Green and ecological planting methods for daylilies

Abstract

This invention discloses a green and ecological planting method for daylilies, belonging to the field of agricultural planting technology. It addresses the technical problems of excessive reliance on chemical fertilizers and traditional mulching in current daylily cultivation, leading to soil ecological imbalance and low fertilizer utilization. The method includes: applying organic fertilizer made from cow and sheep manure and Asteraceae plant residues through aerobic composting, using a layered strip application method. First, fully decomposed organic fertilizer is applied as a base fertilizer layer in the planting furrow, followed by semi-decomposed organic fertilizer that has completed the main fermentation stage but not the later aging stage as a middle layer fertilizer; after land preparation and ridging, wood ash is spread on the ridge surface, followed by the laying of black fully biodegradable mulch film; finally, holes are made in the mulch film for daylily seedling planting. This method is mainly used for sustainable ecological planting of daylilies, improving soil fertility and structure, regulating soil pH and providing potassium, conserving moisture and suppressing weeds while avoiding residual pollution, thereby increasing daylily yield and quality while reducing environmental impact.

Description

Technical Field

[0001] This invention belongs to the field of agricultural planting technology and relates to a green and ecological planting method for daylilies. Background Technology

[0002] In the traditional cultivation practices of daylilies, there are some technical aspects that affect their sustainable production and ecological benefits, which need to be further optimized and improved.

[0003] Firstly, in terms of soil fertilization and fertilization management, conventional planting methods typically rely on chemical fertilizers or untreated farmyard manure. Long-term, excessive application of chemical fertilizers can easily lead to soil compaction, decreased organic matter content, and may disrupt the balance of the soil microbial community, affecting the health of daylily roots and nutrient absorption efficiency. Using ordinary farmyard manure, if not fully decomposed, may introduce weed seeds, pathogens, and insect eggs, posing a risk of seedling burn. Furthermore, its nutrient release rate is difficult to match with crop needs, resulting in low fertilizer utilization. How to construct an organic fertilizer application scheme that can continuously supply nutrients, improve soil structure, and ensure safety and harmlessness is a practical problem in production. In addition, conventional fertilization methods are mostly surface application or one-time furrow application. This extensive approach may lead to uneven nutrient distribution in the soil, or shallow fertilization may fail to meet the needs of crop growth in the later stages, especially in soils with poor water and fertilizer retention capacity, where the sustainability of fertilizer effectiveness faces challenges.

[0004] Secondly, in the areas of ground cover and weed control, the widespread use of polyethylene (PE) plastic film has become a common measure for moisture retention, warming, and weed suppression. However, PE film is difficult to degrade under natural conditions, and the residual fragments accumulate year by year, causing serious "white pollution," damaging the soil's physical structure, hindering water and nutrient transport, and even affecting the growth of subsequent crop rotations. Although biodegradable films are already in use, balancing their degradation performance, strength, and cost, especially ensuring that they can suppress weeds and retain moisture over a sufficiently long period while achieving complete and harmless biodegradation, remains a technical challenge. Furthermore, while the complete enclosure of ordinary plastic film can effectively suppress weeds, it may also hinder gas exchange between the soil and the atmosphere to some extent, affecting root respiration and soil microbial activity.

[0005] Furthermore, in terms of field water management, traditional irrigation relies heavily on growers' experience or fixed schedules, lacking precision. This model struggles to respond accurately and in real-time to changes in soil moisture dynamics and the actual water requirements of crops. Over-irrigation not only wastes precious water resources but can also lead to nutrient leaching and increase the risk of disease; while under-irrigation directly restricts the growth and bud formation of daylilies, affecting yield and quality. Achieving precision irrigation requires comprehensive consideration of multiple factors, including real-time soil moisture, meteorological conditions, and crop growth stages. However, in practice, how to economically and reliably acquire this multi-source, heterogeneous data (such as the spatial variability of soil moisture at the field scale and the uncertainty of future weather) and effectively integrate it into specific, actionable irrigation decision instructions has been a challenge. Single-type data (such as relying on only a few soil sensors) is often insufficient to comprehensively and accurately reflect the overall condition of the field, and without the support of intelligent decision-making models, even if data is obtained, it is difficult to formulate dynamically optimized irrigation plans.

[0006] In summary, existing technologies have certain limitations in areas such as ecological fertilization, sustainable ground cover management, and data-driven intelligent precision irrigation for daylilies. These limitations restrict the development of the daylily industry towards higher resource utilization and lower environmental impact. Summary of the Invention

[0007] One object of the present invention is to solve at least the above-mentioned problems and / or defects, and to provide at least the advantages described below.

[0008] Another objective of this invention is to provide a green and ecological method for cultivating daylilies.

[0009] To address the technical problems of excessive reliance on chemical fertilizers and non-degradable plastic film in the current daylily cultivation process, which leads to soil ecological imbalance, low fertilizer utilization efficiency, and agricultural non-point source pollution.

[0010] This addresses the technical challenges of traditional irrigation methods that rely on manual experience, lack precise decision support, make it difficult to achieve on-demand water supply, and lead to water waste or insufficient irrigation that affects the yield and quality of daylilies.

