Remote sensing identification method for cotton field planting mode distribution state

Abstract

The invention discloses a remote sensing identification method for a cotton field planting mode distribution state, and belongs to the technical field of satellite remote sensing images, and the method comprises the steps: extracting a first cotton field feature remote sensing image of a specific region in a first identification period based on a first cotton field identification data source; classifying farmland in a specific area in the first cotton field feature remote sensing image based on an iterative self-organizing data analysis technology to obtain a suspected cotton field area; the accuracy of the suspected cotton field area is judged based on the actual cotton sowing area provided in the statistical yearbook, and the accurate cotton field area is output; extracting a first-moment cotton field feature remote sensing image and a second-moment cotton field feature remote sensing image of the cotton field area in the second identification period based on the second cotton field identification data source; calculating a normalized water body index difference value between the second moment and the first moment in the second identification period, and identifying a winter and spring irrigation area in the cotton field area; based on the distribution difference and the area difference between the cotton field area and the winter and spring irrigation area, the distribution and the area of the dry-sowing wet-out cotton field are obtained.

Description

Technical Field

[0001] This invention relates to the field of satellite remote sensing image technology, and in particular to a remote sensing identification method for the distribution of cotton field planting patterns. Background Technology

[0002] Cotton, as an important economic crop, is crucial for improving people's livelihoods and agricultural development through sustainable and efficient production. my country is the world's largest producer and consumer of cotton, with over 90% of its cotton produced in the arid inland region of Xinjiang. Over the past two decades, Xinjiang has widely adopted drip irrigation under mulch film to improve water resource utilization efficiency, with significant results. To reduce the stress of soil salinity, Xinjiang has long employed large-scale water leaching techniques, most commonly used in winter and spring irrigation. However, this technique consumes a large amount of freshwater and carries the risks of raising the groundwater level and polluting groundwater bodies.

[0003] Against this backdrop, the dry-seeding and wet-emergence planting model, as another high-yield and efficient measure, has been rapidly promoted and applied. Dry-seeding and wet-emergence involves sowing directly at a suitable temperature for cotton seedling emergence without winter or spring irrigation before sowing. Drip irrigation under mulch is then used for 2-3 days after sowing to replenish soil moisture, followed by normal water management after emergence. Numerous studies have shown that the dry-seeding and wet-emergence model not only saves a significant amount of freshwater resources but also stabilizes and even increases yields. However, because the irrigation quotas for drip irrigation under mulch are relatively small during the growing season, the dry-seeding and wet-emergence planting model inevitably leads to soil salt accumulation in the root zone of cotton fields, thus affecting cotton growth and yield. Furthermore, the degree of this impact increases with the number of consecutive years the dry-seeding and wet-emergence model is used.

[0004] When soil salinity in the root zone of cotton fields accumulates to a certain level, or after several years of continuous application of the dry-seeding, wet-harvesting planting model, it is necessary to implement winter and spring irrigation. Once the salinity has decreased to a certain level, the dry-seeding, wet-harvesting model can be resumed. Therefore, accurately understanding the spatiotemporal distribution patterns of dry-seeding, wet-harvesting cotton fields is a crucial prerequisite for the rational allocation and efficient utilization of oasis water and soil resources. However, conducting farmer surveys, as an important technical method, is not only time-consuming and labor-intensive but also prone to significant errors. Therefore, there is an urgent need to explore a rapid and accurate method for identifying the spatiotemporal distribution patterns of cotton fields to provide scientific support for the optimal allocation of water and soil resources. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention aims to provide a remote sensing identification method for the distribution of cotton planting patterns. This invention, a simple and easy-to-implement remote sensing identification method for dry-seeded and wet-emerged cotton fields suitable for large-scale, long-term applications, is of great value for exploring the spatiotemporal evolution of soil salinity in the root zone of cotton fields and promoting the sustainable and healthy development of oasis agriculture in Xinjiang.

[0006] According to a first aspect of the present invention, a remote sensing identification method for the distribution of cotton planting patterns is provided, comprising the following steps:

[0007] Based on the first cotton field identification data source, extract the first cotton field feature remote sensing images of a specific area within the first identification period;

[0008] Based on iterative self-organizing data analysis technology, farmland in the specific area of ​​the first cotton field feature remote sensing image is classified to obtain suspected cotton field areas.

