Alpine region vegetation irrigation method based on machine learning

Abstract

The invention provides an alpine region vegetation irrigation method based on machine learning, and relates to the technical field of intelligent irrigation, and the method comprises the steps: obtaining environment factor information of a to-be-evaluated region in a target alpine region; according to the environmental factor information, calling a pre-constructed water demand prediction model matched with the vegetation type of the to-be-evaluated region for prediction, and obtaining monthly water demand information of the to-be-evaluated region; performing actual water profit and loss analysis based on the predicted monthly water demand information, and determining and implementing an irrigation strategy according to an analysis result; the water demand prediction model is constructed based on a machine learning algorithm. According to the method, the water demand prediction model is constructed by using the machine learning algorithm, the complex nonlinear relationship among multiple environment variables can be accurately captured, the accuracy of estimating the ecological water demand of vegetation in the alpine region is improved, and more effective irrigation management can be realized.

Description

Technical Field

[0001] This application relates to the field of intelligent irrigation technology, specifically to a method for irrigating vegetation in high-altitude and cold regions based on machine learning. Background Technology

[0002] High-altitude and cold regions, due to their unique climatic conditions and complex ecosystems, have become a key area for research on ecological water demand. The cold climate, uneven spatial and temporal distribution of precipitation, and frequent soil freezing and thawing processes in these regions lead to significant differences in the ecological water demand patterns of vegetation compared to other areas. Current technologies for estimating vegetation ecological water demand primarily rely on remote sensing and meteorological data, combined with the Penman-Monteith equation. Some studies have optimized the estimation results by introducing vegetation coefficients and soil moisture limiting coefficients, but these methods generally suffer from the following problems: The dramatic climate changes, freeze-thaw cycles, and short growing seasons in high-altitude and cold regions significantly impact vegetation water demand, and existing methods struggle to accurately characterize the mechanisms of these environmental variables; existing methods mostly employ statistical methods such as linear regression and structural equation modeling, which are insufficient to accurately simulate the complex nonlinear relationships among multiple variables such as climate, site conditions, and soil factors; and existing methods often rely on empirical formulas or local experimental data for calibration, resulting in poor universality and difficulty in rapidly applying them to ecological water demand assessment and irrigation management in different high-altitude and cold regions. Summary of the Invention

[0003] To at least partially overcome the problems existing in related technologies, this application provides a method for vegetation irrigation in high-altitude and cold regions based on machine learning. The method uses machine learning to learn the impact of various environmental variables on ecological water demand in the target high-altitude and cold region, and then conducts irrigation management based on this.

[0004] Some embodiments of this application provide a machine learning-based vegetation irrigation method for high-altitude cold regions, the method comprising: Obtain environmental factor information for the area to be evaluated in the target high-altitude and cold region; Based on the environmental factor information, a pre-built water demand prediction model that matches the vegetation type of the area to be evaluated is invoked to make a prediction and obtain the monthly water demand information of the area to be evaluated. Based on the predicted monthly water demand information, conduct actual water surplus and deficit analysis, determine irrigation strategies based on the analysis results, and implement them. The water demand prediction model is constructed based on machine learning algorithms.

[0005] In some possible implementations, the pre-construction process of the water demand prediction model includes: Historical multi-source environmental data of the target high-altitude cold region are collected and preprocessed. The historical multi-source environmental data includes remote sensing data, meteorological data, site condition data, and soil data. Based on the preprocessed data, vegetation ecological water demand data that is spatiotemporally matched with historical multi-source environmental data is calculated, and key influencing environmental factors are identified based on the analysis of the impact of various types of environmental data on vegetation ecological water demand. A machine learning model is constructed, using the key environmental factors as input features and vegetation ecological water demand as the target variable. The model is trained based on the preprocessed data and the vegetation ecological water demand data to obtain the water demand prediction model.

[0006] In some possible implementations, principal component analysis is used to analyze the impact of various types of environmental data on vegetation ecological water demand.

[0007] In some possible implementations, the preprocessing process includes: cropping and resampling to ensure that the historical multi-source environmental data meets the spatial resolution and temporal resolution requirements of the target high-altitude and cold regions.

[0008] In some possible implementations, the machine learning model is constructed using a random forest model, a gradient boosting tree model, or an XGBoost model.

[0009] In some possible implementations, the key environmental factors include: temperature, precipitation, soil thickness, DEM, slope, and soil texture.

