Method and device for dynamically monitoring arable soil desertification based on multi-source satellite remote sensing

Abstract

The embodiment of the invention provides a method and a device for dynamically monitoring arable soil desertification based on multi-source satellite remote sensing. The method is applied to the technical field of environmental protection, and comprises the following steps: training a machine learning model by using a constructed inversion data set to obtain a soil sand content inversion model; based on the constructed long-time-sequence cultivated land bare soil data set and the corresponding topographic data, performing inversion by using a soil sand content inversion model to obtain cultivated land soil sand content data of the research area; on the basis of the cultivated land soil sand content data, cultivated land soil desertification early-warning factors are screened from climate data, topographic data, social economic data and soil data of the research area, weights corresponding to the cultivated land soil desertification early-warning factors are calculated, and a cultivated land soil desertification early-warning model is established on the basis of the cultivated land soil desertification early-warning factors and the weights corresponding to the cultivated land soil desertification early-warning factors; and on the basis, arable soil desertification early warning is carried out on the research area. In this way, arable soil desertification dynamic monitoring and risk early warning can be achieved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of environmental protection, and particularly to a method and device for dynamically monitoring the desertification of cultivated land soil based on multi-source satellite remote sensing. Background Art

[0002] The desertification of cultivated land soil poses a severe challenge to the agricultural ecosystem, food production, and the human living environment. Currently, the monitoring of cultivated land soil desertification mainly relies on field surveys and remote sensing technology. Although field surveys have high accuracy, they are time-consuming and costly, making it difficult to achieve large-scale and high-frequency dynamic monitoring. Traditional remote sensing monitoring mostly relies on single-satellite data, which has limitations in temporal and spatial resolution. For example, although the Landsat series of satellites have long-term time-series data, their temporal resolution is low, making it difficult to capture short-term dynamic changes in soil desertification. In addition, single data sources are vulnerable to factors such as cloud cover and vegetation cover in the face of complex surface features, affecting the accuracy and reliability of monitoring results.

[0003] The introduction of multi-source satellite remote sensing technology has brought new opportunities for the monitoring of cultivated land soil desertification. First, the fusion of multi-source satellite data can significantly improve the spatial and temporal resolution of monitoring data. For example, combining high-resolution optical satellite data with high-temporal-resolution MODIS data can not only obtain fine surface information but also achieve continuous monitoring of the soil desertification process. Second, the fusion of multi-source data can give full play to the advantages of different sensors and make up for the deficiencies of single data sources. In addition, multi-source satellite remote sensing technology can also provide richer surface feature information, such as spectral features and temperature features, which helps to deeply explore the causes and evolution laws of soil desertification.

[0004] Carrying out research on the dynamic monitoring of cultivated land soil desertification based on multi-source satellite remote sensing has important practical significance and scientific value. First, this is crucial for improving the accuracy and efficiency of cultivated land soil desertification monitoring. Through the fusion of multi-source data, refined monitoring of the soil desertification process can be achieved, and the distribution, scope, and change trends of desertified land can be grasped in a timely manner. This has important guiding significance for scientifically formulating sand prevention and control policies, rationally planning land use, and effectively carrying out the treatment and restoration of desertified land. Second, this research helps to promote the innovation and development of remote sensing technology in the field of soil desertification monitoring. By continuously optimizing multi-source data fusion algorithms and expanding the application scope of remote sensing monitoring, more advanced technical means and methods can be provided for soil desertification monitoring. In addition, this research also has important ecological and social benefits. Accurate and timely monitoring information on cultivated land soil desertification can provide strong support for ecological security assessment and regional sustainable development decision-making, helping to protect the ecological environment, ensure food security, and promote the sustainable development of the economy and society. Summary of the Invention

[0005] In a first aspect, embodiments of the present disclosure provide a method for dynamically monitoring the desertification of cultivated land soil based on multi-source satellite remote sensing. The method includes:

[0006] Obtain the sand grain content data and laboratory spectral data of the surface soil samples of typical soil types collected within the cultivated land scope of the research area through laboratory measurement, and obtain the multi-source satellite spectral data of the sample points based on the transit images during the sample collection period;

[0007] Using the multi-source satellite spectral data of the sample points, extract the spectral features of the common bands therein, construct the dual-band spectral index between any two bands, and add a third band to the dual-band spectral index to construct a triple-band spectral index. Taking the constructed spectral index, the corresponding terrain data, and the corresponding bands as input features, and the sand grain content data measured in the laboratory as the target output, construct an inversion dataset, and use the inversion dataset to train a machine learning model to obtain a soil sand grain content inversion model;

[0008] Query all available multi-source satellite remote sensing data in all historical periods according to the bare soil period window time corresponding to multiple sub-regions in the research area to obtain the historical remote sensing dataset of the research area; extract the bare soil pixels from the remote sensing data in the historical remote sensing dataset during the selected bare soil synthesis period, and perform bare soil image synthesis to obtain the bare soil image synthesis result corresponding to the time period, so as to construct a long-term bare soil dataset;

[0009] Perform cultivated land extraction processing on the long-term bare soil dataset to obtain a long-term cultivated land bare soil dataset, and perform inversion based on the long-term cultivated land bare soil dataset and the corresponding terrain data using the soil sand grain content inversion model to obtain the cultivated land soil sand grain content data of the research area;

[0010] Based on the cultivated land soil sand grain content data, screen the cultivated land soil desertification warning factors from the climate data, terrain data, social and economic data, and soil data of the research area, calculate the weights corresponding to the cultivated land soil desertification warning factors, and establish a cultivated land soil desertification warning model based on the cultivated land soil desertification warning factors and their corresponding weights;

[0011] Use the cultivated land soil desertification warning model to conduct a warning on the cultivated land soil desertification in the research area.

[0012] In some realizable ways of the first aspect, adding a third band to the dual-band spectral index to construct a triple-band spectral index includes:

[0013] Calculate the correlation coefficient between the dual-band index and the sand grain content data measured in the laboratory, and add a third band to the dual-band spectral index with the correlation coefficient greater than the preset threshold to construct a triple-band spectral index, and supplement the existing spectral index in the constructed triple-band spectral index.

[0014] In some realizable ways of the first aspect, a machine learning model is trained using an inversion dataset to obtain a soil sand content inversion model, including:

[0015] Based on the inversion dataset, the training set and the validation set are divided, and the input features actually used for modeling are screened through a cross-validation algorithm, and the input features are removed from the training set based on this, and the input features actually used for modeling are retained;

[0016] The updated training set is used to train the machine learning model, and hyperparameter tuning is performed to obtain a soil sand content inversion model.