[0011] Therefore, the technical solution provided by this invention is as follows: A green and ecological method for cultivating daylilies includes the following steps: 1) Fertilization: Apply organic fertilizer made from cow and sheep manure and Asteraceae plant residues through aerobic composting; wherein, the Asteraceae plant residues are derived from common Asteraceae plants within the county-level administrative region where the planting site is located; 2) Land preparation and mulching: After preparing the land and forming planting ridges, first spread a layer of wood ash with a thickness of 2-3cm evenly on the soil surface of the ridges; then lay a black fully biodegradable mulch film, which has micro-ventilation holes arranged in a rectangular array. The pore diameter of the micro-ventilation holes is 0.5-1.0mm and the pore density is 300-500 per square meter. 3) Transplanting: Make holes in the plastic film to transplant daylily seedlings; In step 1), the fertilization operation adopts the layered strip application method, which is as follows: In the planting trench dug in the planting ridge, first apply organic fertilizer that has passed the high temperature decomposition period and is fully decomposed as the base fertilizer layer, cover it with soil 3-5cm, then apply organic fertilizer that has passed the main fermentation stage but has not completed the later aging as the middle layer fertilizer, and then cover it with soil and plant the seedlings.

[0012] Preferably, the green and ecological daylily planting method further includes field management, wherein the irrigation step in the field management is controlled by a pre-trained irrigation decision model, specifically including the following operational steps: a) Multi-source data acquisition and fusion: real-time soil temperature and humidity data collected by IoT sensor nodes pre-embedded in the planting ridges; simultaneously, real-time meteorological data and regional weather forecast data for the next 72 hours are acquired by accessing a meteorological data platform through an application programming interface; and multi-source satellite remote sensing data are accessed through a satellite data service interface, the data including at least thermal infrared surface temperature data and microwave surface humidity data for retrieving regional soil moisture status, and normalized vegetation index data for monitoring crop growth. b) Irrigation decision generation: The multi-source heterogeneous data obtained in step a) is spatiotemporally aligned and fused, and then input into the irrigation decision model along with the pre-set water requirement threshold parameters that match the current growth stage of daylilies. The model is trained using a hybrid deep learning algorithm that combines convolutional neural networks and long short-term memory networks, with historical irrigation data, crop growth data and corresponding yield and quality data as labels. It is used to output decision instructions that include specific irrigation time, duration and water volume. c) Command issuance and execution: The decision command generated in step b) is sent to the irrigation controller deployed in the field via a wireless Internet of Things protocol, and the controller drives the solenoid valve of the drip irrigation system to execute the quantitative irrigation.

[0013] Preferably, in the aforementioned green and ecological daylily cultivation method, the daylily seedlings are obtained through the following hybridization and breeding methods: 3.1) Parent selection: Wild daylilies growing at altitudes above 800 meters were selected as the female parent, and superior cultivated varieties with strong disease resistance and a large number of inflorescences were selected as the male parent; 3.2) Artificial hybridization: During the peak flowering period of the female parent, take the healthy and tender flower buds that have opened on the same day, remove the male parts, pollinate them with the new fresh pollen of the male parent, and then isolate them by bagging. 3.3) Seed treatment and sowing: Harvest the hybrid seeds, disinfect and soak them in warm water, and then sow them in a seedling substrate made of organic fertilizer and peat in a volume ratio of 1:3 as described in claim 1; 3.4) Seedling selection and transplanting: When the seedlings have grown to 4-5 true leaves, select strong and healthy individual plants with well-developed root systems and transplant them to the nursery. The following year, based on the plant's resistance, number of inflorescences and bud morphology, select the best-performing strains as seedlings for the planting steps described above.

[0014] Preferably, the pre-training method for the irrigation decision model in the aforementioned green and ecological daylily planting method specifically includes the following steps: Historical datasets are collected and integrated, including: historical multi-source heterogeneous data, which contains historical soil temperature and humidity time-series data, historical meteorological data, and historical satellite remote sensing data, wherein the satellite remote sensing data includes at least historical thermal infrared band surface temperature data, microwave band soil surface humidity data, and normalized difference vegetation index data; historical irrigation operation records aligned with the historical multi-source heterogeneous data in time and space; and historical daylily yield and quality index data corresponding to the historical irrigation operation records. The constructed historical dataset is cleaned, missing values ​​are imputed, and outliers are handled. Then, spatiotemporal alignment and fusion are performed to unify data from different sources and formats into the same spatiotemporal grid. Finally, features are extracted and constructed to generate model input feature vectors containing spatiotemporal features and crop growth stage identifiers. A two-stream hybrid neural network model is constructed, where the first stream is a convolutional neural network (CNN) branch, used to extract spatial distribution features and texture features from the fused spatialized data; the second stream is a long short-term memory network (LSTM) branch, used to extract time-series dependent features from the historical soil temperature and humidity time-series data and historical meteorological time-series data; the high-level features output by the CNN branch and the LSTM branch are concatenated and fused, and then mapped to the irrigation decision action space through a fully connected layer; The preprocessed training dataset is divided into a training set, a validation set, and a test set. The historical irrigation operation records are used as action labels, and the historical yield and quality index data are used as auxiliary supervision signals. A composite loss function, which includes mean squared error loss and a multi-objective reward function based on yield-water use efficiency, is used to perform end-to-end supervised training and optimization of the hybrid deep learning model. Early shutdown and hyperparameter tuning are performed through the validation set to finally obtain the pre-trained irrigation decision model.