[0009] Based on the actual cotton planting area provided in the statistical yearbook, the accuracy of the suspected cotton field area is determined, and the accurate cotton field area is output.

[0010] Based on the second cotton field identification data source, extract the first moment and the second moment of the cotton field feature remote sensing image of the cotton field area during the second identification period;

[0011] Calculate the difference in normalized water index between the second time and the first time for each pixel in the cotton field area during the second identification period, and identify the winter and spring irrigation areas in the cotton field area based on the difference in normalized water index for each pixel.

[0012] Based on the distribution and area differences between the cotton field area and the winter-spring irrigation area, the distribution and area of ​​dry-sown and wet-harvested cotton fields are obtained.

[0013] This invention proposes first identifying all cotton fields within the study area, then identifying cotton fields that underwent winter and spring irrigation before cotton sowing in the current year. The remaining cotton fields, after deducting those that received winter and spring irrigation, are considered dry-sown, wet-opening cotton fields. Based on crop phenological characteristics, cotton exhibits a significant spectral difference compared to other crops during the boll-opening stage; therefore, satellite remote sensing images from this period can be selected for precise identification of cotton fields. Winter and spring irrigation generally occur after the previous year's cotton harvest and before the current year's cotton sowing, when the average daily temperature is above zero degrees Celsius, and are distributed randomly at a regional scale. If, through comparative analysis of two satellite remote sensing images, the surface soil moisture content of some fields (not all fields, thus excluding the influence of rainfall) within the study area is found to be significantly increased, these fields can be identified as winter- and spring-irrigated cotton fields. This invention is applicable to large-scale, long-term identification of dry-sown, wet-opening cotton fields.

[0014] According to an embodiment of the present invention, classifying farmland within a specific area of ​​the first cotton field feature remote sensing image based on iterative self-organizing data analysis technology to obtain suspected cotton field areas includes:

[0015] Import the first cotton field feature remote sensing image into the iterative self-organizing data analysis module;

[0016] Set the unsupervised classification parameters in the iterative self-organizing data analysis module to obtain multiple crop classification results containing different features;

[0017] The crop classification result that best matches the characteristics of the cotton field is selected as the suspected cotton field area through visual interpretation;

[0018] The unsupervised classification parameters include: minimum crop category value, maximum crop category value, maximum number of iterations, pixel transformation threshold, minimum number of pixels in a single category, maximum standard deviation in a single category, minimum difference between different categories, maximum number of categories merged in a single iteration, maximum allowable standard deviation, and maximum allowable bias.

[0019] The characteristics of the cotton field were obtained through analysis of the phenological characteristics of cotton growth.

[0020] According to embodiments of the present invention, obtaining all typical training samples (especially training samples for each crop type) of the study area is difficult for large-scale, long-term cotton field remote sensing identification. Therefore, the present invention uses the iterative self-organizing data analysis technique (ISODATA) in the unsupervised classification method of ENVI 5.6 software to first classify the farmland in the study area, and then selects the category that best matches the characteristics of cotton fields as the suspected cotton field area through visual interpretation.

[0021] According to an embodiment of the present invention, the unsupervised classification parameters refer to:

[0022] Minimum number of crop categories: The minimum number of crop categories in the study area;

[0023] Maximum number of crop categories: The highest number of crop categories in the study area;

[0024] Maximum number of iterations: related to accuracy requirements and the size of the study area;

[0025] Pixel transformation threshold (%): The basis for determining whether to continue the iteration. For a given iteration, if the percentage of pixel transformation for each crop category is less than this value, then the next iteration will not be performed.

[0026] Minimum number of pixels in a single category: If the number of pixels in a category is less than this value, the category is cancelled.

[0027] Maximum standard deviation of a single category: When the standard deviation of all pixels in a category is greater than this value, the category will be split.

[0028] Minimum difference between categories: When the difference between two categories is less than this value, the two categories will be merged.

[0029] Maximum number of categories in a single merge; maximum number of crop categories that can be merged in a single merge.