[0017] In some realizable ways of the first aspect, the remote sensing data in the historical remote sensing dataset within the selected bare soil synthesis period is used to extract bare soil pixels and perform bare soil image synthesis to obtain the bare soil image synthesis result for the corresponding time period, and a long-term bare soil dataset is constructed based on this, including:

[0018] Based on the laboratory spectral data and the spectral response functions of each satellite, the laboratory spectral data is resampled to simulate the spectral characteristics of each sensor band, and a spectral response relationship model between different sensors is established to ensure the implementation of standardized processing and unified scale conversion of cross-sensor data, and the remote sensing data after conversion is used to extract bare soil pixels and perform bare soil image synthesis to obtain the bare soil image synthesis result for the corresponding time period, and a long-term bare soil dataset is constructed based on this.

[0019] In some realizable ways of the first aspect, the long-term bare soil dataset is processed for cultivated land extraction to obtain a long-term cultivated land and bare soil dataset, including:

[0020] A cultivated land mask dataset is made using the annual land use dataset, and based on this, the long-term bare soil dataset is processed for cultivated land extraction to obtain a long-term cultivated land and bare soil dataset.

[0021] In some realizable ways of the first aspect, based on the cultivated land soil sand content data, the early warning factors for cultivated land soil desertification are screened from the climate data, terrain data, social and economic data, and soil data of the research area, and the weights corresponding to the early warning factors for cultivated land soil desertification are calculated, including:

[0022] Based on the cultivated land soil sand content data, the early warning factors for cultivated land soil desertification are screened from the climate data, terrain data, social and economic data, and soil data of the research area through a correlation analysis algorithm;

[0023] The relative importance of each early warning factor for cultivated land soil desertification is calculated through pairwise comparison, and the corresponding weights are obtained based on this.

[0024] In some realizable ways of the first aspect, based on the warning factors of cultivated land soil desertification and their corresponding weights, a warning model for cultivated land soil desertification is established, including:

[0025] Based on the warning factors of cultivated land soil desertification and their corresponding weights, a warning evaluation model for cultivated land soil desertification is established using a comprehensive evaluation algorithm to quantitatively evaluate the risk of cultivated land soil desertification, predict the risk level of cultivated land soil desertification, and establish a warning model for cultivated land soil desertification based on this.

[0026] In the second aspect, the embodiments of the present disclosure provide a device for dynamic monitoring of cultivated land soil desertification based on multi-source satellite remote sensing. The device includes:

[0027] An acquisition module for acquiring the sand content data and laboratory spectral data of the surface soil samples of typical soil types collected within the cultivated land scope of the research area through laboratory measurement, and acquiring the multi-source satellite spectral data of the sample points based on the transit images during the sample collection period;

[0028] A construction module for using the multi-source satellite spectral data of the sample points to extract the spectral characteristics of the common bands therein, constructing a dual-band spectral index between any two bands, and adding a third band to the dual-band spectral index to construct a triple-band spectral index. Using the constructed spectral index, the corresponding terrain data, and the corresponding bands as input features, and the sand content data measured in the laboratory as the target output, an inversion data set is constructed, and the inversion data set is used to train a machine learning model to obtain a soil sand content inversion model;

[0029] An extraction module for querying all available multi-source satellite remote sensing data in all historical periods according to the bare soil period window time corresponding to multiple sub-regions within the research area to obtain the historical remote sensing data set of the research area; extracting bare soil pixels from the remote sensing data in the historical remote sensing data set during the selected bare soil synthesis period and performing bare soil image synthesis to obtain the bare soil image synthesis result corresponding to the time period, and constructing a long-term bare soil data set based on this;

[0030] An inversion module for performing cultivated land extraction processing on the long-term bare soil data set to obtain a long-term cultivated land bare soil data set, and performing inversion using the soil sand content inversion model based on the long-term cultivated land bare soil data set and the corresponding terrain data to obtain the cultivated land soil sand content data of the research area;

[0031] A construction module for screening the warning factors of cultivated land soil desertification from the climate data, terrain data, social and economic data, and soil data of the research area based on the cultivated land soil sand content data, calculating the weights corresponding to the warning factors of cultivated land soil desertification, and establishing a warning model for cultivated land soil desertification based on the warning factors of cultivated land soil desertification and their corresponding weights;

[0032] An early warning module for using a cultivated land soil desertification early warning model to conduct early warning of cultivated land soil desertification in a research area.

[0033] In a third aspect, embodiments of the present disclosure provide an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0034] In a fourth aspect, embodiments of the present disclosure provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method as described above.

[0035] In the embodiments of the present disclosure, a constructed inversion dataset is used to train a machine learning model to obtain a soil sand content inversion model; based on the constructed long-term cultivated land bare soil dataset and the corresponding terrain data, the soil sand content inversion model is used for inversion to obtain the cultivated land soil sand content data of the research area; based on the cultivated land soil sand content data, the cultivated land soil desertification early warning factors are screened from the climate data, terrain data, social and economic data, and soil data of the research area, and the weights corresponding to the cultivated land soil desertification early warning factors are calculated. Based on the cultivated land soil desertification early warning factors and their corresponding weights, a cultivated land soil desertification early warning model is established; the cultivated land soil desertification early warning model is used to conduct early warning of cultivated land soil desertification in the research area. In this way, dynamic monitoring and risk early warning of cultivated land soil desertification can be realized.

[0036] It should be understood that the content described in the summary of the invention section is not intended to limit the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In combination with the accompanying drawings and referring to the following detailed description, the above and other features, advantages and aspects of the embodiments of the present disclosure will become more obvious. The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. In the drawings, the same or similar reference numerals represent the same or similar elements, where:

[0038] Figure 1 is a flowchart of a method for dynamic monitoring of cultivated land soil desertification based on multi-source satellite remote sensing provided by an embodiment of the present disclosure;

[0039] Figure 2 shows a synthesized soil image of a phase provided by an embodiment of the present disclosure;

[0040] Figure 3 shows a schematic diagram of the mapping result of the cultivated land soil sand content of a phase provided by an embodiment of the present disclosure;

[0041] Figure 4 shows a structural diagram of a device for dynamically monitoring the desertification of cultivated land soil based on multi-source satellite remote sensing provided by an embodiment of the present disclosure;

[0042] Figure 5 shows a structural diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure. Detailed implementation manners

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.

[0044] In addition, the term "and / or" in this document is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after.