[0015] Preferably, in the aforementioned green and ecological daylily planting method, step b) irrigation decision generation further includes a data disaster recovery and degradation decision module: When real-time soil temperature and humidity data or key satellite remote sensing data are continuously missing for more than a preset threshold, the irrigation decision model automatically switches to a degraded decision mode. In the downgraded decision-making mode, the model generates irrigation decision instructions based on the last effective soil data, historical meteorological data for the same period, and the 24-hour weather forecast obtained through the application programming interface, combined with the water requirement experience model of the current growth stage of daylilies.

[0016] Preferably, the green and ecological daylily planting method further includes an online optimization step for irrigation decision model parameters: At the end of each growing season, the system collects complete actual irrigation execution data, corresponding multi-source monitoring data, and final yield and quality data for that season to form a new set of training samples. These new samples are used to incrementally learn or fine-tune the pre-trained irrigation decision model, with a focus on optimizing the water demand threshold parameter, so that the model can adapt to the soil water retention characteristics and microclimate environment of a specific field.

[0017] Preferably, in the aforementioned green and ecological daylily cultivation method, after selecting strains with excellent overall performance in step 3.4), a seedling purification and rapid propagation step is added: Rhizome meristems of superior strains were selected as explants for in vitro tissue culture to induce the formation of clustered shoots, and tissue culture seedlings were obtained through rooting culture. After hardening offspring, tissue culture seedlings are transplanted to an isolation propagation nursery, and the resulting vegetatively propagated offspring are used as seedlings for the planting steps described above.

[0018] Preferably, in the green ecological planting method for daylilies, after covering the soil with 3-5cm in step 1), before applying the intermediate layer fertilizer, a layer of isolation and conditioning layer with a thickness of 1-2cm is first laid, which is a mixture of fully decomposed organic fertilizer and bentonite in a weight ratio of 5:1; the intermediate layer fertilizer is applied on top of the isolation and conditioning layer.

[0019] Preferably, in the aforementioned green and ecological daylily cultivation method, the organic fertilizer in step 1) is obtained through the following preparation method: S1. Raw material pretreatment and sorting: The mixture of cow and sheep manure and Asteraceae plant residue is divided into a first raw material part and a second raw material part according to a dry matter weight ratio of 60-70:30-40; the source of Asteraceae plant residue is determined based on the main Asteraceae plant species listed in the official agricultural records, floras or forestry resource survey reports of the county or city administrative division. S2. Asynchronous aerobic fermentation: S2.1 Fermentation of the first raw material: Adjust the moisture content of the first raw material to 55-60% and the carbon-nitrogen ratio (C / N) to 25-30:1, and carry out windrow aerobic composting; when the temperature at the center of the compost pile remains above 55℃ for 10-15 days, it enters the aging stage until it is fully decomposed, and the organic fertilizer used as the bottom fertilizer is obtained; the criteria for determining full decomposition are: the carbon-nitrogen ratio (C / N) is stable in the range of (15-18):1, and the seed germination index (GI) is ≥90%; S2.2 Fermentation of the second raw material: After the first raw material enters the aging stage, the second raw material begins composting fermentation, with the initial fermentation conditions being the same as in S2.1; when the pile of the second raw material experiences a high temperature stage that is continuously above 55°C, and its temperature drops from the peak and stabilizes in the 35-40°C range, active turning is stopped, keeping it in a semi-composted state to obtain the organic fertilizer used as the middle layer fertilizer; the carbon-nitrogen ratio (C / N) of this middle layer fertilizer is 25-30:1; When the first raw material portion reaches complete decomposition, the middle layer fertilizer of the second raw material portion, which is in a semi-decomposed state, is obtained and applied in layers.

[0020] The organic fertilizer that has not completed the later aging process refers to compost material that has completed the main fermentation stage but has not yet reached all the maturity indicators.

[0021] Preferably, in the green and ecological planting method for daylilies, before laying the black fully biodegradable mulch in step 2), a layer of rice husk charcoal or straw fragments with a thickness of 0.5-1.0 cm and a particle size of 2-4 mm is laid on top of the layer of wood ash evenly spread on the ridge surface, and then the mulch is covered. After laying the black fully biodegradable mulch in step 2) and before planting in step 3), add a mulch reinforcement step: evenly cover the surface of the laid mulch with a layer of fine sand and clay mixture with a thickness of 0.5-1.0cm; or, lay a mesh reinforcement layer made of biodegradable plant fibers on the surface of the mulch.

[0022] The present invention has at least the following beneficial effects: This invention utilizes organic fertilizer made from locally sourced Asteraceae plant residues and cow / sheep manure, combined with a stratified strip application method based on crop nutrient requirements. This provides a long-lasting and balanced nutrient supply for daylily growth, improves soil aggregate structure, and reduces reliance on chemical fertilizers. Simultaneously, the use of wood ash in conjunction with a black, fully biodegradable mulch film with micro-perforations effectively suppresses weeds, maintains soil moisture, enhances soil aeration, and avoids the environmental pollution problems caused by traditional plastic mulch residues, thus achieving the goal of ecological planting.

