Method for fine classification of drought grades during corn growth period in northeast China

Abstract

This application provides a method for finely classifying drought levels during the maize growing season in Northeast China. The method includes: using the ESTARFM enhanced spatiotemporal adaptive reflectivity fusion model on the Google Earth Engine platform, fusing Landsat and MODIS data to generate NDVI and LST time-series datasets with a spatial resolution of 30 meters and a temporal resolution of 8 days; constructing the NDVI-LST feature space inversion temperature vegetation drought index (TVDI); verifying accuracy by combining the relative humidity of soil at 0-10 cm depth, ensuring that the TVDI determination coefficient R² is higher than 0.61; dividing the maize growing season into five key stages: sowing-emergence, emergence-jointing, jointing-tasseling, tasseling-milk ripening, and milk ripening-maturity; establishing a staged TVDI dynamic threshold system based on the differences in water sensitivity at each growing season; determining drought conditions pixel-by-pixel according to five levels: moist, normal, mild drought, moderate drought, and severe drought; and generating a spatial distribution map of drought levels during the growing season in maize planting areas of Jilin Province, achieving fine spatiotemporal identification and visualization of drought conditions. This invention aligns with the water requirements of maize during its growth cycle, significantly improving the accuracy of drought level assessment. It can precisely identify high-drought areas in the main maize-producing regions of central and western China, as well as critical risk periods from sowing to emergence and from tasseling to milk stage, providing a scientific basis for precise drought resistance, irrigation regulation, and disaster assessment.

Description

Technical Field

[0001] This application relates to the field of agricultural meteorological disaster risk assessment technology, and in particular to a method for fine classification of drought levels during the maize growing season in Northeast China based on spatiotemporal remote sensing fusion and high spatiotemporal resolution TVDI. Background Technology

[0002] With the intensification of global climate change and the frequent occurrence of extreme drought events, agricultural drought has become the primary meteorological disaster restricting high and stable corn yields. Therefore, conducting precise, efficient, and quantitative agricultural drought monitoring and grading has become an urgent need to ensure food security and sustainable agricultural development. Taking Jilin Province in Northeast China as an example, Jilin is located in the core area of ​​the world-renowned "Golden Corn Belt" and is a crucial commodity grain base in my country. However, influenced by the temperate continental monsoon climate, precipitation is unevenly distributed in time and space, leading to frequent droughts. The main corn-producing areas in the central and western regions are constantly threatened by spring droughts, consecutive spring-summer droughts, and autumn droughts.

[0003] Maize exhibits significantly different sensitivities to water stress at different growth stages. The sowing-emergence stage directly impacts seedling emergence and survival, while the tasseling-milk stage directly determines yield formation, representing a critical period for water management. Drought at different stages can have drastically different effects. Existing drought classification methods mostly employ uniform, fixed thresholds without dynamically adjusting to the water requirements of maize throughout its growth cycle. This leads to discrepancies between drought assessments and actual crop conditions, resulting in large classification biases, poor applicability, and difficulty in supporting phased irrigation, drought early warning, and yield assurance.

[0004] Satellite remote sensing technology possesses advantages such as macroscopic, rapid, dynamic, and large-scale monitoring, and has been widely applied in agricultural drought monitoring. However, existing technologies have significant drawbacks: a single data source cannot simultaneously address spatiotemporal resolution; Landsat offers high spatial accuracy but has a long revisit period and is susceptible to cloud contamination, making it difficult to capture rapid changes in drought conditions during the growing season; MODIS offers high temporal resolution but has a coarse spatial scale, failing to achieve refined drought assessment at the plot level. Furthermore, existing methods often use a uniform TVDI threshold to classify drought levels, neglecting differences in soil moisture, crop physiological water requirements, and climatic background at different growing stages. This leads to inaccurate classifications of mild, moderate, and severe drought, poor consistency with ground-measured soil moisture, and an inability to accurately reflect the drought severity of maize at various stages.