[0010] In some possible implementations, the remote sensing data includes vegetation coverage information data, vegetation type information data, and vegetation growth dynamic monitoring data; The types of meteorological data include temperature, precipitation, wind speed, net radiation, and relative humidity; The types of site condition data include elevation, slope, and aspect; The types of soil data include soil texture, thickness, organic matter content, and soil moisture retention characteristics.

[0011] In some possible implementations, vegetation ecological water requirement data are calculated based on the following expression: in, This indicates the ecological water requirement of vegetation. This represents the plant evapotranspiration calculated based on the Penman formula. Indicates the soil moisture limiting coefficient. This represents the vegetation coefficient.

[0012] In some possible implementations, the process of performing actual water surplus / deficit analysis based on predicted monthly water demand information includes: determining the water deficit based on the difference between the predicted monthly water demand information and the actual precipitation.

[0013] In some possible implementations, determining the irrigation strategy based on the analysis results includes: determining the irrigation threshold of the irrigation strategy based on the water deficit and the vegetation water demand characteristics of the area to be evaluated.

[0014] The machine learning-based vegetation irrigation method for high-altitude cold regions provided in this application obtains environmental factor information of the area to be evaluated in the target high-altitude cold region, and then uses this information to call a pre-constructed water demand prediction model that matches the vegetation type of the area to be evaluated to make predictions, thereby obtaining the monthly water demand information of the area to be evaluated. This method uses machine learning algorithms to construct a water demand prediction model, which can accurately capture the complex nonlinear relationships between multiple environmental variables such as climate, site conditions, and soil factors. It effectively overcomes the problem that existing technologies that rely on statistical methods such as linear regression and structural equation modeling are difficult to accurately simulate these relationships, improves the accuracy of estimating the ecological water demand of vegetation in high-altitude cold regions, and thus facilitates more effective irrigation management.

[0015] Other advantages, objectives, and features of this application will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from an examination of the following, or may be learned from the practice of the invention. Attached Figure Description

[0016] The accompanying drawings are used to provide a further understanding of the technical solutions of this application or the prior art, and constitute a part of the specification. The drawings illustrating embodiments of this application, together with the embodiments of this application, are used to explain the technical solutions of this application, but do not constitute a limitation on the technical solutions of this application.

[0017] Figure 1 A schematic flowchart of a machine learning-based vegetation irrigation method for high-altitude and cold regions provided in one embodiment of this application; Figure 2 This is a flowchart illustrating the construction process of a water demand prediction model in one embodiment of this application. Figure 3 This is a schematic diagram illustrating the structure of an implementation system for a machine learning-based vegetation irrigation method in high-altitude and cold regions, as described in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be described in detail below. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] As described in the background section, high-altitude and cold regions, due to their unique climatic conditions and complex ecosystems, have become a key area for research on ecological water demand. The cold climate, uneven spatial and temporal distribution of precipitation, and frequent soil freezing and thawing processes in high-altitude and cold regions lead to significant differences in the ecological water demand patterns of vegetation compared to general areas. Current techniques for estimating vegetation ecological water demand mainly rely on remote sensing and meteorological data, combined with the Penman-Monteith formula. Some studies have optimized the estimation results by introducing vegetation coefficients and soil moisture limiting coefficients, but these methods generally suffer from the following problems: The dramatic climate changes, freeze-thaw cycles, and short growing seasons in high-altitude and cold regions significantly impact vegetation water demand, and existing methods struggle to accurately characterize the mechanisms of these environmental variables; existing methods mostly employ statistical methods such as linear regression and structural equation modeling, which are insufficient to accurately simulate the complex nonlinear relationships between multiple variables such as climate, site conditions, and soil factors; and existing methods often rely on empirical formulas or local experimental data for calibration, resulting in poor universality and difficulty in rapidly applying them to ecological water demand assessment and irrigation management in different high-altitude and cold regions. Based on this, this application proposes a machine learning-based vegetation irrigation method for high-altitude and cold regions. The method uses machine learning to learn about the impact of various environmental variables on ecological water demand in the target high-altitude and cold region, and then conducts irrigation management accordingly.