[0023] This invention introduces a multi-source data fusion and intelligent decision-making model based on the Internet of Things, meteorology, and satellite remote sensing to achieve real-time and accurate perception and prediction of soil moisture conditions in daylily fields. This model can comprehensively analyze spatiotemporal dynamic information and output scientific irrigation decisions, thereby significantly improving the accuracy and timeliness of irrigation, effectively avoiding water stress or resource waste, providing stable water security for the high-yield and high-quality growth of daylilies, and improving water resource utilization efficiency.

[0024] This invention involves artificially hybridizing highly adaptable wild daylilies from high altitudes with high-yield cultivated varieties, and then using a specially prepared seedling substrate for cultivation and selection. This effectively combines the superior traits of both parents to obtain hybrid seedlings with strong environmental adaptability, disease resistance, and high yield potential. This provides a foundation of superior-quality, genetically rich planting materials for subsequent green and ecological planting, helping to improve the stability and yield potential of the overall planting system.

[0025] This invention designs a complete method encompassing data preprocessing, spatiotemporal fusion, feature engineering, and training of a dual-stream hybrid neural network model. This method fully utilizes the spatiotemporal correlation information of historical multi-source data to effectively extract key features influencing irrigation decisions. The training method enables the model not only to learn the spatiotemporal patterns of soil moisture changes but also to correlate the long-term effects of irrigation operations with final yield and quality, thereby training a high-precision intelligent irrigation model with strong generalization ability and more comprehensive decision-making basis.

[0026] This invention enhances the robustness and reliability of intelligent irrigation systems operating in complex field environments by incorporating a data disaster recovery and degradation decision-making module. When the primary data source becomes abnormal, the system can smoothly switch to backup decision logic, maintaining basic irrigation functions based on available data and experience models. This ensures the continuity of field management operations and avoids the risk of crop drought or complete system shutdown due to data interruption.

[0027] This invention enables the irrigation decision-making model to self-adjust and improve based on actual feedback data from specific fields by performing online optimization and incremental learning of model parameters after each growing season. This allows the model to gradually adapt to the unique soil hydrological characteristics and microclimate conditions of the field, overcoming the "acclimatization" problem that may exist in pre-trained models, and achieving continuous improvement in decision-making accuracy and personalized adaptation under long-term application.

[0028] This invention, by introducing a tissue culture rapid propagation step after hybridization selection, can quickly fix and purify the genetic characteristics of superior strains, obtaining a large number of clonal seedlings with highly consistent genetic traits in a short period of time. This solves the bottleneck of slow segregation and reproduction of hybridization offspring, accelerates the promotion and application of superior new varieties, and provides a high-quality seedling guarantee for large-scale, standardized, and green planting.

[0029] This invention establishes an isolation and conditioning layer composed of well-rotted organic fertilizer and bentonite between the base fertilizer layer and the intermediate fertilizer layer. This physically prevents premature mixing of the upper and lower fertilizer layers, ensuring that nutrients are released according to the designed stages. Simultaneously, this layer buffers the direct impact of any incompletely decomposed substances from the intermediate fertilizer layer on the seedling roots, and utilizes the water- and fertilizer-retaining properties of bentonite to create a more stable and suitable microenvironment for the roots.

[0030] This invention, through the design of an asynchronous aerobic fermentation process, ingeniously achieves the simultaneous production of two organic fertilizer products with different degrees of decomposition—fully decomposed and semi-decomposed—within the same production process by initiating fermentation at different times from the same batch of raw materials. This not only meets the technical requirements of stratified fertilization for a combination of fast-acting and long-acting fertilizers, but also simplifies production management, improves the flexibility and targeting of organic fertilizer preparation, and ensures the smooth implementation of core fertilization steps.

[0031] This invention improves the air and water permeability of the ridge surface and prevents compaction by adding a porous carbonized material or fragment layer to the wood ash. Simultaneously, covering the mulch with a mixture of fine sand and clay or reinforcing it with a biodegradable fiber mesh effectively prevents the mulch from being blown away by wind or prematurely degrading, ensuring the stability of its physical structure during the critical growth period and thus reliably fulfilling its functions of moisture retention and weed suppression. Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Detailed Implementation

[0032] The present invention will now be described in further detail so that those skilled in the art can implement it based on the description.

[0033] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not imply the presence or addition of one or more other elements or combinations thereof.

[0034] According to one embodiment of the present invention, a green and ecological planting method for daylilies includes the following steps: 1) Fertilization: Apply organic fertilizer made from cow and sheep manure and Asteraceae plant residues through aerobic composting; wherein, the Asteraceae plant residues are derived from common Asteraceae plants within the county-level administrative region where the planting site is located; 2) Land preparation and mulching: After preparing the land and forming planting ridges, first spread a layer of wood ash with a thickness of 2-3cm evenly on the soil surface of the ridges; then lay a black fully biodegradable mulch film, which has micro-ventilation holes arranged in a rectangular array. The pore diameter of the micro-ventilation holes is 0.5-1.0mm and the pore density is 300-500 per square meter. 3) Transplanting: Make holes in the plastic film to transplant daylily seedlings; In step 1), the fertilization operation adopts the layered strip application method, which is as follows: In the planting trench dug in the planting ridge, first apply organic fertilizer that has passed the high temperature decomposition period and is fully decomposed as the base fertilizer layer, cover it with soil 3-5cm, then apply organic fertilizer that has passed the main fermentation stage but has not completed the later aging as the middle layer fertilizer, and then cover it with soil and plant the seedlings.