[0030] Maximum allowable standard deviation: During the iteration process, the criterion for determining whether a cell can be merged into a certain category is used. If the standard deviation after merging is greater than this value, the merging is canceled.

[0031] Maximum allowable deviation: During the iteration process, the criterion for determining whether a cell can be merged into a category is that if the deviation between the cell and the category mean is greater than this value, the merging is canceled.

[0032] According to an embodiment of the present invention, based on the actual cotton planting area provided in the statistical yearbook, the accuracy of the suspected cotton field area is determined, and the accurate cotton field area is output, including:

[0033] Obtain the area of ​​the suspected cotton field in the suspected cotton field region;

[0034] Compare the suspected cotton field area with the actual cotton planting area to calculate the relative error;

[0035] If the relative error is greater than 5%, the suspected cotton field area needs to be re-acquired and the accuracy of the suspected cotton field area needs to be reassessed.

[0036] If the relative error is less than or equal to 5%, then the accurate cotton field area is output.

[0037] According to an embodiment of the present invention, if the relative error is greater than 5%, the unsupervised classification parameters are adjusted (mainly the minimum and maximum values ​​of crop categories and the pixel transformation threshold are adjusted) and reclassification is performed until the relative error is less than 5%.

[0038] According to an embodiment of the present invention, in order to obtain better classification results and reduce the computation time of a single classification, the large-scale study area is usually subdivided into several sub-study areas according to cities and counties, and the farmland to be classified in each sub-study area is selected as the classification object.

[0039] According to an embodiment of the present invention, the first cotton field identification data source is a dataset of remote sensing images of the specific area acquired during the first identification period;

[0040] The first identification period is the cotton boll-opening stage and the period with low cloud cover.

[0041] The first remote sensing image of the cotton field features is a remote sensing image of the specific area acquired during the first identification period after atmospheric correction, radiometric calibration, cropping, and projection transformation processing.

[0042] According to an embodiment of the present invention, the first cotton field identification data source can be a Landsat-8OLI, Landsat-7ETM, or Landsat-5TM remote sensing image dataset with a resolution of 30m and a revisit period of 16 days within the first identification period.

[0043] According to embodiments of the present invention, atmospheric correction, radiometric calibration, clipping, and projection transformation can be performed using ENVI 5.6 software.

[0044] According to embodiments of the present invention, the key to accurately extracting cotton field distribution information lies in selecting remote sensing images of the appropriate time phase. Based on the analysis of cotton growth phenological characteristics, there are significant differences in the spectrum of cotton compared to other crops during the boll-opening period. Therefore, the cotton boll-opening period (generally from late August to early September for northern Xinjiang; and from early September to late September for southern Xinjiang) is generally selected as the optimal time for cotton field identification.

[0045] According to an embodiment of the present invention, when downloading satellite remote sensing images, remote sensing images that meet the identification time requirements and have low cloud cover are selected first; otherwise, remote sensing images from other nearby time periods are selected.

[0046] According to an embodiment of the present invention, a second cotton field feature remote sensing image of the cotton field area within a second identification period is extracted based on a second cotton field identification data source, comprising a first-time cotton field feature remote sensing image and a second-time cotton field feature remote sensing image, including:

[0047] The second identification period is the winter and spring irrigation identification period, which can be the period from the cotton harvest of the previous year to the sowing of the current year when the average daily temperature is consistently above zero degrees Celsius.

[0048] Based on the satellite transit cycle of the second cotton field identification data source and the weather conditions of the cotton field area, the winter and spring irrigation identification period is divided into multiple second identification periods;

[0049] The beginning of each second identification period is taken as the first moment;

[0050] The end of each second identification period is taken as the second moment;

[0051] Acquire remote sensing images of cotton field features at the first and second time points.

[0052] According to an embodiment of the present invention, the normalized water index difference between a second time point and a first time point within a second identification period is calculated for each pixel in the cotton field area, and the winter-spring irrigation area within the cotton field area is identified based on the normalized water index difference of each pixel, including:

[0053] Based on the remote sensing image of the cotton field features at the first moment, the normalized water index at the first moment is calculated.

[0054] Based on the remote sensing image of cotton field features at the second time point, the normalized water index at the second time point is calculated.