[0020] like Figure 1 As shown, in one embodiment, the machine learning-based vegetation irrigation method for high-altitude and cold regions proposed in this application includes: Step S110: Obtain environmental factor information for the area to be evaluated in the target high-altitude region. Specifically, for example, the target high-altitude region is the Qinghai-Tibet Plateau, and the area to be evaluated is a certain area A in this region. The environmental factor information includes multi-dimensional information such as meteorological factors, site factors, and soil factors, such as temperature and precipitation information (corresponding to meteorological factors), soil thickness and soil texture information (corresponding to soil factors), digital elevation model (DEM) information and slope information (corresponding to site factors). And it is easy to understand that this information can be obtained through satellite remote sensing, on-site observation, field surveys, etc.

[0021] Next, step S120 is performed, in which a pre-built water demand prediction model matching the vegetation type of the area to be evaluated is called to predict the monthly water demand information of the area to be evaluated based on the environmental factor information. It should be noted that in the technical scenario of this application, the vegetation type is classified based on ecological restoration considerations, such as restoring grassland or shrubland; different vegetation types correspond to different water demand prediction models. The environmental factor information corresponding to the area to be evaluated (such as area A) is input into the corresponding type of water demand prediction model for prediction processing. Based on the characteristics of the water demand prediction model itself, the monthly water demand information of the area to be evaluated can be output, such as the monthly water demand of 253 cubic meters. The water demand prediction model here is built based on machine learning algorithms. For example, random forest, gradient boosting tree, or XGBoost algorithms can be used. Relevant historical data required for prediction and evaluation are selected to build and train the model. This process will be illustrated later and will not be detailed here.

[0022] Continue back Figure 1 Then proceed to step S130, where an actual water surplus / deficit analysis is performed based on the predicted monthly water demand information from step S120, and an irrigation strategy is determined and implemented based on the analysis results.

[0023] Specifically, in some implementations, the process of conducting actual water surplus and deficit analysis based on predicted monthly water demand information includes: determining the water deficit based on the difference between the predicted monthly water demand information and the actual precipitation; that is, after obtaining the predicted monthly water demand information, comparing and analyzing it with the actual precipitation data of the target area in the same period, calculating the difference between the predicted monthly water demand and the actual recorded precipitation, thereby determining the water deficit in a specific time period.

[0024] Furthermore, in one embodiment, determining an irrigation strategy based on the analysis results includes: determining the irrigation threshold for the irrigation strategy based on the water deficit and the vegetation water demand characteristics of the area to be assessed. This not only takes into account the supporting role of natural precipitation in vegetation growth but also quantifies the additional water required to meet the healthy growth of vegetation. In this way, the water supply and demand status of the area to be assessed can be more accurately evaluated at different times, potential water shortage risks can be identified, and targeted irrigation strategies can be formulated based on these analysis results to ensure that the growth needs of vegetation are fully met while conserving water resources.

[0025] The machine learning-based vegetation irrigation method for high-altitude cold regions provided in this application obtains environmental factor information of the area to be evaluated in the target high-altitude cold region, and then uses this information to call a pre-constructed water demand prediction model that matches the vegetation type of the area to be evaluated to make predictions, thereby obtaining the monthly water demand information of the area to be evaluated. This method uses machine learning algorithms to construct a water demand prediction model, which can accurately capture the complex nonlinear relationships between multiple environmental variables such as climate, site conditions, and soil factors. It effectively overcomes the problem that existing technologies that rely on statistical methods such as linear regression and structural equation modeling are difficult to accurately simulate these relationships, improves the accuracy of estimating the ecological water demand of vegetation in high-altitude cold regions, and thus facilitates more effective irrigation management.

[0026] To facilitate understanding of the technical solution of this application, the pre-construction process of the water demand prediction model in this application will be introduced and explained below.

[0027] like Figure 2 As shown, in some embodiments, the pre-construction process of the water demand prediction model in this application includes: Step S210 collects and preprocesses historical multi-source environmental data of the target high-altitude cold region. The historical multi-source environmental data includes remote sensing data, meteorological data, site condition data and soil data.

[0028] Specifically, remote sensing data includes vegetation cover information, vegetation type information, and vegetation growth dynamic monitoring data; meteorological data includes temperature, precipitation, wind speed, net radiation, and relative humidity; site condition data includes altitude, slope, and aspect; and soil data includes soil texture, thickness, organic matter content, and soil moisture retention characteristics.

[0029] Specifically, some of the data sources mentioned above are shown in Table 1 below: Table 1. List of Data Types and Sources The preprocessing process in this application mainly includes: cropping and resampling to ensure that historical multi-source environmental data meets the spatial and temporal resolution requirements of the target high-altitude and cold regions, such as unifying the spatial resolution to 1km and the temporal resolution to monthly. For example, in actual implementation, meteorological station data, including temperature, precipitation, wind speed, net radiation, and relative humidity, can be further collected, and meteorological reanalysis data (ERA5, GLDAS) can be used to supplement areas with insufficient spatial resolution.

