A crop drought detection method based on remote sensing images
By dividing crop growth stages based on remote sensing images and meteorological data and constructing a composite drought index, the problems of incomplete data integration and failure to consider differences in growth stages in existing technologies are solved, thus achieving efficient and accurate crop drought detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GANSU AGRI UNIV
- Filing Date
- 2026-04-17
- Publication Date
- 2026-07-14
AI Technical Summary
Existing crop drought detection technologies suffer from incomplete data integration, insufficient consideration of differences in water sensitivity at different growth stages of crops, and the need to reconstruct models for cross-crop adaptation, resulting in insufficient detection accuracy and flexibility in practical applications.
By acquiring remote sensing images, meteorological and crop basic data, we can divide crops into five key growth stages, calculate the water sensitivity index and dynamic baseline without water stress, construct a composite drought index, achieve scientific determination of drought level, and adjust only the water sensitivity index parameter when adapting to different crops.
It improves detection accuracy and crop adaptation flexibility, reduces cross-crop adaptation time, and enhances detection adaptation efficiency and accuracy.
Smart Images
Figure SMS_5 
Figure QLYQS_1 
Figure QLYQS_2
Abstract
Description
Technical Field
[0001] This invention relates to the field of agricultural drought monitoring technology, specifically a method for detecting crop drought based on remote sensing images. Background Technology
[0002] Drought monitoring of crops is a key technology for ensuring agricultural production and optimizing water resource allocation. With the development of remote sensing and agricultural information technology, multi-source data fusion and index model construction have become the core development directions in this field. Currently, multispectral and thermal infrared remote sensing images provided by satellites such as Sentinel-2 and Landsat, combined with meteorological observation data, are widely used in the calculation and monitoring of drought indices. At the same time, the water sensitivity characteristics of crops at different growth stages are gradually attracting attention and becoming an important consideration for improving the accuracy of monitoring. Various monitoring methods based on vegetation indices and temperature indices are constantly emerging, providing technical support for large-scale drought monitoring.
[0003] Existing drought detection technologies still have significant limitations, making it difficult to meet the needs of precise and universal production. For example, the published patent CN108613933A, "A Spatiotemporal Dynamic Monitoring Method for Forest Drought Based on Multi-Source Remote Sensing Data Fusion," only focuses on the fusion of MODIS and Landsat remote sensing data, without integrating basic crop data and failing to consider the differences in water sensitivity at different growth stages, leading to discrepancies between the detection results and the actual water stress state of crops. Another example is the published patent CN101790955B, "A Method and Device for Controlling Irrigation Based on Crop Water Deficiency," which uses a fixed logic for calculating the crop water stress index, failing to adapt to dynamic changes in growth stages, and requiring a reconstruction of the detection process when switching to different crops, resulting in insufficient flexibility. Furthermore, most methods suffer from limited data dimensions or complex parameter adjustments, making it difficult to balance detection accuracy with ease of practical application. Summary of the Invention
[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a crop drought detection method based on remote sensing images. This method solves the problems commonly found in existing technologies, such as incomplete data dimension integration, insufficient consideration of the differences in water sensitivity at different growth stages of crops, and the need to reconstruct models for cross-crop adaptation, which lead to insufficient detection accuracy and flexibility in practical applications.
[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a method for detecting crop drought based on remote sensing images, comprising the following steps: S1. Acquire remote sensing image data, meteorological data, and basic crop data for the entire growing season of crops in the target area; S2. Extract NDVI based on remote sensing image time series data and construct time series curves. Identify and divide the five key growth stages—seedling stage, jointing stage, heading stage, grain filling stage, and maturity stage—by identifying curve inflection points. S3. For each critical reproductive period, calculate the water sensitivity index and dynamic baseline without water stress. S4. Based on the aforementioned water sensitivity index and dynamic water-free stress baseline, construct a composite drought index; S5. Determine the drought level of crops based on the preset threshold of the compound drought index.
[0006] Preferably, the remote sensing image data in S1 includes Sentinel-2 multispectral images with red-edge / near-infrared bands and Landsat-8 thermal infrared images; meteorological data includes daily average air temperature and relative humidity; and basic crop data includes crop variety, yield data, planting density, and soil type.