[0045] In response to the problems in the background art, the embodiments of the present disclosure provide a method, device, equipment, and storage medium for dynamically monitoring the desertification of cultivated land soil based on multi-source satellite remote sensing. The following will, with reference to the accompanying drawings, explain in detail a method, device, equipment, and storage medium for dynamically monitoring the desertification of cultivated land soil based on multi-source satellite remote sensing provided by the embodiments of the present disclosure through specific embodiments.

[0046] Figure 1 shows a flowchart of a method for dynamically monitoring the desertification of cultivated land soil based on multi-source satellite remote sensing provided by an embodiment of the present disclosure. As Figure 1 shown, the method 100 for dynamically monitoring the desertification of cultivated land soil may include the following steps:

[0047] S110, obtain the sand grain content data and laboratory spectral data of the surface soil samples of typical soil types collected within the cultivated land scope of the research area through laboratory measurement, and obtain the multi-source satellite spectral data of the sample points based on the passing images during the sample collection period.

[0048] In some embodiments, surface soil samples can be collected within the cultivated land area of the research region to ensure that the samples can cover typical soil types. Thereafter, the sand content data and laboratory spectral data of the samples are measured through the laboratory, and satellite spectral data of the sample points are obtained based on the transit images during the sample collection period (multi-source satellite remote sensing data can be selected and screened according to the image quality).

[0049] Meanwhile, topographic data with a spatial resolution of not less than 30m in the research region, annual land use data, and auxiliary data such as historical annual climate data, socioeconomic data, and soil data in the research region are queried.

[0050] S120. Using the multi-source satellite spectral data of the sample points, extract the spectral characteristics of the common bands therein, construct the dual-band spectral index between any bands, and add a third band to the dual-band spectral index to construct a triple-band spectral index. Taking the constructed spectral index, the corresponding topographic data, and the corresponding bands as input features, and the sand content data measured in the laboratory as the target output, construct an inversion dataset, and use the inversion dataset to train a machine learning model to obtain a soil sand content inversion model.

[0051] In some embodiments, the multi-source satellite spectral data of the sample points can be used to extract the spectral characteristics of the common bands therein, construct the dual-band spectral index between any bands, then calculate the correlation coefficient between the dual-band index and the sand content data measured in the laboratory, and add a third band to the dual-band spectral index with a correlation coefficient greater than the preset threshold to construct a triple-band spectral index, and supplement the existing spectral index in the constructed triple-band spectral index. Then, taking the constructed spectral index and the supplemented spectral index, the corresponding topographic data, and the corresponding bands as input features, and the sand content data measured in the laboratory as the target output, construct an inversion dataset. Based on the inversion dataset, divide the training set and the validation set, screen the input features actually used for modeling through the cross-validation algorithm, and perform feature elimination on the training set accordingly, retain the input features actually used for modeling, use the updated training set to train the machine learning model, and perform hyperparameter tuning to obtain a soil sand content inversion model.

[0052] That is to say, the multi-source satellite spectral data of the sample points are used here to construct a dual-band spectral index (such as difference index, ratio index, normalized index, square root index, etc.) between arbitrary bands, and the correlation coefficient between the dual-band spectral index and the measured sand content data is calculated. For the dual-band spectral index with high correlation, a third band is added on its basis to construct a three-band spectral index. The spectral index related to the existing soil texture research is supplemented to the constructed three-band spectral index, and the inversion data set is constructed in combination with the terrain data and the measured sand content data. The inversion data set is divided into a training set and a test set by proportional stratified sampling. The modeling features constructed using the training set are used together with the original bands for feature screening through cross-validation, and the screened features are used to construct a soil sand content inversion model, and hyperparameter tuning is performed to further optimize the model (the accuracy evaluation standard of the model is: R 2 , RMSE, RPD).

[0053] It is worth noting that here, on the basis of conventional spectral indices, all available two-band spectral indices are systematically compared. In addition to the difference, ratio and normalized indices, the square root index is also introduced. And based on the spectral indices that have already had good results, a three-band index is further constructed to explore deeper spectral features in multispectral data.

[0054] S130, query the available multi-source satellite remote sensing data of all historical periods according to the bare soil period window time corresponding to multiple sub-areas in the study area, and obtain the historical remote sensing data set of the study area; extract bare soil pixels from the remote sensing data in the historical remote sensing data set within the selected bare soil synthesis period, and perform bare soil image synthesis to obtain the bare soil image synthesis result of the corresponding time period, so as to construct a long-term bare soil data set.

[0055] In some embodiments, the influencing factors of the bare soil period, such as latitude, climate, and the growth cycle of major crops, can be comprehensively considered to set the remote sensing image query time (i.e., the bare soil period window time) for multiple sub-areas in the study area. According to the scope of the study area and the set bare soil period window time, the multi-source satellite remote sensing data available in all historical periods are queried to obtain the historical remote sensing data set of the study area. Thereafter, the laboratory spectral data is resampled based on the laboratory spectral data and the spectral response function of each satellite, the band characteristics of each sensor are simulated, and a spectral response relationship model between different sensors is established to ensure the standardized processing and unified scale conversion of cross-sensor data, and the converted remote sensing data is subjected to bare soil pixel extraction, and bare soil image synthesis is performed to obtain the bare soil image synthesis result of the corresponding time period, thereby constructing a long-time series bare soil data set.

[0056] As an example, to ensure the spectral consistency between laboratory spectral data and multi-source satellite remote sensing data, i.e., data from different remote sensing sensors, adaptation conversion was carried out here based on the laboratory spectral data and the spectral response functions of each satellite to match the spectral response ranges of different remote sensing sensors such as Landsat 5, Landsat 8, and Landsat 9. Through this process, a regression model was established to determine the band conversion coefficients between different sensors, thereby ensuring the consistency of laboratory data and remote sensing data within the same spectral range.

[0057] In addition, a set of rules for extracting bare soil pixels was constructed here, and the bare soil index threshold was adjusted based on the satellite spectral data of the sample points to improve its applicability in the study area. By sorting and screening the obtained remote sensing data sets year by year, it was ensured that the data for each year could cover the corresponding cultivated land area. For the missing parts of the data, they were supplemented by expanding the query time range or increasing the data sources. At the same time, considering the spectral response differences between different sensors, the spectral response conversion coefficients were calculated to uniformly convert the time-series remote sensing image data obtained by different sensors into image data consistent with the sensor used for modeling, thus ensuring the consistency and comparability of the data. The bare soil pixels of each image data were extracted, and finally, the bare soil pixel sequences of each image data were obtained. Based on this, a long-term bare soil data set was constructed, providing high-quality data support for the inversion model of soil sand content.