[0030] The preprocessing process also includes calibration and verification based on the acquired information such as soil texture, thickness, and organic matter content, combined with remote sensing soil moisture products (SMAP, ESA CCI).

[0031] Furthermore, it should be noted that, considering the application scenarios of this application, the spatiotemporal variation characteristics of vegetation in high-altitude and cold regions can be analyzed based on the vegetation cover (characterized by NDVI) and vegetation type information in the aforementioned data. Combined with multi-temporal data, the growth dynamics of vegetation can be monitored to further obtain information on key growth periods. Additionally, the impact of topographic data such as altitude, slope, and aspect on vegetation water requirements can be analyzed.

[0032] Step S220: Based on the preprocessed data, calculate the vegetation ecological water demand data that matches the historical multi-source environmental data in time and space, and determine the key influencing environmental factors based on the analysis of the impact of various types of environmental data on vegetation ecological water demand. It is readily understood that changes in vegetation ecological water demand are influenced by a variety of environmental factors, including climatic factors (such as temperature and precipitation), soil properties, and site conditions. To comprehensively assess the contribution of each environmental factor to ecological water demand, and considering engineering practicalities, this application identifies key influencing environmental factors based on an analysis of the impact of various types of environmental data on vegetation ecological water demand. Specifically, as a practical implementation method, principal component analysis (PCA) can be used to analyze the impact of various types of environmental data on vegetation ecological water demand. Dimensionality reduction techniques reduce redundant information in the data, compressing multiple environmental factors into a few principal components, thus retaining the main information affecting ecological water demand. This process helps extract the most representative environmental factors and reduces the complexity of subsequent analyses.

[0033] For example, in some embodiments, the types of key environmental factors identified are: temperature, precipitation, soil thickness, DEM, slope, and soil texture.

[0034] Next, step S230 is performed to construct a machine learning model. This model uses key environmental influencing factors as input features and vegetation ecological water demand as the target variable. It is trained based on preprocessed data and vegetation ecological water demand data to obtain a water demand prediction model. As mentioned earlier, the machine learning model can be constructed using a random forest model, a gradient boosting tree model, or an XGBoost model. For example, in this embodiment, the random forest algorithm is used to construct the water demand prediction model.

[0035] Random Forest is an ensemble learning algorithm, an extension of the bagging method. It improves the accuracy and robustness of a model by constructing multiple decision trees and combining their predictions. Its core idea is to "integrate multiple weak learners to form a strong learner," and it is widely used in machine learning tasks such as classification, regression, and feature selection. This application primarily focuses on a regression application. The Random Forest algorithm itself is a well-known existing technology, and its training methods, parameter tuning, and other implementation principles can be found in existing publicly available technical materials. This application focuses on explaining the dataset construction process in the implementation of the technical solution.

[0036] As mentioned above, the input features correspond to preprocessed historical multi-source environmental data, specifically including temperature, precipitation, soil thickness, DEM, slope, and soil texture. The target variable corresponds to vegetation ecological water demand, which is calculated in this application based on the following expression: (1) In expression (1), This indicates the ecological water requirement of vegetation. This represents the plant evapotranspiration calculated based on the Penman formula. Indicates the soil moisture limiting coefficient. This represents the vegetation coefficient.

[0037] Furthermore, plant transpiration Calculated based on the following expression, (2) In expression (2), Δ represents plant evapotranspiration (mm / d); Δ represents the slope of the saturated vapor pressure curve (kPa / °C). For net radiation (MJ / m²·d), G is soil heat flux density (MJ / m²·d), which is usually ignored in the non-humid season; γ is the gas constant (kPa / °C); T is the air temperature (°C). The wind speed at a height of 2 meters (m / s); and These are the saturated vapor pressure and the actual vapor pressure (kPa), respectively, calculated using air temperature and relative humidity.