[0005] Furthermore, traditional drought classification lacks standardized, automated, and operational processes, resulting in low processing efficiency and coarse spatial representation, making it impossible to achieve dynamic mapping and quantitative evaluation of drought levels over long time periods, large areas, and different growth stages. Therefore, how to establish a refined drought classification method that conforms to the growth cycle of maize, is based on high spatiotemporal resolution data, and employs dynamic thresholds has become a key technical problem that urgently needs to be solved in the field of agricultural drought remote sensing monitoring. Summary of the Invention

[0006] The purpose of this application is to provide a method for finely classifying drought levels during the growing season of maize in Northeast China, which solves the problems of traditional drought level classification using a uniform fixed threshold, not conforming to the water demand pattern of the growing season, insufficient spatiotemporal resolution, low judgment accuracy, and inability to achieve fine-grained drought identification at the plot level, so as to achieve accurate, dynamic, fine-grained classification and spatial expression of drought levels throughout the entire growing season of maize.

[0007] To achieve the above objectives, this application provides a method for finely classifying drought levels during the maize growing season in Northeast China. This method includes:

[0008] The data were collected using the GEE platform, including Landsat and MODIS series remote sensing data, NOAA meteorological data, ground-based measured data of soil relative humidity in the 0-10cm depth, and data on maize planting areas and growth stages.

[0009] The remote sensing data is preprocessed by radiometric calibration, atmospheric correction, cloud masking, cloud removal, resampling to 30 meters, and unified projection coordinate system.

[0010] The ESTARFM enhanced spatiotemporal adaptive reflectivity fusion model was used to fuse Landsat and MODIS data to generate NDVI and LST time series datasets with a spatial resolution of 30 meters and a temporal resolution of 8 days.

[0011] A triangular feature space of NDVI-LST was constructed, and the high spatiotemporal consistency temperature vegetation drought index (TVDI) was obtained by inversion. The accuracy was verified using the relative humidity of soil at 0-10 cm depth, ensuring the accuracy of the determination coefficient R. 2 All are above 0.61;

[0012] The growth period of corn is strictly divided into five key stages: sowing-emergence, emergence-jointing, jointing-tasseling, tasseling-milk ripening, and milk ripening-maturity.

[0013] A TVDI dynamic drought threshold system was established according to the growth period, and the drought situation was determined pixel by pixel according to the five-level standard of wet, normal, mild drought, moderate drought and severe drought.

[0014] Complete spatial mapping and statistical analysis of drought levels at each growth stage, and output a spatial distribution map of drought levels at each growth stage in maize planting areas.

[0015] The beneficial effects achieved by this application are as follows:

[0016] (1) Based on the ESTARFM algorithm, multi-source remote sensing data are fused to generate a 30-meter / 8-day high spatiotemporal resolution TVDI dataset. The determination coefficient R² > 0.61 with soil relative humidity is used to realize high-precision dynamic monitoring of drought during the maize growth period. The drought level is refined according to the five key growth stages of maize, which conforms to the physiological water demand pattern of maize and accurately reflects the drought stress characteristics of different growth stages. With TVDI as the core drought index, a five-level standardized classification system of "humid, normal, mild drought, moderate drought and severe drought" is established, and the evaluation results are unified and comparable. The spatial expression of pixel-level drought level can accurately identify the drought distribution pattern in different regions and different growth stages, and provide a scientific basis for precision irrigation and regional drought resistance. The method is highly versatile and applicable to maize planting areas in different climate zones. It has a wide range of applications, high portability, and strong promotion and application value. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart illustrating a method for refining drought severity levels during the maize growing season in Northeast China, as described in this application.

[0019] Figure 2 This is a framework diagram of a method for refining drought levels during the maize growing season in Northeast China, as proposed in this application.

[0020] Figure 3 This is a drought level map of Jilin Province from late April to September 2023, representing an embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0022] like Figure 1 and Figure 2 As shown, this application proposes a dynamic assessment method for drought disaster risk in maize based on spatiotemporal data fusion. The method includes the following steps:

[0023] Step S1: Collect research data.

[0024] The study area was determined to be the entire Jilin Province. The administrative boundary vector data of Jilin Province (shp format) was imported and named roi for subsequent data cropping and spatial statistics.

[0025] Import the following dataset into the GEE platform:

[0026] Landsat series data: LANDSAT / LE07 / C02 / T1_L2, LANDSAT / LC08 / C02 / T1_L2, selected from maize growing season (late April to September) from 2009 to 2023.