[0058] It should be noted that here, based on the screening of images during the bare soil period, a complete set of bare soil extraction technical processes was formed using multiple bare soil spectral characteristics. The index threshold was further corrected in combination with the sample data, and the distribution information of bare soil in a large area and long time series was extracted by integrating multi-source remote sensing data.

[0059] S140, the cultivated land extraction process was carried out on the long-term bare soil data set to obtain the long-term cultivated land bare soil data set. Based on the long-term cultivated land bare soil data set and the corresponding topographic data, the soil sand content inversion model was used for inversion to obtain the cultivated land soil sand content data of the study area.

[0060] In some embodiments, the cultivated land mask data set can be made using the annual land use data set. Based on this, the cultivated land extraction process is carried out on the long-term bare soil data set to obtain the long-term cultivated land bare soil data set. Based on the long-term cultivated land bare soil data set and the corresponding topographic data, the soil sand content inversion model is used for inversion to obtain the cultivated land soil sand content data of the study area.

[0061] As an example, refer to S130. Here, by synthesizing the sequence of bare soil pixels in the long-time series cultivated land bare soil dataset year by year, the median or mean method is used to ensure numerical stability, and the annual bare soil synthetic image data of the study area is obtained. Combining the cultivated land data of the corresponding year, the annual synthetic results of cultivated land bare soil in the study area are further generated, and the original spectral information is retained to obtain the long-time series cultivated land bare soil dataset. Using the constructed inversion model of soil sand content, based on the spectral information of the cultivated land bare soil synthetic image data in the long-time series cultivated land bare soil dataset and the corresponding topographic data, the spatial distribution map of the soil sand content of the cultivated land in the study area is generated.

[0062] It should be noted that here the corresponding relationship between the soil sand content and the spectral data is extended to the time dimension, and the regression model is applied to the long-time series bare soil synthesis results to realize the inversion of the soil sand content across years, providing a dynamic monitoring means for the long-term changes of the soil properties in the study area.

[0063] S150. Based on the data of the soil sand content of cultivated land, select the early warning factors for cultivated land soil desertification from the climate data, topographic data, social and economic data, and soil data of the study area, calculate the weights corresponding to the early warning factors for cultivated land soil desertification, and establish an early warning model for cultivated land soil desertification based on the early warning factors for cultivated land soil desertification and their corresponding weights.

[0064] In some embodiments, based on the data of the soil sand content of cultivated land, the early warning factors for cultivated land soil desertification can be selected from the climate data, topographic data, social and economic data, and soil data of the study area through the correlation analysis algorithm. Each pair of early warning factors for cultivated land soil desertification is compared, and their relative importance is calculated, and the corresponding weights are obtained accordingly. Based on the early warning factors for cultivated land soil desertification and their corresponding weights, a comprehensive evaluation algorithm is used to establish an early warning evaluation model for cultivated land soil desertification, quantitatively evaluate the risk of cultivated land soil desertification, predict the risk level of cultivated land soil desertification, and establish an early warning model for cultivated land soil desertification based on this. Exemplarily, the above process can be specifically expanded as follows:

[0065] (1) Selection of early warning factors for cultivated land soil desertification: For the sample data, obtain the potential influencing factors of soil desertification such as climate data, social and economic data, and soil data within the corresponding time, perform resampling, and obtain the correlation degree values between each factor and the data of the soil sand content of cultivated land through methods such as correlation analysis, and screen the early warning factors for cultivated land soil desertification according to the values.

[0066] (2) Determination of the weights of the early warning factors for cultivated land soil desertification: Each pair of early warning factors is compared, and their relative importance is calculated to obtain the weights. A consistency test is carried out to ensure that the judgment results are reasonable. Through this process, the influence of each early warning factor on soil desertification is quantified, providing a basis for the early warning model.

[0067] (3) Establishment of the early warning evaluation model for cultivated land soil desertification: Based on the selected early warning factors and their weights, combined with relevant mathematical models and algorithms, quantitatively evaluate the risk of cultivated land soil desertification. Predict the risk levels (mild, moderate, severe) of soil desertification through the model.

[0068] (4) Construction of the early warning model for cultivated land soil desertification: Construct an early warning model based on early warning factors, combine data such as climate, terrain, social economy, and soil, dynamically monitor the risk level of soil desertification, and provide data support for prevention and control measures.

[0069] It should be noted that this provides a new approach for the early warning and precise prevention and control of cultivated land soil desertification. Based on the high-precision long-time-series dataset of the sand content in cultivated land soil, long-term evaluation of cultivated land soil desertification can be achieved, and the future situation of cultivated land soil desertification in the research area can be predicted using the new dataset.

[0070] S160. Use the early warning model for cultivated land soil desertification to conduct early warning of cultivated land soil desertification in the research area.

[0071] In summary, according to the embodiments of the present disclosure, at least the following technical effects are achieved:

[0072] (1) Data integration and high-precision inversion: The high-precision inversion model for soil sand content constructed in the present disclosure provides accurate and robust soil sand content estimation capabilities. Compared with the prior art, by collecting and processing multi-source satellite remote sensing data, laboratory spectral data, and long-term meteorological, terrain, and other auxiliary data, the present disclosure can better meet the requirements for obtaining soil texture information in different regions and time periods, and provide more reliable soil desertification monitoring results.

[0073] (2) Long-time-series data and cross-year monitoring: By constructing a long-time-series cultivated land bare soil dataset, the present disclosure realizes the functions of historical data synthesis and continuous monitoring based on satellite images, overcoming the limitations of traditional methods in data missing and regional coverage. This function provides cross-year dynamic monitoring capabilities for soil desertification, enabling the present disclosure to have strong depth and breadth in the analysis of soil property changes in different years.

[0074] (3) Automatic early warning and evaluation: Through the screening of early warning factors and the weighted evaluation model for cultivated land soil desertification in the present disclosure, refined grading and real-time monitoring of the risk of cultivated land soil desertification are realized. Compared with the prior art, the risk level prediction model based on multiple factors in the present disclosure can provide more timely and regionally targeted desertification prevention and control information, helping users formulate reasonable soil management measures in advance.

[0075] (4) Easy-to-use multi-dimensional data visualization support: The spatio-temporal distribution map of the sand content in cultivated land soil and the desertification risk level map generated by this disclosure provide the product with rich visualization information output capabilities. Combining diverse visualization means such as the dynamic change map of remote sensing images and the soil desertification trend map of geographical locations can visually present soil changes and provide strong data support for user decision-making.