[0038] Specifically, the saturated vapor pressure is calculated using the air temperature (T) based on the following expression: (3) Specifically, the actual vapor pressure is calculated based on the relative humidity (RH) and saturated vapor pressure using the following expression: (4) Specifically, the wind speed at a height of 2 meters is estimated based on the following expression: (5) In expression (5), At a height of 2 meters, Z1 and Z2 represent the wind speed at a known height (e.g., 10 meters or other), where Z1 and Z2 are the heights (in meters) of the known wind speed measurement point and the target height, respectively. α is the wind speed index, which is usually related to surface roughness. In practical applications, the value of the wind speed index α depends on the surface roughness. For rough surfaces such as mountainous areas and forests, α is usually taken as 0.14, a value verified in multiple studies.

[0039] In the above calculations, meteorological data include air temperature (T), wind speed (U2), relative humidity (RH), and net radiation (R). n (See Table 1 for examples.) The original sources of these data are the National Meteorological Science Data Center and the Tibetan Plateau Science Data Center, among others. All data underwent cropping and resampling to adapt to the spatial resolution (1 km) and temporal requirements of the study area.

[0040] Specifically, regarding the soil moisture limiting coefficient in expression (1) It is determined based on the following expression: (6) In expression (6), θ represents the actual soil moisture content (unit: mm), θz represents the permanent wilting point, and θc represents the critical water content (usually 75% of field capacity). The original sources of these data are shown in Table 1.

[0041] Specifically, vegetation coefficients are typically estimated using remote sensing data, particularly through the Normalized Difference Vegetation Index (NDVI) to reflect the growth status of vegetation. Changes in vegetation coefficient values ​​are directly related to vegetation growth status and water requirements. In this application, regarding the vegetation coefficient in expression (1)... It is determined based on the following expression: (7) In expression (7), NDVI is the normalized vegetation index of the actual vegetation, reflecting the growth status of the vegetation; NDVI min (5th percentile) and NDVI max (95th percentile) represents the NDVI values ​​for the bare soil area and the fully covered area, respectively.

[0042] Thus, as described above, in the process of Figure 2After completing the construction process of the water demand prediction model shown, the required assessment and prediction model can be obtained. In this process, through the preprocessing of historical multi-source environmental data and the determination of key influencing environmental factors, the model not only considers the impact of factors such as drastic climate change, freeze-thaw cycles, and short growing seasons unique to high-altitude and cold regions, but also can quickly adapt to the ecological water demand assessment and irrigation management needs of different high-altitude and cold regions, thus enhancing the model's versatility and applicability.

[0043] In terms of the overall solution, in practical application, the actual water deficit analysis based on predicted monthly water demand information can accurately identify water shortages. Combined with the water demand characteristics of vegetation in the area to be assessed, the irrigation threshold for the irrigation strategy is determined, enabling the scientific formulation and efficient implementation of irrigation strategies. This helps improve water resource utilization efficiency, ensures the healthy growth of vegetation in high-altitude and cold regions, and promotes ecological environmental protection. Therefore, this invention not only solves the problems existing in the prior art but also demonstrates significant technical effects in improving the accuracy of vegetation ecological water demand estimation and optimizing irrigation management.

[0044] The shortcomings of the above solutions are the result of the inventor's practical experience and careful research. Therefore, the discovery process of the above problems and the solutions proposed in this application below should be considered the inventor's contributions to this application. Furthermore, in some embodiments, the process of water deficit analysis and irrigation threshold determination includes: After completing the water demand forecast, a further water supply and demand analysis is conducted, which includes the following steps: The Water Deficit Index (WBI) is defined as ETc - P, where P is the precipitation during the same period. When WBI > 0, it indicates a water deficit, requiring artificial water replenishment. Furthermore, by combining the precipitation and vegetation growth cycles of spring, summer, and autumn in high-altitude regions, the water supply and demand relationship in different months can be analyzed to identify key water replenishment periods.

[0045] Furthermore, dynamic irrigation thresholds can be set based on model prediction results. For example, when the predicted ETC exceeds the local water resource carrying capacity or WBI exceeds a certain critical value (such as 30 mm / month), an irrigation early warning mechanism can be triggered. In addition, differentiated irrigation strategies can be formulated according to the water demand characteristics of different vegetation types (such as alpine meadows, shrublands, and coniferous forests), such as adjusting irrigation frequency and irrigation quotas.

[0046] Based on the above embodiments, in practical applications, the technical solution of this application can be realized through system integration and intelligent irrigation management, that is, by deploying the above model into an ecological protection and agricultural water management system in high-altitude and cold regions; specifically, as shown in... Figure 3 As shown, the system includes: Data access module 301: used to access multi-source data such as weather stations, remote sensing platforms, and soil sensors in real time, and to achieve dynamic updates of input; Model prediction module 302: Used to call the trained machine learning model to automatically predict the ETC in the region and generate a water balance map.