[0027] MODIS series data: MODIS / 061 / MOD09A1 (8-day composite surface reflectance), MODIS / 061 / MOD11A2 (8-day composite surface temperature).

[0028] Nighttime light data: NOAA / VIIRS nighttime light imagery.

[0029] Meteorological data: Daily precipitation and temperature data from NOAA NCEI stations, interpolated to a 1km grid.

[0030] Socioeconomic data: Jilin Provincial Statistical Yearbook (corn yield per unit area, irrigated area, GDP, agricultural machinery power, etc.), spatial distribution data of corn planting areas (30m resolution).

[0031] Step S2: Preprocess the collected data.

[0032] The remote sensing data underwent radiometric calibration, atmospheric correction, cloud masking, resampling to 30 meters, and unified projection coordinate system preprocessing; meteorological data were interpolated to a 1-kilometer resolution raster using the inverse distance weighting method; and statistical data were spatially unfolded into a 30-meter resolution raster based on the administrative center point.

[0033] Step S3: Spatiotemporal data fusion based on the ESTARFM algorithm.

[0034] Nietupski et al. shared spatiotemporal fusion Python code on GitHub. The code follows many of the same theories and assumptions as image fusion using the ESTARFM algorithm, but differs in some details. This paper is based on the GEE platform and refers to Nietupski et al.'s code, modifying the correlation coefficients and dates to make it usable for generating high spatiotemporal resolution NDVI and LST.

[0035] Step S3 includes the following sub-steps:

[0036] Step S310: Find similar pixels.

[0037] Specifically, within a sliding window, pixels similar to the center pixel are searched, and the similarity is defined as:

[0038]

[0039] Where, F(x) i ,y i F(x, t0) is the i-th pixel within the w×w pixel search window of the fine-resolution image at time t0. w / 2 ,y w / 2 ,t0) is the center pixel in the same region at time t0, σ(P) is the standard deviation of pixel values ​​in a given region, and n is the number of land use types.

[0040] Step S320: Calculate the weights and coefficients.

[0041] Specifically, if a pixel meets the similarity condition, it will be classified as a homogeneous pixel with a spectrum similar to the center pixel. The weights and regression coefficients are calculated using the center pixel, as shown in the following formula:

[0042]

[0043]

[0044]

[0045] Among them, W i D represents the weight of the i-th similar pixel; i It is an index that combines spectral and spatial similarity; d i R is the distance between the i-th pixel and the center pixel within the window; i It is the spectral correlation coefficient between the fine-resolution and coarse-resolution pixels of the i-th similar pixel.

[0046] Step S330: Predict fine-resolution pixels.

[0047] Specifically, the difference between the values ​​of similar pixels and coarse-resolution data is used to predict the center pixel at another time t. p Value:

[0048]

[0049] Where N is the number of similar pixels within the search window; W i V represents the weight of the i-th similar pixel; i It is the regression coefficient between the selected fine-resolution and coarse-resolution pixels; C(x) i ,y i ) is the location of the coarse-resolution image (x i ,y i The pixel at ().

[0050] Step S340: Calculate the time weights of the two prediction results.

[0051]

[0052] Step S350: Fuse the prediction results.

[0053] Specifically, the two prediction results are weighted and combined according to time weights to obtain the final prediction value:

[0054]

[0055] Among them, T k (k∈(m,n)) is the time weight of a single prediction.

[0056] Step S4: High spatiotemporal resolution TVDI drought index inversion.

[0057] Sandholt et al. discovered multiple straight lines in the vegetation index-land surface temperature characteristic space, and based on this, proposed the Temperature Vegetation Drought Index (TVDI) to monitor land surface humidity.

[0058] Step S4 includes the following sub-steps:

[0059] Step S410: NDVI calculation based on Landsat and MODIS data.

[0060] Specifically, NDVI measures vegetation density and health by utilizing the reflection and absorption characteristics of different wavelengths of solar radiation by plant chlorophyll. Therefore, the NDVI index is widely used for regional drought monitoring. The formula for calculating NDVI is as follows:

[0061]

[0062] in, and These represent the reflectance values ​​in the near-infrared and infrared bands, respectively. NDVI values ​​range from -1 to 1. A high NDVI value (close to +1) indicates healthy, dense green vegetation; a low NDVI value (close to 0 or -1) indicates less vegetation and bare land or water.