[0055] Calculate the difference in the normalized water index between the second time point and the first time point, and identify the area where the difference in the normalized water index exceeds the change threshold as the winter and spring irrigation area.

[0056] According to an embodiment of the present invention, the Normalized Difference Water Index (NDWI) is selected to identify cotton fields irrigated in winter and spring. This method utilizes the difference in reflectance of water bodies in different wavelength bands to highlight water body characteristics, and achieves the purpose of extracting water bodies by setting a threshold. Based on NDWI > 0, all water bodies within the study area at a certain time can be extracted, mainly including lakes, rivers, and winter- and spring-irrigated farmland. Since lakes and rivers are relatively fixed, while irrigation areas over large scales exhibit significant randomness in both time and space, this invention applies time series analysis. First, based on the difference in NDWI between the first and second moments of satellite transit, cotton fields with significantly increased surface water content during the second identification period are identified. Then, combined with rainfall data during the period, it is determined which cotton fields underwent winter and spring irrigation.

[0057] According to an embodiment of the present invention, the expression for the normalized water index is:

[0058]

[0059] In the formula, ρ G Green band surface reflectance;

[0060] ρ NIR Near-infrared band surface reflectance;

[0061] NDWI is the normalized water index.

[0062] According to an embodiment of the present invention, within a relatively long second identification period, the relative timing of winter and spring irrigation and the changes in NDWI can be divided into the following three scenarios:

[0063] (1) Winter and spring irrigation occur within a short period of time before the second moment of satellite transit. At the second moment, the cotton field still has a water layer or the surface soil moisture content is still very high, and the NDWI changes greatly during the period.

[0064] (2) Winter and spring irrigation occur within a short period of time after the first moment of satellite transit. Therefore, the surface soil moisture content of cotton fields is at a low level due to factors such as infiltration and evaporation at the second moment, and the NDWI change is small during the period.

[0065] (3) If winter and spring irrigation occur in the middle of the second identification period, the surface soil moisture content of the cotton field at the second moment and the changes in NDWI during the period will be relatively moderate. In general, the longer the remote sensing identification period, the smaller the changes in NDWI of the cotton field during the winter and spring irrigation period.

[0066] According to an embodiment of the present invention, based on the above situation, the step of calculating the difference in the normalized water index between the second time point and the first time point, and identifying the area where the difference in the normalized water index exceeds a change threshold as a winter-spring irrigation area, includes:

[0067] If the second identification period does not exceed 3 days, the change threshold is set to 0.15, and the area where the difference in the normalized water index is higher than 0.15 is the winter and spring irrigation area;

[0068] Alternatively, if the second identification period is between 4 and 7 days, the change threshold is set to 0.10, and the area where the difference in the normalized water index is higher than 0.10 is the winter-spring irrigation area;

[0069] Alternatively, if the second identification period exceeds 7 days, the change threshold is set to 0.05, and the area with a normalized water index difference higher than 0.05 is the winter-spring irrigation area.

[0070] According to an embodiment of the present invention, the second cotton field identification data source is a dataset of remote sensing images of the cotton field area acquired during the second identification period;

[0071] The second cotton field feature remote sensing image is a remote sensing image of the specific area acquired during the second identification period after atmospheric correction, radiometric calibration, cropping, and projection transformation processing.

[0072] According to an embodiment of the present invention, the second cotton field identification data source can be selected from the MOD09GA dataset with a resolution of 500m and a revisit period of 1 day during the second identification period as the identification data source for winter and spring irrigated cotton fields, and the selected remote sensing image data can be preprocessed using ENVI 5.6 software, including atmospheric correction, radiometric calibration, cropping, and projection transformation.

[0073] According to an embodiment of the present invention, the second identification period needs to coincide with the actual implementation of winter and spring irrigation in the local area. Generally, it can be selected from the period after the cotton harvest of the previous year to the sowing of the current year, when the average daily temperature is consistently above zero degrees Celsius.

[0074] According to an embodiment of the present invention, obtaining the distribution and area of ​​dry-sown and wet-harvested cotton fields based on the distribution and area differences between the cotton field area and the winter-spring irrigation area includes:

[0075] By subtracting the winter and spring irrigation areas from the cotton field area, the distribution of the dry-sown and wet-harvested cotton fields can be obtained.