[0035] In traditional daylily cultivation, a common practice is to apply chemical compound fertilizer as base fertilizer in a single application to the planting furrow after land preparation, followed by covering with ordinary black polyethylene plastic film. While this method has a positive effect on moisture retention and soil warming for a certain period, it has significant long-term drawbacks. Continuous and excessive application of chemical fertilizers easily leads to soil compaction, disrupting soil aggregate structure, making it difficult to increase soil organic matter content, and potentially inhibiting the activity of beneficial microorganisms in the soil. At the same time, traditional polyethylene film is extremely difficult to degrade under natural conditions, and its residual fragments accumulate in the topsoil year after year, forming serious white pollution. This not only hinders water and root development but also affects the growth of subsequent crop rotations, posing a long-term threat to the health of the entire farmland ecosystem. How to maintain soil fertility and control weeds without relying on synthetic chemicals, while simultaneously avoiding farmland environmental pollution, has become a practical challenge in production.

[0036] In this embodiment, organic fertilizer made from cow and sheep manure and locally common Asteraceae plant residues, such as mugwort or dandelion stalks, through a standardized aerobic composting process is selected for fertilization. In practice, a layered strip application method is adopted. First, organic fertilizer that has undergone high-temperature composting and reached full composting standards is applied as a base fertilizer layer at the bottom of the pre-dug planting trench. This layer of fertilizer is stable, providing basic nutrients and improving the soil environment at the bottom of the trench. Then, about four centimeters of soil is covered to separate the base fertilizer layer from the top. Next, another type of organic fertilizer is applied on top of the soil. This type of fertilizer has also undergone the high-temperature process of the main fermentation stage, but has not yet undergone long-term post-fermentation aging and is in a semi-composted state, serving as a middle layer fertilizer. This semi-composted fertilizer will continue to decompose slowly in the soil, thus continuously releasing nutrients during the later stages of crop growth. Finally, the soil is covered again and the crops are planted.

[0037] In the land preparation and surface management stage, after the land is ridged, a layer of wood ash about 2.5 cm thick is evenly spread on the soil surface of the ridges. Wood ash not only provides mineral elements such as potassium, but also moderately adjusts the soil pH. Next, a specially made black, fully biodegradable mulch film is laid. This mulch film has numerous tiny air pores regularly distributed on it, with a diameter of about 0.8 mm, approximately 400 pores per square meter. These micro-air pores allow for a certain degree of gas exchange between the soil and the atmosphere while ensuring the mulch film's primary functions of suppressing weeds and conserving moisture, helping to maintain an aerobic environment for root respiration and soil microorganisms. Finally, holes are made in the laid mulch film, and daylily seedlings are planted.

[0038] The traditional chemical fertilizer plus polyethylene mulch method was used as a control group in parallel planting on the same plot. After a complete growing cycle, the physical properties of the soil in the experimental area using the new method differed. Due to the continuous effect of organic fertilizer and the addition of wood ash, the soil felt looser and its aeration improved. The fully biodegradable mulch began to break down and decompose after the growing season, and by the time of the next cultivation season, it had largely integrated into the soil, with no obvious plastic residue. In contrast, the soil condition in the traditional control group did not show significant improvement, and a large number of difficult-to-break polyethylene mulch fragments were removed after harvest.

[0039] This embodiment establishes a planting foundation that more closely resembles the natural cycle. By applying localized organic fertilizers, particularly through the innovative method of layered application of organic fertilizers at different decomposed levels, the process of nutrient release in nature is simulated. This aims to provide a balanced and sustained nutrient supply for daylilies throughout their entire growth cycle, fundamentally avoiding the damage to soil structure caused by chemical fertilizers. The use of a fully biodegradable mulch film with micro-perforations effectively suppresses weeds and retains soil moisture while addressing the core environmental problem of permanent pollution from traditional plastic mulch films. Furthermore, the perforated design mitigates the potential inhibition of soil respiration caused by fully enclosed covering. Therefore, the main benefit of this method is the simultaneous achievement of soil fertility and environmental protection, providing a feasible ecological technological path for the sustainable production of daylilies.

[0040] According to one embodiment of the present invention, a green and ecological planting method for daylilies, preferably, further includes field management, wherein the irrigation step in the field management is controlled by a pre-trained irrigation decision model, specifically including the following operational steps: a) Multi-source data acquisition and fusion: real-time soil temperature and humidity data collected by IoT sensor nodes pre-embedded in the planting ridges; simultaneously, real-time meteorological data and regional weather forecast data for the next 72 hours are acquired by accessing a meteorological data platform through an application programming interface; and multi-source satellite remote sensing data are accessed through a satellite data service interface, the data including at least thermal infrared surface temperature data and microwave surface humidity data for retrieving regional soil moisture status, and normalized vegetation index data for monitoring crop growth. b) Irrigation decision generation: The multi-source heterogeneous data obtained in step a) is spatiotemporally aligned and fused, and then input into the irrigation decision model along with the pre-set water requirement threshold parameters that match the current growth stage of daylilies. The model is trained using a hybrid deep learning algorithm that combines convolutional neural networks and long short-term memory networks, with historical irrigation data, crop growth data and corresponding yield and quality data as labels. It is used to output decision instructions that include specific irrigation time, duration and water volume. c) Command issuance and execution: The decision command generated in step b) is sent to the irrigation controller deployed in the field via a wireless Internet of Things protocol, and the controller drives the solenoid valve of the drip irrigation system to execute the quantitative irrigation.