[0063] Step S420: Calculate the surface temperature based on Landsat and MODIS data.

[0064] Specifically, based on the Google Earth Engine (GEE) platform, first load the MOD11A2 dataset from MODIS and select the "LST_Day_1km" band, then load the Landsat 8 Collection 2 Level 2 dataset and select the b10 band (thermal infrared band) to calculate the land surface temperature data. The specific steps and formulas are as follows:

[0065] First, the digital designation (DN) value of the thermal infrared band needs to be converted into a radiance value (L):

[0066]

[0067] Where L is the radiance value (unit: W / m2·sr·μm). and These are the radiometric calibration gain (approximately 3.3420E-04 W / m²·sr·μm·DN) and offset (approximately 0.1 W / m²·sr·μm) obtained from Landsat8 metadata, respectively, where Q is the DN value in the thermal infrared band.

[0068] Finally, the surface temperature was calculated using a single-channel algorithm. The formula is as follows:

[0069]

[0070] Where LST is the surface temperature (unit: degrees Celsius), and K1 and K2 are calibration constants for the thermal infrared band obtained from Landsat8 metadata (set to 607.76 W / m^2·sr·μm and 1260.56 K, respectively).

[0071] Step S430, high spatiotemporal resolution TVDI inversion.

[0072] Specifically, the calculation formula is as follows:

[0073]

[0074]

[0075]

[0076] In the formula: It refers to the surface temperature; The minimum surface temperature under the same NDVI conditions, corresponding to NDVI— Wet edges in feature space; The maximum surface temperature under the same NDVI conditions, corresponding to NDVI- Dry edges in the feature space. NDVI = 1 on dry edges and NDVI = 0 on wet edges; a, b, c, and d are coefficients obtained by fitting remote sensing data. For each pixel, NDVI is used to determine the coefficients based on T. s In T s The position of the NDVI trapezoid is used to calculate the TVDI. The larger the TVDI, the lower the soil moisture; the smaller the TVDI, the higher the soil moisture.

[0077] Step S440, TVDI fusion accuracy verification.

[0078] Specifically, there is a significant correlation between TVDI and soil moisture (SM), especially for soil relative humidity in the 0-10cm range. Therefore, we used SM data in the 0-10cm range to verify the accuracy of TVDI obtained by fusing NDVI and LST using the ESTARFM algorithm.

[0079] Step S5: Divide the five key growth stages of maize.

[0080] Specifically, the Food and Agriculture Organization of the United Nations (FAO) recommends dividing the maize growth period into the initial growth stage, development stage, mid-growth stage, and late-growth stage. Based on the growth characteristics of maize in Northeast China, the start and end times of each growth stage for maize in Jilin Province have been determined as follows:

[0081] Sowing-emergence: Late April to mid-May;

[0082] Emergence to jointing: Late May to mid-June;

[0083] Jointing-heading: Late June to mid-July;

[0084] Heading to milk stage: Late July to mid-August;

[0085] Milk ripening to maturity: Late August to late September.

[0086] Step S6: Calculate the TVDI index by reproductive period.

[0087] Specifically, based on the time range of the corn growth period, TVDI data for the corresponding time period is extracted, and the average TVDI value for each growth period is calculated.

[0088] Step S7: Refine the drought level based on the threshold.

[0089] Specifically, due to differences in climatic conditions across different regions, the methods for classifying drought levels often vary depending on the study area. Qi Shuhua et al., based on soil relative humidity data (0-10 cm) observed at meteorological stations nationwide, verified the correlation between this data and the TVDI (Total Volume Index) during the same period. The results showed that TVDI effectively indicates the surface soil moisture status. Based on this, they classified drought conditions into five levels: moist, normal, mild drought, moderate drought, and severe drought, and applied this classification to drought monitoring and mapping during the crop growing season nationwide. This study follows this classification standard, dividing the drought status during the maize growing season in Jilin Province into five levels: moist (TVDI value between 0 and 0.2), normal (TVDI value between 0.2 and 0.4), mild drought (TVDI value between 0.4 and 0.6), moderate drought (TVDI value between 0.6 and 0.8), and severe drought (TVDI value between 0.8 and 1).