[0076] The area of ​​the dry-sown and wet-harvested cotton fields was calculated based on the distribution of the dry-sown and wet-harvested cotton fields.

[0077] The formula for calculating the area of ​​dry-sown and wet-grown cotton fields is as follows:

[0078] DSWB Y =C Y -WI Y -SI Y

[0079] In the formula, DSWB Y The area of ​​cotton fields in the study area where the dry-seeding and wet-laying method was applied in year Y;

[0080] C Y This represents the cotton planting area in year Y;

[0081] WI Y This indicates the area of ​​cotton fields that underwent winter irrigation before cotton planting in year Y;

[0082] SI Y This indicates the area of ​​cotton fields that underwent spring irrigation before cotton planting in year Y.

[0083] According to an embodiment of the present invention, the distribution of the dry-sown and wet-harvested cotton fields can be obtained by deducting the winter and spring irrigation areas within the cotton field area, and can be processed using ArcGIS or other image processing software.

[0084] The beneficial effects of this invention are as follows:

[0085] This invention proposes a simple, feasible, and applicable method for large-scale, long-term identification of dry-sown, wet-emerged cotton fields. The method first employs an unsupervised classification algorithm, combined with multi-temporal remote sensing imagery and regional agricultural statistics, to accurately extract cotton fields. Based on this, a normalized water index model is constructed, combined with meteorological data, to remotely identify cotton fields that have undergone winter and spring irrigation, thereby extracting the spatiotemporal distribution information of dry-sown, wet-emerged cotton fields. This method has significant application value for evaluating and exploring the application effects of the dry-sown, wet-emerged planting pattern, especially its impact on soil salinization in the root zone of cotton fields, and for promoting the sustainable and healthy development of oasis agriculture in Xinjiang. Attached Figure Description

[0086] The present invention includes the following figures:

[0087] Figure 1 This is a schematic diagram showing the daily average temperature and precipitation during the winter and spring seasons of 2000-2001, as well as the remote sensing identification period of cotton irrigation fields during the winter and spring seasons, according to an embodiment of the present invention.

[0088] Figure 2 This is a spatiotemporal distribution map of dry-sown and wet-harvested cotton fields in northern Xinjiang in 2001, according to an embodiment of the present invention.

[0089] Figure 3 This is a spatiotemporal distribution map of dry-sown and wet-harvested cotton fields in northern Xinjiang in 2010, as described in this embodiment of the invention.

[0090] Figure 4 This is a spatiotemporal distribution map of dry-sown and wet-harvested cotton fields in northern Xinjiang in 2020, as described in this embodiment of the invention. Detailed Implementation

[0091] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings.

[0092] Example: Remote sensing identification of cotton planting patterns in northern Xinjiang from 2001 to 2020, with the remote sensing identification steps in 2001 as an example.

[0093] (1) Based on the first cotton field identification data source (Landsat-8OLI, Landsat-7ETM, and Landsat-5TM remote sensing image datasets with 30m resolution and a revisit period of 16 days provided by the U.S. Geological Survey), the first cotton field feature remote sensing images of northern Xinjiang during the first identification period (from late August to early September each year) were extracted. In this embodiment, the first cotton field feature remote sensing images extracted are remote sensing feature images of the entire northern Xinjiang LANDSAT series satellite remote sensing images with row and column numbers of 146 / 29, 145 / 29, 144 / 29, 143 / 29, and 142 / 29, respectively.

[0094] (2) Use ENVI 5.6 software to perform atmospheric correction, radiometric calibration, cropping, projection transformation and other preprocessing on the remote sensing image of the first cotton field in northern Xinjiang;

[0095] (3) The iterative self-organizing data analysis technique (ISODATA) in the unsupervised classification method of ENVI 5.6 software was used to classify farmland in the remote sensing image of the first cotton field in northern Xinjiang, and suspected cotton field areas were obtained. In order to obtain better classification results and reduce the computation time of a single classification, the entire northern Xinjiang region was further subdivided into 5 sub-study areas according to cities and counties. Farmland that needs to be classified in each sub-study area was selected as the classification object. The specific process is as follows:

[0096] 1) Import the remote sensing image of the first cotton field into the iterative self-organizing data analysis module;

[0097] 2) Set the unsupervised classification parameters in the iterative self-organizing data analysis module to obtain multiple crop classification results containing different features, including:

[0098] Minimum and maximum values ​​for crop categories:

[0099] The minimum and maximum values ​​for crop categories in the satellite remote sensing image with row and column numbers 146 / 29 are 5 and 24, respectively.

