, and cotton. 150 time series observations were carried out during the complete crop growing season. Ring knife sampling was carried out on the surface soil to observe bulk density, soil moisture, plant height, leaf area index, plant water content, plant transpiration rate, chlorophyll content, and canopy temperature.

[0044] The microwave remote sensing satellite data of the present invention uses the Level-1GRD data product based on the IW mode produced by the Sentinel-1 remote sensing satellite, and the optical remote sensing satellite data is provided by the Sentinel-2 remote sensing satellite. The spatial resolution of the band data contained in the MODIS sensor can reach up to 250m, and includes 1km resolution data. Through the comprehensive utilization of multi-source remote sensing satellite data, the spatial resolution and temporal frequency of the observation are greatly improved. The MODIS vegetation index data with high temporal frequency in the GEE platform is used as an important basis for the data fusion algorithm, providing data support for the construction of the spatiotemporal fusion algorithm of satellite remote sensing data.

[0045] The data preprocessing work in this embodiment mainly includes the collection and preprocessing of remote sensing data, water consumption data, and auxiliary data. Based on data collection and preprocessing, the crop growth characteristics zoning work is carried out. First, the area of ​​5 major crop planting areas is extracted based on medium and high resolution remote sensing images, and the spatial distribution of planting in the main crop production areas is obtained. Based on MODIS images, commonly used vegetation indices such as NDVI and EVI are calculated. Combined with the spatial distribution data of planting in the main crop production areas and crop phenology, the crop growth period time series MODIS vegetation index data set is obtained. Based on this data and combined with the spatiotemporal variation characteristics of crop growth characteristics, crop growth characteristics zoning is carried out.

[0046] On the basis of collecting and sorting out the crop water consumption and growth characteristic data and related auxiliary data, combined with time series remote sensing data, the spectral response law of the dynamic changes of crop water consumption and growth characteristic parameters is analyzed through sensitive spectral index screening, crop growth process standard curve and other methods. On this basis, according to the comprehensive comparison of product production efficiency and accuracy, the Gaussian curve fitting method, stepwise regression analysis method and other methods are used to screen out the best crop water consumption and growth characteristic monitoring model, and combined with the crop zoning results, the crop water consumption and growth characteristic monitoring model is constructed by zoning.

[0047] The multi-source remote sensing data quantitative inversion of regional crop water consumption data in the present invention mainly includes basic background data, farmland collection data, remote sensing image data and other data. The data forms mainly include vector data, attribute data, raster data, original image data, image product data, original observation data, corrected observation data, metadata and dictionary data, etc.

[0048] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for fusion of multi-source remote sensing data to quantitatively invert regional crop water consumption, characterized in that: The method comprises the following steps: S1, data collection; S2, data assimilation; S3, data fusion; S4. Form a multi-source heterogeneous database.

2. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 1 is characterized by: The data collected in S1 includes fixed-point collection data, mobile collection data, drone remote sensing data and satellite remote sensing data.

3. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 1, characterized in that: In the data assimilation process in S2, a crop water consumption model is constructed by inputting in-situ detection data and crop driving data as well as observed field sampling data and ground verification data, and the crop morphology data and vegetation index data of satellite remote sensing are assimilated to invert the regional crop water consumption information.

4. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 1, characterized in that: In the data fusion process in S3, the MODIS main growth period high temporal resolution vegetation index and the Sentinel-2 high spatial resolution vegetation index are analyzed by fusion algorithm through spatiotemporal data to obtain the high spatial resolution vegetation index of the regional crop growth stage. The information after data assimilation analysis is used as a prediction variable, and regression analysis is performed in combination with a neural network model to construct a neural network regression model.

5. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 1, characterized in that: In the process of forming a multi-source heterogeneous database in S4, the fused data is applied at the regional scale to obtain the regional crop water consumption data after assimilation analysis of each pixel, and a long-sequence multimodal data processing and integration method is established to achieve regional scale daily high spatial resolution crop water consumption data, build an irrigation big data storage structure and indexing method, and form a high-performance storage and retrieval multi-source heterogeneous database and data products for complex irrigation data systems.

6. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 2, characterized in that: The fixed-point data collection is completed using soil moisture sensors and meteorological sensors.

7. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 2, characterized in that: The mobile collected data includes phenotypic data and physiological data. The phenotypic data includes leaf area index data, plant height data and root depth data, and the physiological data includes plant water content data, transpiration rate data and canopy temperature data.

8. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 2, characterized in that: The UAV remote sensing data includes multispectral data and hyperspectral data. The multispectral data includes canopy characteristic parameters, morphological characteristic parameters and vegetation index extraction parameters. The hyperspectral data includes sensitive band data, spectral index data and stress index inversion data.

9. The method for fusion of multi-source remote sensing data for quantitative inversion of regional crop water consumption according to claim 2, characterized in that: The satellite remote sensing data includes optical remote sensing data and microwave remote sensing data. The optical remote sensing data includes spectrum characteristic parameters, texture characteristic parameters and vegetation index extraction parameters. The microwave remote sensing data includes temperature characteristic parameters, surface characteristic parameters and drought characteristic parameters.

Images