[0076] For the convenience of further understanding, the cultivated land soil desertification dynamic monitoring method 100 provided by this disclosure will be described in detail below in combination with a specific embodiment, as follows:

[0077] (1) Data collection and processing

[0078] 1.1) Soil data collection

[0079] In April 2021, 162 surface soil samples (0 - 15 cm deep) were collected in the farmland area of the black soil region by the random sampling method. Within each sampling point, five soil sub-samples were collected in a circular area with a diameter of 10 meters by the five-point method, and these sub-samples were mixed to form a composite sample. 1 kg of composite sample was collected at each point. After air-drying indoors, gravel and plant residues were removed through a 2-mm sieve. An appropriate amount of the sample was taken, a dispersant and deionized water were added, the particles were dispersed by ultrasonic oscillation, then washed, and a dispersing medium was added for background measurement. The scattered light signal was collected using a laser particle size analyzer, and the measurement was repeated 2 - 3 times for calibration and analysis to determine the particle size distribution. According to the soil texture classification principle, the sand content (53 - 2000 μm) in the sample was quantitatively analyzed. In 2022 and 2023, the soil data and the corresponding laboratory spectral data collected from the cultivated land in the study area were obtained through the ASD FieldSpec4 Hi-Res spectrometer in the soil measurement laboratory, with 243 spectral data points. The spectral resolution of the laboratory spectral data of the soil samples was 3 nm in the 400 - 1000 nm band and 8 nm in the 1001 - 2500 nm band. These laboratory spectral data served as the main data source for spectral conversion. At the same time, Landsat 8 data covering the sampling points and matching the sampling time was selected, and the spectral data of the corresponding pixels (30 m resolution) were extracted through resampling.

[0080] 1.2) Temporal remote sensing image collection

[0081] For the black soil region, 4 bare soil period intervals were set according to the latitude range, as shown in Table 1 specifically. The surface reflectance (SR) data of Landsat 5, 8, and 9 sensors were screened, and these data were archived in the C02 / T1_L2 dataset in GEE. Clouds and snow were removed from the images through the QA_PIXEL band to ensure the accuracy and availability of the data.

[0082] Table 1

[0083]

[0084] 1.3) Supplementary data

[0085] The digital elevation data with a 30-meter resolution of NASA DEM was adopted, and the slope and aspect data were calculated accordingly. Meanwhile, the annual cultivated land dataset with a 30-meter resolution in the black soil area was collected. The annual average temperature and annual precipitation data, with a resolution of 1 km, were sourced from the National Earth System Science Data Center of China. In addition, the ERA5 land monthly average dataset of the European Centre for Medium-Range Weather Forecasts (ECMWF) was selected as the source of other meteorological data, including the monthly average air pressure and wind speed data during the bare soil period, and were uniformly resampled to a 1-km resolution in Google Earth Engine (GEE). The gross domestic product (GDP) and population density (POP) data had a spatial resolution of 1 km. The spatial distribution data of soil types was sourced from the "1:1,000,000 Soil Map of the People's Republic of China" compiled and published by the National Soil Census Office in 1995, which adopted the traditional Chinese "National Soil Taxonomic System".

[0086] (2) Construction of the soil sand content inversion model

[0087] 2.1) Construction of the optimal spectral index

[0088] The spectral information of the sampling points was obtained by selecting the remote sensing image data passing by during the sampling period. This information had a certain correlation with the surface properties of the soil and could mitigate the influence of soil moisture, surface roughness, and atmospheric conditions. Here, four two-band indices were selected: the ratio index, the difference index, the normalized index, and the square root index, to construct the spectral index related to the soil sand content. The mathematical expressions of these spectral indices are as follows:

[0089] RI(RBi,RBj)=RBi / RBj (1)

[0090] DI(RBi,RBj)=RBi - RBj (2)

[0091] NDI(RBi,RBj)=(RBi - RBj) / (RBi + RBj) (3)

[0092]

[0093] Adding the third band in the specific sensitive area to the two-band spectral index can significantly improve the estimation accuracy and enhance the anti-interference ability. Therefore, several three-band spectral indices were selected here:

[0094]

[0095]

[0096] TBI3(RBi, RBj, RBz) = (RBi - RBj) - (RBj - RBz) (7)

[0097]

[0098] 2.2) Random Forest Regression Model

[0099] The random forest regression model is a non - linear machine learning model based on decision trees. This algorithm is widely used to predict the spatial distribution of soil parameters. In cases where environmental factors have a significant impact and there is no obvious linear relationship between variables, the random forest algorithm usually has higher prediction accuracy compared to traditional statistical and geostatistical models. It has the advantages of handling high - dimensional data, outputting feature importance, improving accuracy through result averaging, and supporting fast parallel computing.

[0100] The features used include six original spectral bands of Landsat data, 300 constructed spectral indices, and geographical variables such as altitude, elevation, and slope aspect. Through the forward feature selection method, first, the variable with the highest random forest score is selected as the initial input, and then the variables are sorted according to their importance and added in turn until further addition no longer improves the model performance. The final feature set selects the top ten features through this method for subsequent modeling.

[0101] After feature selection, to improve the model accuracy, the hyperparameters of the random forest regression model are tuned. After identifying the most important features, the best combination of hyperparameters is searched through GridSearchCV to ensure the maximization of model performance. The optimized model parameters will be uploaded to the cloud through the Python API GeeMap, making it convenient to directly call the final model in subsequent analysis and applications.

[0102] 2.3) Accuracy Analysis

[0103] 162 samples are divided into a calibration set and a validation set in a ratio of 3:1, where 121 are training samples and 41 are test samples. The coefficient of determination (R 2 ), root mean square error (RMSE), and relative prediction deviation (RPD) are calculated as indicators to evaluate the performance of different models.

[0104]

[0105]

[0106]

[0107]

[0108] where n is the number of samples, is the average value of the measured values, y i is the measured value, is the predicted value, and SD is the standard deviation. Generally, a well-developed model usually has high R 2 and RPD values, as well as low RMSE. Here, the RPD values are divided into five levels to explain the model performance: RPD < 1.4 (unreliable), 1.4 ≤ RPD < 1.8 (average), 1.8 ≤ RPD < 2.0 (good), 2.0 ≤ RPD < 2.5 (very good), RPD ≥ 2.5 (excellent).

[0109] 2.4) Comparison of Model Accuracy

[0110] Table 2

[0111]

[0112] (3) Construction of Long-Term Cultivated Land Bare Soil Dataset

[0113] 3.1) Production of Satellite Image Time Series

[0114] The study area is large and the bare soil periods at different geographical locations vary due to climate reasons. Therefore, the time range for screening image data is set according to latitude here to minimize the influence of factors such as snow cover and vegetation cover and obtain as much available remote sensing image data as possible. To ensure full coverage of the study area, the bare soil image dataset is queried and selected every five years, which can achieve a good coverage effect in this area. For areas with data gaps (such as high-altitude, cloudy or snowy areas), the time series extension method is adopted to select adjacent dates to supplement the dataset. After filtering the image dataset, it is integrated and cropped to obtain a long-time series image set for the study area. Finally, the collection of bare soil image data for the study area in different time periods is realized through GEE.