[0100] The minimum and maximum values ​​for crop categories in the satellite remote sensing image with row and column number 145 / 29 are 5 and 21, respectively.

[0101] The minimum and maximum values ​​for crop categories in the satellite remote sensing image with row and column numbers 144 / 29 are: 5 and 21.

[0102] The minimum and maximum values ​​for crop categories in the satellite remote sensing image with row and column numbers 143 / 29 are: 5 and 21.

[0103] The minimum and maximum values ​​for crop categories in the satellite remote sensing image with row and column numbers 142 / 29 are: 5 and 21.

[0104] Maximum number of iterations: 30;

[0105] Pixel transformation threshold (%): 5;

[0106] Minimum number of pixels per category: 1;

[0107] Maximum standard deviation for a single category: 1;

[0108] Minimum difference between categories: 5;

[0109] Maximum number of categories in a single merge: 2;

[0110] Maximum allowable standard deviation: default;

[0111] Maximum allowable deviation: default;

[0112] 3) Select the crop classification result that best matches the characteristics of cotton fields through visual interpretation as the suspected cotton field area: In the ISODATA classification results, select the category that best matches the characteristics of cotton fields through visual interpretation, compare the farmland area of ​​this category with the actual cotton planting area of ​​the study area in the statistical yearbook, if the relative error is greater than 5%, adjust the above parameters (mainly adjust the minimum and maximum values ​​of crop categories and the pixel transformation threshold) and reclassify until the relative error is less than 5%, and obtain the suspected cotton field area;

[0113] (4) Based on the actual cotton planting area provided in the statistical yearbook, determine the accuracy of suspected cotton field areas and output the accurate cotton field areas;

[0114] (5) Based on the second cotton field identification data source (the MOD09GA remote sensing image dataset with a resolution of 500m and a revisit period of 1 day provided by the National Aeronautics and Space Administration (NASA), the data for the second identification period (the period for identifying winter-irrigated cotton fields was from October 1 to November 3, 2000, and the period for identifying spring-irrigated cotton fields was from March 17 to April 15, 2001) was extracted. Figure 1 The first and second time-specific remote sensing images of cotton fields in the cotton field area are used. In this embodiment, the second cotton field feature remote sensing image is a remote sensing image with row and column numbers h23v04 and h24v04 from the MODIS series satellite remote sensing image that fully covers northern Xinjiang.

[0115] (6) The remote sensing images of the second cotton field in northern Xinjiang were preprocessed using ENVI 5.6 software, including atmospheric correction, radiometric calibration, cropping, and projection transformation.

[0116] (7) Calculate the Normalized Difference Water Index (NDWI) based on the remote sensing image data of cotton field characteristics at the first and second time points during the second identification period:

[0117]

[0118] In the formula ρ G With ρ NIR These represent the surface reflectance in the green band and near-infrared band, respectively.

[0119] Based on the difference between the NDWI values ​​at the first and second time points, cotton fields with significantly increased surface water content during the second identification period are identified. Then, combined with the rainfall data during the period, it is determined which cotton fields have implemented winter-spring irrigation: when the remote sensing identification period is no more than 3 days, or between 4 and 7 days, or more than 7 days, if the NDWI of a cotton field at the second time point is 0.15, 0.10, or 0.05 higher than the NDWI at the first time point, respectively, then it is considered that the cotton field has implemented winter-spring irrigation during the period.

[0120] (8) By deducting the winter and spring irrigation areas within the cotton field area, the distribution and area of ​​the dry-sown and wet-harvested cotton field area can be obtained.

[0121] Following the above process, the distribution of cotton planting patterns in northern Xinjiang from 2001 to 2020 was obtained, and the results are as follows.