[0041] In the field management of daylilies, irrigation has long relied on growers' personal experience or simple fixed schedules. This traditional method often fails to accurately capture real-time dynamic changes in soil moisture and cannot effectively respond to fluctuations in weather conditions, easily leading to inappropriate irrigation timing, either wasting water resources and nutrient leaching, or causing water stress in crops and affecting their growth. To solve this problem, this application proposes an irrigation control method based on an intelligent decision-making model. The implementation is as follows: IoT sensor nodes are pre-installed in the field to continuously collect soil temperature and humidity data; simultaneously, a programming interface is used to connect to an authoritative meteorological platform to obtain real-time weather information and regional forecasts for the next three days; in addition, satellite data services are connected to obtain remote sensing data reflecting regional soil moisture and crop growth, such as thermal infrared, microwave, and vegetation index data. All these multi-source, multi-type data are aligned and fused in time and space to form a comprehensive field status view. This view, along with the water requirement threshold parameters of the daylily's current growth stage, is input into a pre-trained irrigation decision-making model. This model is constructed using a hybrid deep learning algorithm combining convolutional neural networks and long short-term memory networks, and uses historical irrigation records and corresponding yield and quality data as training labels. After analyzing the input information, the model outputs specific irrigation instructions, including the optimal start time, duration, and irrigation volume. These instructions are sent to the irrigation controller in the field via a wireless IoT protocol. The controller then activates the solenoid valves of the drip irrigation system to execute precise, quantitative irrigation.

[0042] This embodiment elevates irrigation management from experience-driven to data- and model-driven. By integrating and analyzing heterogeneous information from multiple sources, it enables irrigation decisions to dynamically adapt to continuous changes in soil, weather, and the crop itself. Its beneficial effects are mainly reflected in significantly improving the accuracy and timeliness of irrigation operations, avoiding ineffective water resource dissipation, and creating a soil moisture environment more suited to the physiological needs of daylilies. This provides a stable and reliable water guarantee for healthy crop growth and the formation of yield and quality, while also improving the intelligence level and utilization efficiency of agricultural water resource management.

[0043] According to one embodiment of the present invention, a green and ecological planting method for daylilies is preferably provided, wherein the daylily seedlings are obtained through the following hybridization and breeding method: 3.1) Parent selection: Wild daylilies growing at altitudes above 800 meters were selected as the female parent, and superior cultivated varieties with strong disease resistance and a large number of inflorescences were selected as the male parent; 3.2) Artificial hybridization: During the peak flowering period of the female parent, take the healthy and tender flower buds that have opened on the same day, remove the male parts, pollinate them with the new fresh pollen of the male parent, and then isolate them by bagging. 3.3) Seed treatment and sowing: Harvest the hybrid seeds, disinfect and soak them in warm water, and then sow them in a seedling substrate made of organic fertilizer and peat in a volume ratio of 1:3 as described in claim 1; 3.4) Seedling selection and transplanting: When the seedlings have grown to 4-5 true leaves, select strong and healthy individual plants with well-developed root systems and transplant them to the nursery. The following year, based on the plant's resistance, number of inflorescences and bud morphology, select the best-performing strains as seedlings for the planting steps described above.

[0044] In daylily cultivation, the long-term use of single cultivars or seedlings propagated by simple division may lead to varietal degeneration, decreased stress resistance, and difficulty in adapting to diverse ecological environments or resisting disease attacks, thus limiting further improvements in yield and quality. Existing seedlings often struggle to balance adaptability and high yield. To address this problem, this application proposes a hybridization breeding method to obtain superior seedlings. The implementation method is as follows: First, wild daylily plants growing in mountainous areas above 800 meters in altitude are selected as the female parent, as they typically possess stronger environmental adaptability and stress-resistant genes; simultaneously, superior cultivars with strong disease resistance and abundant inflorescences, bred artificially, are selected as the male parent. During the peak flowering period of the female parent plants, healthy young flower buds that have opened that day are selected, artificially emasculated, and immediately pollinated with fresh pollen from the male parent, then bagged to prevent contamination by other pollen. After successful hybridization, the seeds are harvested, disinfected, and soaked in warm water before being sown in a specific seedling substrate, which is a mixture of the aforementioned organic fertilizer and peat moss in a specific ratio. When the seedlings have grown four to five true leaves, select the robust individuals with well-developed root systems and transplant them to a nursery for further cultivation. In the second year, based on the actual performance of these plants in the field, including their resistance to diseases, the number of inflorescences produced, and the morphological characteristics of the flower buds, comprehensively evaluate and select the best-performing strains, which can then be used as seedlings for subsequent field planting.