[0115] 3.2) Spectral Transformation

[0116] Through numerical regression, the OLI bands are compared with the TM and ETM+ OLI-2 bands to achieve the conversion of multi-source data into OLI time-series data. With the help of the corresponding spectral response functions, the soil laboratory spectral data are resampled into the TM, ETM+, OLI, and OLI-2 bands. Then, the conversion coefficients between the TM, ETM+, OLI-2, and OLI bands are obtained through polynomial regression fitting. Using these conversion coefficients, the TM, ETM+, and OLI-2 image data in the dataset can be converted into OLI time-series image data, and the fitting parameters of the reflectance between TM and OLI-2, R OLI = a·R TM + b, R OLI = c·R OLI2 + d. Specifically, it can be shown in Table 3 as follows.

[0117] Table 3

[0118]

[0119]

[0120] 3.3) Bare soil pixel extraction

[0121] Combining the spectral characteristics of bare soil pixels and jointly applying the NDVI, NBR2, and VNSIR indices to more effectively identify bare soil pixels. And based on the sample points, the index thresholds are appropriately adjusted. Finally:

[0122]

[0123]

[0124] VNSIR = 1 - [(2 * Red - Green - Blue)+3 * (Swir 2 - Nir)] (15)

[0125] DI 1 = Green - Blue (16)

[0126] DI 2 = Red - Green (17)

[0127]

[0128] 3.4) Calculate the temporal soil spectral reflectance of each bare soil pixel

[0129] After extracting the bare soil pixels from the image datasets at different time periods, the temporal soil spectral reflectance corresponding to the time is obtained. Using the obtained data stack, the per-pixel statistical results of the bare soil in the study area can be calculated, and the percentage of the number of times each pixel has bare soil can also be obtained.

[0130] 3.5) Synthetic soil image

[0131] To reduce the error caused by extreme values, the median statistical method is used to calculate the synthetic reflectance of each pixel. For the past forty years, the data is systematically processed, and finally the synthetic soil image of the study area is generated, and the spectral bands of the original data are accurately retained. The synthetic soil image of the first phase can be as Figure 2 shown.

[0132] 3.6) Cultivated land extraction and processing

[0133] Using the annual land use dataset with a resolution of 30 meters for the past 40 years, the cultivated land area data is obtained year by year to produce a cultivated land mask dataset, so as to identify the stable cultivated land area in the soil synthesis result. The non-cultivated land area is excluded by the cultivated land mask, and finally the cultivated land synthetic soil image dataset of the study area for the past 40 years is obtained.

[0134] (4) Digital mapping of the sand content in cultivated land soil

[0135] For the spectral data of each cultivated land soil synthetic image, calculate the indices required for modeling and supplement the topographic information, and input them into the pre-constructed inversion model of the sand content in soil, and finally obtain the mapping results of the sand content in cultivated land soil of the study area for the past forty years. The mapping results of the sand content in cultivated land soil of the first phase can be as Figure 3 shown.

[0136] (5) Construction of a warning model for cultivated land soil desertification

[0137] 5.1) Selection of warning factors for soil desertification

[0138] 1. First, collect the climate data and socio-economic data required for the warning evaluation model of cultivated land soil desertification. These data include climate factors such as temperature, precipitation, wind speed, and socio-economic factors such as land use intensity. Taking the sand content in cultivated land soil as the mother sequence set and other variables as the sub-sequence set.

[0139] 2. Determine the optimal index set: Select the optimal values of each index to form the optimal index set.

[0140] 3. Standardization processing: Since the units of different indexes are different, it is necessary to convert the values of each index into dimensionless values through standardization processing. Common standardization methods include linear normalization, Z-score standardization, etc.

[0141] 4. Calculate the correlation degree: Calculate the correlation coefficient between each index of each evaluation object and the optimal index. Taking grey correlation analysis as an example, evaluate the similarity between each index of each object and the optimal index. The higher the correlation coefficient, the closer the evaluation object is to the optimal index. For the k-th index of the i-th evaluation object, its correlation degree with the optimal index can be calculated by the following formula:

[0142]

[0143] where, Δ 0k is the difference of the reference sequence (optimal index), Δ ik is the difference of the k-th index of the i-th evaluation object, and ρ is the resolution coefficient.

[0144] 5.2) Determination of the weights of the warning factors for cultivated land soil desertification

[0145] In the stage of determining the weights of the warning factors, various weighting methods such as the Analytic Hierarchy Process (AHP) and the entropy method can be used to weight each warning factor according to the importance of the indexes to quantify its influence in the model. Taking the Analytic Hierarchy Process as an example, the main process includes:

[0146] 1. Construct a judgment matrix: Compare the importance of different warning factors pairwise to construct a judgment matrix.

[0147] 2. Calculate the weights: Calculate the weights by the eigenvalue method, or use the entropy method to determine the weights according to the information entropy of each factor.

[0148] 3. Consistency test: Use the consistency test to verify the rationality of the judgment matrix. If the consistency requirement is not met, the judgment matrix needs to be adjusted and the weights need to be recalculated.

[0149] 4. Weight correction: On the basis of considering expert experience and actual data, make necessary corrections to the calculated weights to ensure the practicality and scientificity of the model.

[0150] 5.3) Establishment of the warning evaluation model for cultivated land soil desertification

[0151] After determining the weights of the factors, a warning evaluation model for cultivated land soil desertification can be established based on the comprehensive evaluation method.

[0152] Multiply the dimensionless values of each warning factor by their weights and sum them up to obtain a comprehensive score, and divide the desertification risk levels (such as mild, moderate, severe) based on the score range. Verify the accuracy of the model through field data, and optimize and adjust the model parameters and weights to ensure its accuracy and reliability in practical applications.

[0153] 5.4) Establishment of the warning model for cultivated land soil desertification

[0154] The establishment of the cultivated land soil desertification early warning model is based on the cultivated land soil desertification early warning evaluation model, and through methods such as time series analysis and machine learning, it predicts the future desertification trend to achieve early warning. Models such as ARIMA and LSTM neural networks can be used to process time series data and predict the future change trend of cultivated land soil desertification. Use the existing soil desertification monitoring data to train the early warning model so that it can identify the impacts of climate factors, land use changes, etc. on soil desertification. Compare the model prediction results with historical data to evaluate the accuracy and stability of the model, and make adjustments according to the error situation. Based on the trained model, output the future trend of soil desertification, and classify the desertification risk within a certain period in the future according to the early warning level, and provide targeted treatment suggestions.