[0122] Table 1 shows the area of ​​cotton fields and dry-sown / wet-transplanted cotton fields in northern Xinjiang from 2001 to 2020; Table 2 shows the area of ​​cotton fields that implemented dry-sown / wet-transplanted methods for several consecutive years from 2001 to 2020; Table 3 shows the cumulative area of ​​cotton fields that implemented dry-sown / wet-transplanted methods for several years from 2001 to 2020. Figure 2 , Figure 3 , Figure 4 The temporal and spatial distribution of dry-sown and wet-harvested cotton fields in northern Xinjiang in 2001, 2010, and 2020 are presented respectively.

[0123] Table 1. Area of ​​cotton fields and dry-sown / wet-transplanted cotton fields in northern Xinjiang, 2001-2020

[0124]

[0125] Table 2. Area of ​​cotton fields in northern Xinjiang that implemented dry sowing and wet transplanting for consecutive years from 2001 to 2020

[0126]

[0127]

[0128] Table 3. Cumulative Area of ​​Cotton Fields in Northern Xinjiang Implemented Dry-Sowing and Wet-Seedling Over Multiple Years (2001-2020)

[0129]

[0130] The above embodiments are only used to illustrate the present invention and are not intended to limit the present invention. Those skilled in the art can make various changes and modifications without departing from the essence and scope of the present invention. Therefore, all equivalent technical solutions also fall within the protection scope of the present invention.

[0131] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A remote sensing method for identifying the distribution of cotton planting patterns, characterized in that, Includes the following steps: Based on the first cotton field identification data source, extract the first cotton field feature remote sensing images of a specific area within the first identification period; Based on iterative self-organizing data analysis technology, farmland in the specific area of ​​the first cotton field feature remote sensing image is classified to obtain suspected cotton field areas. Based on the actual cotton planting area provided in the statistical yearbook, the accuracy of the suspected cotton field area is determined, and the accurate cotton field area is output. Based on the second cotton field identification data source, extract the first moment and the second moment of the cotton field feature remote sensing image of the cotton field area during the second identification period; Calculate the difference in normalized water index between the second time and the first time for each pixel in the cotton field area during the second identification period, and identify the winter and spring irrigation areas in the cotton field area based on the difference in normalized water index for each pixel. Based on the distribution and area differences between the cotton field area and the winter-spring irrigation area, the distribution and area of ​​dry-sown and wet-harvested cotton fields are obtained.

2. The remote sensing identification method for cotton field planting pattern distribution according to claim 1, characterized in that: The iterative self-organizing data analysis technique is used to classify farmland within a specific area of ​​the first cotton field feature remote sensing image to obtain suspected cotton field areas, including: Import the first cotton field feature remote sensing image into the iterative self-organizing data analysis module; Set the unsupervised classification parameters in the iterative self-organizing data analysis module to obtain multiple crop classification results containing different features; The crop classification result that best matches the characteristics of the cotton field is selected as the suspected cotton field area through visual interpretation; The unsupervised classification parameters include: minimum crop category value, maximum crop category value, maximum number of iterations, pixel transformation threshold, minimum number of pixels in a single category, maximum standard deviation in a single category, minimum difference between different categories, maximum number of categories merged in a single iteration, maximum allowable standard deviation, and maximum allowable bias. The characteristics of the cotton field were obtained through analysis of the phenological characteristics of cotton growth.

3. The remote sensing identification method for cotton field planting pattern distribution according to claim 1, characterized in that: The process of determining the accuracy of suspected cotton field areas based on the actual cotton planting area provided in the statistical yearbook, and outputting the accurate cotton field areas, includes: Obtain the area of ​​the suspected cotton field in the suspected cotton field region; Compare the suspected cotton field area with the actual cotton planting area to calculate the relative error; If the relative error is greater than 5%, the suspected cotton field area needs to be re-acquired and the accuracy of the suspected cotton field area needs to be reassessed. If the relative error is less than or equal to 5%, then the accurate cotton field area is output.