[0090] Step S8: Output the spatial distribution data of drought levels during the maize growing season.

[0091] Specifically, raster data and spatial distribution maps of drought levels for each growth stage are generated for drought monitoring and drought relief decision-making.

[0092] As a specific embodiment of the present invention, the invention is implemented on the GEE (Google Earth Engine) platform, a comprehensive platform integrating scientific analysis and geographic information data visualization. This platform provides rich APIs and tools, facilitating the viewing, calculation, processing, and analysis of large-scale GIS data, including various images. The specific implementation method of the present invention is as follows:

[0093] Step T1: Import the collected remote sensing data of the study area into the gee platform.

[0094] Specifically, import the remote sensing data of the study area into the GEE platform. The remote sensing data is a shapefile named roi. Import the collected remote sensing data of the study area into the GEE platform for subsequent cropping of the study area.

[0095] Step T2: Import the cloud removal function on the gee platform.

[0096] Specifically, the cloud removal function is used to remove clouds from the remote sensing images to be processed, so as to avoid cloud cover causing distortion of data of corn drought-related assessment indicators (such as vegetation index and surface temperature), ensure the accuracy of subsequent assessment results, and meet the technical specifications for corn drought disaster risk assessment.

[0097] Step T3: Select a dataset from the remote sensing data of the study area.

[0098] As a specific embodiment of the present invention, the dataset selected is data from the maize growth period (late April to September) from 2009 to 2023. Six typical years were selected: 2009, 2012, 2015, 2018, 2021, and 2023, covering five key growth stages of maize: sowing-emergence, emergence-jointing, jointing-tasseling, tasseling-milk ripening, and milk ripening-maturity. Images with cloud cover of less than 15% were selected, and after cloud removal, resampling and a unified projection were defined. The unified spatial resolution was 30 meters, and the temporal resolution was 8 days.

[0099] Step T4: Input the Landsat and MODIS remote sensing images into the OL-STARFM model to generate a highly spatiotemporally fused image.

[0100] Among them, the OL-STARFM model is an existing spatiotemporal fusion model used to generate 8-day, 30-meter resolution NDVI and LST time series data.

[0101] Step T5: Calculate the NDVI and LST time-series datasets based on the fused image and construct the NDVI-LST feature space.

[0102] Step T6: Import the normalization function into the gee platform.

[0103] Specifically, the normalization function is used to perform normalization calculations using a maximum-minimum normalization model. The maximum-minimum normalization model is used to standardize NDVI, LST, TVDI, and other risk assessment-related indicators.

[0104] Step T7: Perform the calculation of the heat index LST to obtain the heat index LST.

[0105] Step T8: Construct the NDVI-LST feature space and invert the Temperature Vegetation Drought Index (TVDI) to characterize drought intensity.

[0106] Step T9: Perform the calculation of the greenness index NDVI to obtain the NDVI time series data.

[0107] Step T10: Enter basic information on the gee platform.

[0108] Specifically, the basic information includes the study area, image bands, and spatial resolution.

[0109] Step T11: Rename the image bands in the basic information according to the image band renaming function.

[0110] Step T12: Export the analysis results data.

[0111] Specifically, export the analysis results data on the GEE platform.

[0112] This invention presents a refined method for classifying drought levels during the maize growing season in Northeast China. This method closely integrates the water requirement patterns of maize during its growing season to determine drought levels, accurately reflecting the spatiotemporal distribution characteristics of drought during this period. Compared to traditional drought monitoring and classification methods, this method utilizes high spatiotemporal resolution remote sensing data to achieve pixel-by-pixel dynamic monitoring. It offers convenient data acquisition, a clear calculation process, refined spatial representation, and high degree of result visualization, making it more suitable for the quantitative identification and spatial display of drought at different maize growing seasons in a region.

[0113] To verify the applicability and reliability of the method, the entire maize-growing area of ​​Jilin Province was selected as the study area. The maize growth period (late April to September) from 2009 to 2023 was chosen, with six typical years (2009, 2012, 2015, 2018, 2021, and 2023) as representatives. The study covered five key growth stages: sowing-emergence, emergence-jointing, jointing-tasseling, tasseling-milk ripening, and milk ripening-maturity. A refined classification and verification of drought levels during the maize growth period was conducted. Figure 3 ).