[0155] It should be noted that the following content can be used to replace some of the above steps:

[0156] Regarding remote sensing data sets: In addition to Landsat data, Sentinel data with higher spatial resolution, high-resolution data, and hyperspectral data with higher spectral resolution can also be used as data sources to improve the spatio-temporal resolution of the data set through data fusion.

[0157] Eliminate the differential impacts between different remote sensing sensors: Images taken by different sensors at the same geographical location at the same time can be used for overlapping observations. By analyzing the radiation response relationship of the image data in these overlapping areas, the calibration factors between the sensors can be calculated.

[0158] Dynamically adjust the parameters of the index formula: For the spectral characteristics of different regions, the parameters of the index formula can be dynamically adjusted according to the actual observation situation to cope with the spectral response differences brought about by environmental changes. In addition to constructing dual-band and triple-band indices, methods such as the first derivative and second derivative between dual-bands can also be tried to construct new spectral indices.

[0159] Construction of regression models: In addition to the random forest algorithm, more advanced deep learning models such as DNN, CNN, and LSTM can also be considered, and trained in combination with the spatio-temporal characteristics of remote sensing data to improve the prediction accuracy of the model.

[0160] Bare soil extraction methods: In addition to screening bare soil based on the threshold of a single image index, all available data can be obtained, and by constructing a time series spectral index set, the time series change characteristics can be used to screen bare soil, or the maximum and minimum values of the index in the time series and the differences from other ground objects can be used for bare soil extraction.

[0161] These alternative contents can provide more flexibility and adaptability to meet the soil desertification monitoring needs in different scenarios.

[0162] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present disclosure is not limited by the described action sequence, because according to the present disclosure, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present disclosure.

[0163] The above is the introduction to the method embodiments. The following further illustrates the solution of the present disclosure through device embodiments.

[0164] Figure 4 The structural diagram of a dynamic monitoring device for cultivated land soil desertification based on multi-source satellite remote sensing provided by an embodiment of the present disclosure is shown, as Figure 4 shown, the dynamic monitoring device 400 for cultivated land soil desertification may include:

[0165] An acquisition module 410, configured to acquire the sand grain content data and laboratory spectral data of the surface soil samples of typical soil types collected within the cultivated land scope of the research area through laboratory measurement, and acquire the multi-source satellite spectral data of the sample points based on the passing images during the sample collection period.

[0166] A construction module 420, configured to use the multi-source satellite spectral data of the sample points to extract the spectral features of the common bands therein, construct the dual-band spectral index between any two bands, and add a third band to the dual-band spectral index to construct a triple-band spectral index. Using the constructed spectral index, the corresponding terrain data, and the corresponding bands as input features, and the sand grain content data measured in the laboratory as the target output, construct an inversion data set, and use the inversion data set to train a machine learning model to obtain a soil sand grain content inversion model.

[0167] An extraction module 430, configured to query all available multi-source satellite remote sensing data in all historical periods according to the bare soil period window time corresponding to multiple sub-regions within the research area to obtain the historical remote sensing data set of the research area; extract the bare soil pixels from the remote sensing data in the historical remote sensing data set during the selected bare soil synthesis period, and perform bare soil image synthesis to obtain the bare soil image synthesis result corresponding to the time period, so as to construct a long-term bare soil data set.

[0168] An inversion module 440, configured to perform cultivated land extraction processing on the long-term bare soil data set to obtain a long-term cultivated land bare soil data set, and perform inversion using the soil sand grain content inversion model based on the long-term cultivated land bare soil data set and the corresponding terrain data to obtain the cultivated land soil sand grain content data of the research area.

[0169] A building module 450 is configured to screen out early warning factors for cultivated land soil desertification from climate data, terrain data, socio-economic data, and soil data of a research area based on the sand content data of cultivated land soil, calculate the weights corresponding to the early warning factors for cultivated land soil desertification, and establish an early warning model for cultivated land soil desertification based on the early warning factors for cultivated land soil desertification and their corresponding weights.

[0170] An early warning module 460 is configured to perform early warning of cultivated land soil desertification on the research area by using the early warning model for cultivated land soil desertification.

[0171] It can be understood that Figure 4 each module / unit in the cultivated land soil desertification dynamic monitoring device 400 shown has the function of implementing Figure 1 each step in the cultivated land soil desertification dynamic monitoring method 100 shown, and can achieve its corresponding technical effects. For the sake of brevity, they will not be described in detail here.

[0172] Figure 5 The structure diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure is shown. The electronic device 500 is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 500 may also represent various forms of mobile devices, such as, personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0173] As Figure 5 shown, the electronic device 500 may include a computing unit 501, which may perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 may also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0174] Multiple components in the electronic device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a magnetic disk, an optical disc, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the electronic device 500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0175] The computing unit 501 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 executes the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 can be implemented as a computer program product, including a computer program, which is tangibly contained in a computer-readable medium, such as the storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the method 100 described above can be executed. Alternatively, in other embodiments, the computing unit 501 can be configured to execute method 100 by any other suitable means (e.g., by means of firmware).

[0176] The various embodiments described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs, which can be executed and / or interpreted on a programmable system including at least one programmable processor, the programmable processor can be a dedicated or general-purpose programmable processor, can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0177] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program codes can be executed entirely on the machine, partially on the machine, executed partially on the machine and partially on a remote machine as an independent software package, or executed entirely on a remote machine or server.

[0178] In the context of the present disclosure, a computer-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a computer-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0179] It should be noted that the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute method 100 and achieve the corresponding technical effects achieved by the method of the embodiments of the present disclosure. For the sake of brief description, details are not repeated herein.

[0180] In addition, the present disclosure also provides a computer program product, which includes a computer program that implements method 100 when executed by a processor.

[0181] In order to provide interaction with a user, the above-described embodiments can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or an LCD (liquid crystal display) monitor); and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0182] The embodiments described above can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with embodiments of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected with each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0183] The computer system can include a client and a server. The client and the server are generally far from each other and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, a server of a distributed system, or a server incorporating blockchain.

[0184] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this is not limited herein.