[0007] Preferably, the criteria for defining the critical reproductive period in S2 are as follows: Seedling stage: NDVI 0.3-0.5, rising gradually; During the growth spurt stage: NDVI rises rapidly to above 0.7; During the heading stage: NDVI reaches its peak value of 0.8-0.9, and the curve remains stable; Grouting period: NDVI 0.6-0.8, decreasing slowly; Maturity stage: NDVI rapidly drops below 0.3.
[0008] Preferably, in step S3, the moisture sensitivity index is calculated using the Stewart water production function model: , where Y is the actual crop yield, Ym is the potential crop yield, λ_i is the water sensitivity index of the i-th growth stage, E_i is the actual evapotranspiration of the i-th growth stage, E_m is the potential evapotranspiration of the i-th growth stage, and n is the number of critical growth stages.
[0009] Preferably, the CWSI-WS calculation formula in S4 is: Where CWSI-WS is the dynamic adaptive composite drought index for the growing season, λ_i is the water sensitivity index for the i-th growing season, Tmin,i is the canopy temperature without water stress in the i-th growing season, Tmax,i is the canopy temperature under extreme drought in the i-th growing season, Tc,i is the actual canopy temperature in the i-th growing season, and i is the growing season number.
[0010] Preferably, the drought level classification in S5 is: no drought, mild drought, moderate drought, and severe drought.
[0011] Preferably, the crops include wheat, corn, and peanuts, and when adapting to different crops, only the water sensitivity index parameter at each growth stage is adjusted.
[0012] (III) Beneficial Effects This invention provides a method for detecting crop drought based on remote sensing images. It has the following beneficial effects: 1. This invention integrates three types of data: remote sensing images, meteorological data, and crop data. By using NDVI time-series curve inflection point identification, it accurately divides the five key growth stages of crops. Combined with the Stewart water production function model, it calculates the water sensitivity index for each growth stage and constructs a composite drought index that integrates a dynamic baseline without water stress and the weight of growth stage requirements, thus enabling scientific determination of drought levels. At the same time, it can adapt to different crops by only adjusting the water sensitivity index parameters. This not only solves the detection bias problem caused by existing technologies relying on single data and ignoring the differences in water requirements during growth stages, but also improves the flexibility of crop adaptation and detection adaptability. Detailed Implementation
[0013] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0014] Example 1: This invention provides a method for detecting crop drought based on remote sensing images, comprising the following steps: S1. Acquire remote sensing image data, meteorological data, and basic crop data for the entire growing season of crops in the target area; S2. Extract NDVI based on remote sensing image time series data and construct time series curves. Identify and divide the five key growth stages—seedling stage, jointing stage, heading stage, grain filling stage, and maturity stage—by identifying curve inflection points. S3. For each critical reproductive period, calculate the water sensitivity index and dynamic baseline without water stress. S4. Based on the aforementioned water sensitivity index and dynamic water-free stress baseline, construct a composite drought index; S5. Determine the drought level of crops based on the preset threshold of the compound drought index.
[0015] The remote sensing image data in S1 includes Sentinel-2 multispectral images with red-edge / near-infrared bands and Landsat-8 thermal infrared images; meteorological data includes daily average air temperature and relative humidity; and basic crop data includes crop varieties, yield data, planting density, and soil type.
[0016] S2 Critical Fertility Period Delineation Criteria: Seedling stage: NDVI 0.3-0.5, rising gradually; During the growth spurt stage: NDVI rises rapidly to above 0.7; During the heading stage: NDVI reaches its peak value of 0.8-0.9, and the curve remains stable; Grouting period: NDVI 0.6-0.8, decreasing slowly; Maturity stage: NDVI rapidly drops below 0.3.
[0017] In S3, the moisture sensitivity index is calculated using the Stewart water production function model: , where Y is the actual crop yield, Ym is the potential crop yield, λ_i is the water sensitivity index of the i-th growth stage, E_i is the actual evapotranspiration of the i-th growth stage, E_m is the potential evapotranspiration of the i-th growth stage, and n is the number of critical growth stages.