[0045] This embodiment creates artificial germplasm with superior characteristics through targeted gene combination and systematic screening. It breaks down the utilization barriers between wild resources and cultivated varieties, organically combining target traits such as adaptability and high yield. Its beneficial effects are that it can obtain new daylily seedlings with richer genetic basis, stronger stress resistance, and greater high-yield potential, providing a superior starting point for subsequent ecological planting materials and helping to enhance the production stability and output potential of the entire planting system.

[0046] According to one embodiment of the present invention, a green ecological planting method for daylilies, preferably, includes the following steps in the pre-training method of the irrigation decision model: Historical datasets are collected and integrated, including: historical multi-source heterogeneous data, which contains historical soil temperature and humidity time-series data, historical meteorological data, and historical satellite remote sensing data, wherein the satellite remote sensing data includes at least historical thermal infrared band surface temperature data, microwave band soil surface humidity data, and normalized difference vegetation index data; historical irrigation operation records aligned with the historical multi-source heterogeneous data in time and space; and historical daylily yield and quality index data corresponding to the historical irrigation operation records. The constructed historical dataset is cleaned, missing values ​​are imputed, and outliers are handled. Then, spatiotemporal alignment and fusion are performed to unify data from different sources and formats into the same spatiotemporal grid. Finally, features are extracted and constructed to generate model input feature vectors containing spatiotemporal features and crop growth stage identifiers. A two-stream hybrid neural network model is constructed, where the first stream is a convolutional neural network (CNN) branch, used to extract spatial distribution features and texture features from the fused spatialized data; the second stream is a long short-term memory network (LSTM) branch, used to extract time-series dependent features from the historical soil temperature and humidity time-series data and historical meteorological time-series data; the high-level features output by the CNN branch and the LSTM branch are concatenated and fused, and then mapped to the irrigation decision action space through a fully connected layer; The preprocessed training dataset is divided into a training set, a validation set, and a test set. The historical irrigation operation records are used as action labels, and the historical yield and quality index data are used as auxiliary supervision signals. A composite loss function, which includes mean squared error loss and a multi-objective reward function based on yield-water use efficiency, is used to perform end-to-end supervised training and optimization of the hybrid deep learning model. Early shutdown and hyperparameter tuning are performed through the validation set to finally obtain the pre-trained irrigation decision model.

[0047] The main challenge in developing efficient intelligent irrigation decision-making models is how to effectively transform historical data from diverse sources and in different formats into learnable information for the model, and train a model capable of understanding complex spatiotemporal relationships and making accurate predictions. Simple models often struggle to capture the nonlinear dynamics of multi-factor interactions in the agricultural environment. To address this challenge, this application details a pre-training method for irrigation decision-making models. The implementation is as follows: First, a systematic collection and integration of historical datasets is required, including accumulated soil temperature and humidity data arranged in time series, meteorological data, and satellite remote sensing data. The remote sensing data includes at least surface temperature, surface soil moisture, and vegetation indices. Simultaneously, historical irrigation operation records perfectly aligned with these data in time and space are collected, along with corresponding data on the final yield and quality indicators of daylilies. Next, these raw data are cleaned, missing and outlier values ​​are removed, and data from different sources are aligned and fused onto the same spatiotemporal grid. Based on this, feature engineering is performed to construct a model input feature vector containing spatiotemporal features and crop growth stage identifiers. The model architecture employs a two-stream hybrid neural network. One branch is a convolutional neural network, responsible for extracting spatial distribution and texture features from the fused spatialized data; the other branch is a long short-term memory network, responsible for extracting time-dependent features from soil and meteorological time-series data. The high-level features extracted by the two branches are concatenated and then mapped to the final irrigation decision action space through a fully connected layer. During the training phase, the preprocessed dataset is divided into training, validation, and test sets. Historical irrigation records are used as the primary action labels, while yield and quality data are used as auxiliary supervision signals. A composite loss function combining mean squared error loss and a multi-objective reward function is employed for end-to-end supervised training and optimization of the model. The training process is monitored through the validation set to prevent overfitting and adjust hyperparameters, ultimately resulting in a reliable pre-trained model.

[0048] This embodiment provides a complete and efficient technical approach for building an intelligent decision-making model capable of deeply understanding the spatiotemporal dynamics of the farmland environment. It tightly integrates data engineering with an advanced deep learning architecture. Its beneficial effects lie in its ability to train an irrigation model with powerful feature extraction capabilities and high-precision decision-making capabilities. This model can not only effectively learn from complex historical data but also better adapt to real-world agricultural environmental changes after deployment, providing a core intelligent driving engine for achieving precision irrigation and enhancing the technological advancement and reliability of the entire irrigation management system.

[0049] According to one embodiment of the present invention, a green ecological planting method for daylilies, preferably, step b) irrigation decision generation further includes a data disaster recovery and degradation decision module: When real-time soil temperature and humidity data or key satellite remote sensing data are continuously missing for more than a preset threshold, the irrigation decision model automatically switches to a degraded decision mode. In the downgraded decision-making mode, the model generates irrigation decision instructions based on the last effective soil data, historical meteorological data for the same period, and the 24-hour weather forecast obtained through the application programming interface, combined with the water requirement experience model of the current growth stage of daylilies.