4. The remote sensing identification method for cotton field planting pattern distribution according to claim 1, characterized in that: The first cotton field identification data source is a dataset of remote sensing images of the specific area acquired during the first identification period; The first identification period is the cotton boll-opening stage and the period with low cloud cover. The first remote sensing image of the cotton field features is a remote sensing image of the specific area acquired during the first identification period after atmospheric correction, radiometric calibration, cropping, and projection transformation processing.

5. The remote sensing identification method for cotton field planting pattern distribution according to claim 1, characterized in that: Based on the second cotton field identification data source, second cotton field feature remote sensing images of the cotton field area within the second identification period are extracted, including cotton field feature remote sensing images at the first time and cotton field feature remote sensing images at the second time, including: The second identification period is the winter and spring irrigation identification period, which can be the period from the cotton harvest of the previous year to the sowing of the current year when the average daily temperature is consistently above zero degrees Celsius. Based on the satellite transit cycle of the second cotton field identification data source and the weather conditions of the cotton field area, the winter and spring irrigation identification period is divided into multiple second identification periods; The beginning of each second identification period is taken as the first moment; The end of each second identification period is taken as the second moment; Acquire remote sensing images of cotton field features at the first and second time points.

6. The remote sensing identification method for cotton field planting pattern distribution according to claim 1, characterized in that: Calculate the normalized water index difference between the second and first times within the second identification period for each pixel in the cotton field area, and identify the winter-spring irrigation area within the cotton field area based on the normalized water index difference for each pixel, including: Based on the remote sensing image of the cotton field features at the first moment, the normalized water index at the first moment is calculated. Based on the remote sensing image of cotton field features at the second time point, the normalized water index at the second time point is calculated. Calculate the difference in the normalized water index between the second time point and the first time point, and identify the area where the difference in the normalized water index exceeds the change threshold as the winter and spring irrigation area.

7. The remote sensing identification method for cotton field planting pattern distribution according to claim 6, characterized in that: The expression for the normalized water index is: In the formula, ρ G Green band surface reflectance; ρ NIR Near-infrared band surface reflectance; NDWI is the normalized water index.

8. The remote sensing identification method for cotton field planting pattern distribution according to claim 6, characterized in that: The step of calculating the difference in the normalized water index between the second time point and the first time point, and identifying areas where the difference in the normalized water index exceeds a change threshold as winter-spring irrigation areas, includes: If the second identification period does not exceed 3 days, the change threshold is set to 0.15, and the area where the difference in the normalized water index is higher than 0.15 is the winter and spring irrigation area; Alternatively, if the second identification period is between 4 and 7 days, the change threshold is set to 0.10, and the area where the difference in the normalized water index is higher than 0.10 is the winter-spring irrigation area; Alternatively, if the second identification period exceeds 7 days, the change threshold is set to 0.05, and the area with a normalized water index difference higher than 0.05 is the winter-spring irrigation area.

9. The remote sensing identification method for cotton field planting pattern distribution according to claim 5, characterized in that: The second cotton field identification data source is a dataset of remote sensing images of the cotton field area acquired during the second identification period; The second cotton field feature remote sensing image is a remote sensing image of the specific area acquired during the second identification period after atmospheric correction, radiometric calibration, cropping, and projection transformation processing.

10. The remote sensing identification method for cotton field planting pattern distribution according to claim 1, characterized in that: The process of obtaining the distribution and area of ​​dry-sown, wet-harvested cotton fields based on the distribution and area differences between the cotton field area and the winter-spring irrigation area includes: By subtracting the winter and spring irrigation areas from the cotton field area, the distribution of the dry-sown and wet-harvested cotton fields can be obtained. The area of ​​the dry-sown and wet-harvested cotton fields was calculated based on the distribution of the dry-sown and wet-harvested cotton fields. The formula for calculating the area of ​​dry-sown and wet-grown cotton fields is as follows: DSWB Y =C Y -WI Y -SI Y In the formula, DSWB Y The area of ​​cotton fields in the study area where the dry-seeding and wet-laying method was applied in year Y; C Y This represents the cotton planting area in year Y; WI Y This indicates the area of ​​cotton fields that underwent winter irrigation before cotton planting in year Y; SI Y This indicates the area of ​​cotton fields that underwent spring irrigation before cotton planting in year Y.

Images