[0047] Decision support module 303: Used to determine whether to start irrigation equipment based on irrigation thresholds and to provide irrigation suggestions (such as irrigation time, water volume, etc.).

[0048] Automation control module 304: Used to connect to water-saving irrigation systems such as drip irrigation and sprinkler irrigation to achieve remote control and precision irrigation.

[0049] pass Figure 3 The system shown can achieve real-time monitoring, prediction, and scheduling of ecological water demand in high-altitude and cold regions, significantly improving water resource utilization efficiency and ecosystem stability. In summary, this invention integrates multi-source data such as remote sensing, meteorology, topography, and soil data to construct a machine learning-based ecological water demand prediction model. Combined with water surplus / deficit analysis, it formulates scientific irrigation thresholds and strategies, achieving intelligent and precise management of vegetation irrigation in high-altitude and cold regions.

[0050] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0051] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0052] It should be noted that in the description of this application, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means at least two.

[0053] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0054] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A machine learning-based alpine region vegetation irrigation method, characterized by, The method comprises the following steps: acquiring environmental factor information of a target alpine region to be evaluated; based on the environmental factor information, calling a pre-constructed water demand prediction model matched with the vegetation type of the region to be evaluated to make a prediction, and obtaining monthly water demand information of the region to be evaluated; based on the predicted monthly water demand information, performing actual water balance analysis, determining an irrigation strategy according to the analysis result, and implementing the irrigation strategy; wherein the water demand prediction model is constructed based on a machine learning algorithm.

2. The machine learning based alpine region vegetation irrigation method of claim 1, wherein, The pre-construction process of the water demand prediction model comprises the following steps: collecting historical multi-source environmental data of the target alpine region and performing preprocessing, wherein the historical multi-source environmental data comprises remote sensing data, meteorological data, site condition data and soil data; based on the preprocessed data, calculating vegetation ecological water demand data matched with the historical multi-source environmental data in time and space, and determining key environmental factors based on an analysis of the influence degree of each type of environmental data on the vegetation ecological water demand; constructing a machine learning model, taking the key environmental factors as input features and the vegetation ecological water demand as a target variable, and training the model based on the preprocessed data and the vegetation ecological water demand data to obtain the water demand prediction model.

3. The machine learning based alpine region vegetation irrigation method of claim 2, wherein, The influence degree of each type of environmental data on the vegetation ecological water demand is analyzed by principal component analysis.

4. The machine learning based alpine region vegetation irrigation method of claim 2, wherein, The preprocessing process comprises the following steps: performing cropping and resampling processing to make the historical multi-source environmental data meet the spatial resolution requirement and the time resolution requirement of the target alpine region.

5. The machine learning based alpine region vegetation irrigation method of claim 2, wherein, The machine learning model is constructed by selecting a random forest model, a gradient boosting tree model or an XGBoost model.

6. The machine learning based alpine region vegetation irrigation method of claim 2, wherein, The types of the key environmental factors are as follows: air temperature, precipitation, soil thickness, DEM, slope and soil texture.

7. The machine learning-based alpine region vegetation irrigation method according to claim 6, wherein the remote sensing data comprises vegetation coverage information data, vegetation type information data and vegetation growth dynamic monitoring data; the types of the meteorological data comprise air temperature, precipitation, wind speed, net radiation and relative humidity; the types of the site condition data comprise altitude, slope and slope direction; the types of the soil data comprise soil texture, thickness, organic matter content and soil water holding characteristics.

8. The machine learning based alpine region vegetation irrigation method of claim 6, wherein, The vegetation ecological water demand data is calculated based on the following expression: wherein, represents the ecological water requirement of vegetation, represents the plant transpiration amount determined based on the Penman formula calculation, represents the soil moisture limiting coefficient, represents the vegetation coefficient.

9. The machine learning based alpine region vegetation irrigation method of claim 1, wherein, The process of performing actual water balance analysis based on the predicted monthly water demand information comprises the following steps: determining a water gap based on the difference between the predicted monthly water demand information and the actual precipitation.

10. The machine learning based alpine region vegetation irrigation method of claim 9, wherein, The process of determining an irrigation strategy based on the analysis result comprises the following steps: determining an irrigation threshold of the irrigation strategy based on the water gap and the vegetation water demand characteristics of the region to be evaluated.

Images