[0018] The formula for calculating CWSI-WS in S4 is: Where CWSI-WS is the dynamic adaptive composite drought index for the growing season, λ_i is the water sensitivity index for the i-th growing season, Tmin,i is the canopy temperature without water stress in the i-th growing season, Tmax,i is the canopy temperature under extreme drought in the i-th growing season, Tc,i is the actual canopy temperature in the i-th growing season, and i is the growing season number.
[0019] The drought levels in S5 are classified as: no drought, mild drought, moderate drought, and severe drought.
[0020] Crops include wheat, corn, and peanuts. When adapting to different crops, only the water sensitivity index parameters at each growth stage are adjusted.
[0021] Comparative Example 1: Acquire multispectral remote sensing images of the target area during the crop growing season and extract NDVI time-series data; The NDVI value at the heading stage (a critical period of crop growth) was selected as the core detection indicator, and a fixed threshold was preset. Drought level is determined directly based on threshold values: no drought (NDVI>0.8), mild drought (0.7-0.8), moderate drought (0.6-0.7), and severe drought (<0.6). It does not distinguish between other growth stages of crops, nor does it incorporate meteorological data and basic crop data.
[0022] Comparative Example 2: Acquire thermal infrared remote sensing images and meteorological data of crops during the growing season, and retrieve the average canopy temperature throughout the entire growth period; Establish fixed baselines for water stress (mean canopy temperature over the entire growth period) and extreme drought; The single crop water stress index (CWSI) is calculated based on a fixed baseline, without considering the differences in water sensitivity at different growth stages. The drought level threshold is the same as that used in the previous example, and the drought level is determined directly by the CWSI value.
[0023] Comparative Example 3: Collect long-term meteorological data (historical data on precipitation, temperature, evaporation, and soil moisture content) for the target area. The Palmer Drought Index (PDSI) is calculated based on meteorological data, and its core reliance is on the balance between precipitation and evaporation. It does not incorporate remote sensing image data and does not distinguish crop growth stage characteristics; Drought levels are determined based on preset PDSI thresholds: no drought (PDSI>-0.5), mild drought (-1.5 to-0.5), moderate drought (-3.0 to-1.5), and severe drought (<-3.0).
[0024] Experimental example: 1. Experimental Design Unified experimental setting: Select the same target area (temperate monsoon climate, loam soil type), and use wheat, corn and peanut as the test objects to cover the entire growing season (wheat: October to June of the following year; corn: June to October; peanut: May to September).
[0025] Unified judgment criteria: drought levels are divided into four categories: no drought, mild drought, moderate drought, and severe drought; data sources are unified (remote sensing images and meteorological data are all from the same official data source, and basic crop data are collected uniformly in the field).
[0026] Evaluation indicators: Three core quantitative indicators were selected: detection accuracy (the degree of fit between the judgment result and the actual water stress state), growth period adaptation fit (the degree of matching of the method with the differences in water requirements of crops during key growth periods), and crop adaptation time (the time required for parameter adjustment / model reconstruction to adapt to different crops).
[0027] 2. Experimental Procedure Data preparation: Collect Sentinel-2 multispectral images, Landsat-8 thermal infrared images, daily average temperature / relative humidity meteorological data, and basic data such as crop varieties, planting density, and soil type for the three crops in the target area during the complete growing season. Perform unified preprocessing on all data (radiometric correction, outlier removal, and interpolation completion).
[0028] 3. Method execution: According to the method flow of the embodiment (the present invention), data extraction, growth period division, parameter calculation, composite index construction, and drought level determination are completed in sequence. When adapting to three crops, only the water sensitivity index parameter is adjusted.
[0029] Following the core steps of Comparative Example 1 (single NDVI index method), Comparative Example 2 (traditional fixed baseline CWSI method), and Comparative Example 3 (Pamer drought index method), the drought level of three crops was determined independently, strictly adhering to the original technical logic of each method.
[0030] 4. Result Collection: Detection accuracy: The actual soil moisture content and crop wilting status were obtained through field sampling (30 sampling points / crop) as the true values, and the matching ratio between the judgment results of each method and the true values was calculated.