[0114] This application also provides a computer storage medium storing computer instructions, which, when invoked, execute the aforementioned corn drought disaster risk assessment method. The computer storage medium includes one or more program instructions, which are executed by a processor to perform a calculation method for corn drought disaster risk assessment.

[0115] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions that, when executed on a computer, cause the computer to perform the aforementioned calculation method for assessing drought disaster risk in maize.

[0116] This invention provides a processor for processing the above-described calculation method for drought disaster risk assessment of maize.

[0117] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0118] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0119] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0120] Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).

[0121] The beneficial effects achieved by this application are as follows:

[0122] (1) The method for finely classifying drought levels during the growth period of maize in Northeast China constructed in this invention closely follows the water stress response pattern of maize at each growth stage and can accurately reflect the spatiotemporal distribution characteristics of drought levels during the maize growth period. Compared with traditional drought monitoring and classification methods, this method realizes pixel-by-pixel dynamic monitoring and level determination based on high spatiotemporal resolution remote sensing data. It is convenient for data acquisition, has a clear calculation process, fine spatial expression, and high degree of result visualization, and is more suitable for fine discrimination and spatial display of drought levels at different growth stages of maize in the region.

[0123] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0124] In the description of this application, the word "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0125] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for fine classification of drought grades during the growth period of northeast corn, characterized in that, Includes the following steps: Landsat and MODIS series remote sensing data, as well as NOAA meteorological data, 0-10cm soil relative humidity data, and maize planting area and population density data were collected based on the GEE platform. The images were preprocessed by radiometric calibration, atmospheric correction, cloud masking, resampling to 30 meters, and unifying the projection coordinate system. The ESTARFM algorithm was used to fuse Landsat and MODIS data to generate NDVI and LST time series datasets with a spatial resolution of 30 meters and a temporal resolution of 8 days. Construct the NDVI-LST triangular feature space, invert the TVDI temperature vegetation drought index, and verify it using the relative humidity of 0-10cm soil to ensure that the determination coefficient R² > 0.61; The growth period of maize is divided into five stages: sowing-emergence, emergence-jointing, jointing-tasseling, tasseling-milk ripening, and milk ripening-maturity. A dynamic threshold system for TVDI was established based on the reproductive stage, and drought conditions were classified into five levels: wet, normal, mild drought, moderate drought, and severe drought. Based on the dynamic threshold of each growth stage, the drought level is determined pixel by pixel, and a spatial distribution map of drought level during the growth stage of maize in Northeast China is generated.

2. The method according to claim 1, wherein, The formula for calculating TVDI is: Where T s For pixel surface temperature, The lowest temperature for fitting the wet edge. The highest temperature for dry edge fitting.

3. The method for refined classification of drought levels during the maize growing season in Northeast China according to claim 1, characterized in that, The criteria for classifying drought levels are as follows: Humidity: 0 ≤ TVDI < 0.2; Normal: 0.2 ≤ TVDI < 0.4; Mild drought: 0.4 ≤ TVDI < 0.6; Moderate drought: 0.6 ≤ TVDI < 0.8; Severe drought: 0.8≤TVDI≤1.

4. The method for refined classification of drought levels during the maize growing season in Northeast China according to claim 1, characterized in that, The start and end dates of each reproductive period are as follows: Sowing-emergence: Late April to mid-May; Emergence to jointing: Late May to mid-June; Jointing-heading: Late June to mid-July; Heading to milk stage: Late July to mid-August; Milk ripening to maturity: Late August to late September.

5. The method for refined classification of drought levels during the maize growing season in Northeast China according to claim 1, characterized in that, TVDI showed a significant negative correlation with the relative humidity of the 0-10cm soil layer, with a determination coefficient R² higher than 0.61, indicating that it could stably reflect the surface soil moisture status.

6. The method for refined classification of drought levels during the maize growing season in Northeast China according to claim 1, characterized in that, This method is applicable to corn-growing areas in different climate zones and terrain conditions in Northeast China, and has wide applicability.

Images