[0185] The above specific embodiments do not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. A method for dynamic monitoring of cultivated land soil desertification based on multi-source satellite remote sensing, characterized in that: The method comprises: Obtain sand content data and laboratory spectral data of surface soil samples of typical soil types collected within the scope of cultivated land in the study area through laboratory measurement, and obtain multi-source satellite spectral data of sample points based on transit images during the sample collection period; Using multi-source satellite spectral data of sample points, the spectral features of the common bands are extracted, a two-band spectral index between arbitrary bands is constructed, and a third band is added to the two-band spectral index to construct a three-band spectral index. The constructed spectral index, the corresponding terrain data and the corresponding bands are used as input features, and the sand content data measured in the laboratory is used as the target output to construct an inversion data set. The inversion data set is used to train the machine learning model to obtain the soil sand content inversion model. According to the bare soil window time corresponding to multiple sub-areas in the study area, all available multi-source satellite remote sensing data in all historical periods are queried to obtain the historical remote sensing data set of the study area; bare soil pixels are extracted from the remote sensing data in the historical remote sensing data set within the selected bare soil synthesis period, and bare soil image synthesis is performed to obtain the bare soil image synthesis results of the corresponding time period, thereby constructing a long-term bare soil data set; The long-term bare soil dataset is processed by arable land extraction to obtain a long-term bare soil dataset. Based on the long-term bare soil dataset and the corresponding terrain data, the soil sand content inversion model is used to perform inversion to obtain the sand content data of the cultivated soil in the study area. Based on the data of cultivated land soil sand content, the cultivated land soil desertification early warning factors were screened from the climate data, topographic data, socio-economic data and soil data of the study area, and the corresponding weights of the cultivated land soil desertification early warning factors were calculated. Based on the cultivated land soil desertification early warning factors and their corresponding weights, the cultivated land soil desertification early warning model was established. The cultivated land soil desertification early warning model is used to provide cultivated land soil desertification early warning in the study area.

2. The method according to claim 1, characterized in that The adding of a third band to the dual-band spectral index to construct a three-band spectral index includes: The correlation coefficient between the two-band index and the sand content data measured in the laboratory was calculated, and a third band was added to the two-band spectral index with a correlation coefficient greater than a preset threshold to construct a three-band spectral index, and the existing spectral index was supplemented in the constructed three-band spectral index.

3. The method according to claim 1, characterized in that The inversion data set is used to train the machine learning model to obtain the soil sand content inversion model, including: Based on the inversion data set, the training set and the validation set are divided. The input features actually used for modeling are screened through the cross-validation algorithm, and the features of the training set are eliminated to retain the input features actually used for modeling. The machine learning model is trained using the updated training set, and hyperparameters are tuned to obtain the soil sand content inversion model.

4. The method according to claim 1, characterized in that The method extracts bare soil pixels from the remote sensing data in the historical remote sensing data set within the selected bare soil synthesis period, synthesizes bare soil images, obtains bare soil image synthesis results for the corresponding time period, and constructs a long-term bare soil data set, including: Based on the laboratory spectral data and the spectral response functions of each satellite, the laboratory spectral data are resampled, the band characteristics of each sensor are simulated, and a spectral response relationship model between different sensors is established to ensure the standardized processing and unified scale conversion of cross-sensor data. The bare soil pixels are extracted from the converted remote sensing data, and bare soil images are synthesized to obtain the bare soil image synthesis results for the corresponding time period, thereby constructing a long-term bare soil dataset.

5. The method according to claim 1, characterized in that The method of performing cultivated land extraction processing on the long time series bare soil dataset to obtain the long time series cultivated land bare soil dataset includes: The cultivated land mask dataset is made using the annual land use dataset, and the cultivated land is extracted from the long-term bare soil dataset based on it to obtain the long-term cultivated land bare soil dataset.

6. The method according to claim 1, characterized in that Based on the cultivated land soil sand content data, the cultivated land soil desertification early warning factors are screened from the climate data, topographic data, socio-economic data, and soil data of the study area, and the weights corresponding to the cultivated land soil desertification early warning factors are calculated, including: Based on the data of sand content in cultivated land soil, the early warning factors of cultivated land soil desertification were screened from the climate data, topographic data, socio-economic data and soil data of the study area through the correlation analysis algorithm. Each cultivated land soil desertification warning factor is compared pairwise, and its relative importance is calculated to obtain the corresponding weight.

7. The method according to claim 1, characterized in that The method of establishing a farmland soil desertification early warning model based on the farmland soil desertification early warning factors and their corresponding weights includes: Based on the cultivated land soil desertification warning factors and their corresponding weights, a cultivated land soil desertification warning evaluation model was established using a comprehensive evaluation algorithm to quantitatively evaluate the cultivated land soil desertification risk, predict the risk level of cultivated land soil desertification, and based on this, establish a cultivated land soil desertification warning model.

8. A dynamic monitoring device for cultivated land soil desertification based on multi-source satellite remote sensing, characterized in that: The device comprises: The acquisition module is used to obtain the sand content data and laboratory spectral data of the surface soil samples of typical soil types collected within the cultivated land of the study area through laboratory measurement, and to obtain the multi-source satellite spectral data of the sample points based on the transit images during the sample collection period; A construction module is used to use multi-source satellite spectral data of sample points to extract spectral features of common bands, construct a two-band spectral index between any bands, and add a third band to the two-band spectral index to construct a three-band spectral index. The constructed spectral index, the corresponding terrain data and the corresponding bands are used as input features, and the sand content data measured in the laboratory is used as the target output to construct an inversion data set. The inversion data set is used to train the machine learning model to obtain the soil sand content inversion model; The extraction module is used to query the available multi-source satellite remote sensing data of all historical periods according to the bare soil period window time corresponding to multiple sub-areas in the study area, and obtain the historical remote sensing data set of the study area; extract bare soil pixels from the remote sensing data in the historical remote sensing data set within the selected bare soil synthesis period, and perform bare soil image synthesis to obtain the bare soil image synthesis result of the corresponding time period, so as to construct a long-term bare soil data set; The inversion module is used to extract the cultivated land from the long-term bare soil dataset to obtain the long-term cultivated land bare soil dataset. The soil sand content inversion model is used to perform inversion based on the long-term cultivated land bare soil dataset and the corresponding terrain data to obtain the cultivated land soil sand content data in the study area. Establish a module for screening cultivated land soil desertification early warning factors from the climate data, topographic data, socio-economic data and soil data of the study area based on cultivated land soil sand content data, and calculate the weights corresponding to the cultivated land soil desertification early warning factors, and establish a cultivated land soil desertification early warning model based on the cultivated land soil desertification early warning factors and their corresponding weights; The early warning module is used to use the cultivated land soil desertification early warning model to provide cultivated land soil desertification early warning in the study area.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to make a computer execute the method according to any one of claims 1 to 7.

Images