[0031] Growth period fit: Compare the matching degree between the determination results of the five key growth periods of each method and the actual water-sensitive stages of crops, and quantify them according to the fit ratio.
[0032] Crop adaptation time: Record the operation time (in hours) required for each method to adapt from wheat to corn and peanut.
[0033] Data summary: The average value of the results of various indicators for the three crops was taken as the final experimental data.
[0034] The specific data is shown in Table 1 below: Table 1 In summary: Detection accuracy: The present invention (92.6%) is 25.3%, 18.1% and 22.9% higher than comparative examples one, two and three, respectively. This is because it integrates remote sensing, meteorological and crop data and dynamic baselines, avoiding the bias of single data or fixed thresholds.
[0035] The compatibility of the reproductive period: This invention (94.8%) is significantly higher than the other pairs. The core reason is that it accurately divides the five key reproductive periods and matches them with water-sensitive weights, which is in line with the growth pattern of the organism.
[0036] Crop adaptation time: This invention only requires 1.5 hours to complete cross-crop adaptation, which is much shorter than the average pair (6.8-9.5 hours). Since only the water sensitivity index parameter needs to be adjusted, there is no need to reconstruct the model, and the adaptation efficiency is greatly improved.
[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for detecting crop drought based on remote sensing images, characterized in that, Includes the following steps: S1. Acquire remote sensing image data, meteorological data, and basic crop data for the entire growing season of crops in the target area; S2. Extract NDVI based on remote sensing image time series data and construct time series curves. Identify and divide the five key growth stages—seedling stage, jointing stage, heading stage, grain filling stage, and maturity stage—by identifying curve inflection points. S3. For each critical reproductive period, calculate the water sensitivity index and dynamic baseline without water stress. S4. Based on the aforementioned water sensitivity index and dynamic water-free stress baseline, construct a composite drought index; S5. Determine the drought level of crops based on the preset threshold of the compound drought index.
2. The method for detecting crop drought based on remote sensing images according to claim 1, characterized in that: The remote sensing image data in S1 includes Sentinel-2 multispectral images with red-edge / near-infrared bands and Landsat-8 thermal infrared images; meteorological data includes daily average air temperature and relative humidity; and basic crop data includes crop varieties, yield data, planting density, and soil type.
3. The method for detecting crop drought based on remote sensing images according to claim 1, characterized in that: The criteria for defining the critical reproductive period in S2 are as follows: Seedling stage: NDVI 0.3-0.5, rising gradually; During the growth spurt stage: NDVI rises rapidly to above 0.7; During the heading stage: NDVI reaches its peak value of 0.8-0.9, and the curve remains stable; Grouting period: NDVI 0.6-0.8, decreasing slowly; Maturity stage: NDVI rapidly drops below 0.
3.
4. The method for detecting crop drought based on remote sensing images according to claim 1, characterized in that: In S3, the moisture sensitivity index is calculated using the Stewart water production function model: , where Y is the actual crop yield, Ym is the potential crop yield, λ_i is the water sensitivity index of the i-th growth stage, E_i is the actual evapotranspiration of the i-th growth stage, E_m is the potential evapotranspiration of the i-th growth stage, and n is the number of critical growth stages.
5. The method for detecting crop drought based on remote sensing images according to claim 1, characterized in that: The CWSI-WS calculation formula in S4 is as follows: Where CWSI-WS is the dynamic adaptive composite drought index for the growing season, λ_i is the water sensitivity index for the i-th growing season, Tmin,i is the canopy temperature without water stress in the i-th growing season, Tmax,i is the canopy temperature under extreme drought in the i-th growing season, Tc,i is the actual canopy temperature in the i-th growing season, and i is the growing season number.
6. The method for detecting crop drought based on remote sensing images according to claim 1, characterized in that: The drought levels in S5 are classified as: no drought, mild drought, moderate drought, and severe drought.
7. The method for detecting crop drought based on remote sensing images according to claim 1, characterized in that: The crops mentioned include wheat, corn, and peanuts. When adapting to different crops, only the water sensitivity index parameters at each growth stage are adjusted.