[0050] In the actual operation of IoT-based and sensor-based smart irrigation systems, situations such as sensor failures, network transmission interruptions, or temporary unavailability of data services are difficult to completely avoid. Once critical real-time soil data or satellite remote sensing data streams are interrupted, the intelligent decision-making model that relies on these data inputs will not function properly, potentially causing the entire irrigation system to shut down and crops to face the risk of water shortage. This is a reliability problem that must be solved for smart agricultural systems to become practical. To address this, this application introduces a data disaster recovery and degradation decision-making module. Its implementation is as follows: The system continuously monitors data streams from field sensors and satellite data interfaces. When it detects that, for example, soil temperature and humidity data or critical satellite soil moisture data are continuously missing for more than a preset threshold, such as three data acquisition cycles, the irrigation decision-making model will not report an error or stop working, but will automatically trigger a switching mechanism to enter a degradation decision-making mode. In this mode, the model no longer relies on the missing real-time data stream, but instead relies on the last batch of effectively collected soil data, historical meteorological data stored in the system for the same period, and combined with the 24-hour weather forecast information still available through the application programming interface. This available information will be combined with an empirical water requirement model for daylilies at their current growth stage to generate a simplified yet still data-driven set of irrigation decision instructions, thereby maintaining the basic operation of irrigation.

[0051] This embodiment endows the intelligent irrigation system with resilience in the face of data anomalies. Through pre-set backup decision logic, it ensures the continuity of system function when the main data path fails. Its beneficial effects include significantly enhancing the practicality and reliability of the intelligent irrigation system in complex field environments, reducing production risks caused by technical failures, ensuring the stable operation of crop water management activities, and making the high-tech intelligent system more adaptable to the stringent requirements of actual agricultural production.

[0052] According to one embodiment of the present invention, a green and ecological planting method for daylilies, preferably, further includes an online optimization step for irrigation decision model parameters: At the end of each growing season, the system collects complete actual irrigation execution data, corresponding multi-source monitoring data, and final yield and quality data for that season to form a new set of training samples. These new samples are used to incrementally learn or fine-tune the pre-trained irrigation decision model, with a focus on optimizing the water demand threshold parameter, so that the model can adapt to the soil water retention characteristics and microclimate environment of a specific field.

[0053] A pre-trained irrigation decision-making model, once deployed to a specific field, may not perfectly match the unique soil water retention capacity, local microclimate, and other specific conditions of that field because its training data typically comes from a broader or historical environment. With increasing planting years, interannual climate variation may also cause the model's preset parameters to gradually deviate from their optimal values, leading to a slow decline in decision-making accuracy. This is an adaptive challenge faced by the model in long-term applications. To address this issue, this application establishes an online optimization step for model parameters. The implementation is as follows: After each daylily growing season, the system automatically collects and archives complete operational records for that season, including the specific execution details of each irrigation, the corresponding multi-source monitoring data sequences, and the yield and quality data measured at harvest. These data collectively constitute a new set of training samples with feedback. Subsequently, the system uses these new samples from the field to incrementally learn or fine-tune the deployed pre-trained irrigation decision-making model. The optimization focuses particularly on the threshold parameters within the model related to the water requirements of daylilies at different growth stages. Through this continuous, feedback-based learning, the model can gradually understand and adapt to the soil hydrological characteristics and unique microclimate patterns of specific fields, thereby achieving localized improvement and long-term maintenance of decision-making performance.

[0054] This embodiment endows the intelligent irrigation system with the ability to self-evolve. It treats each production cycle as a learning opportunity, using actual results to calibrate and improve the model. Its beneficial effect lies in effectively overcoming the "incompatibility" problem that may arise in pre-trained models in specific application scenarios, ensuring that the intelligent decision-making system can maintain high accuracy and adaptability over the long term. This allows irrigation management to become increasingly precise in matching the actual conditions of specific fields, achieving a sustainable improvement in the level of intelligence.

[0055] According to one embodiment of the present invention, a green and ecological planting method for daylilies, preferably, after screening out strains with excellent overall performance in step 3.4), adds a seedling purification and rapid propagation step: Rhizome meristems of superior strains were selected as explants for in vitro tissue culture to induce the formation of clustered shoots, and tissue culture seedlings were obtained through rooting culture. After hardening offspring, tissue culture seedlings are transplanted to an isolation propagation nursery, and the resulting vegetatively propagated offspring are used as seedlings for the planting steps described above.

[0056] Superior yellow-flowered plantlets obtained through artificial hybridization, while exhibiting outstanding overall traits, show genetic segregation in their offspring through seed propagation, making it difficult to maintain phenotypic consistency. Traditional division propagation methods are not only slow and have low propagation coefficients, making it difficult to obtain large quantities of seedlings in a short period, but also may carry and spread soil-borne diseases. This is a bottleneck problem in rapidly converting superior germplasm resources into large-scale production materials. Therefore, this application adds a seedling purification and rapid propagation step after hybridization. The implementation method is as follows: Healthy, sterile rhizome meristems are selected as explants from the selected hybrid lines with excellent overall performance. Under sterile conditions, the explants are inoculated into a specific in vitro tissue culture medium. By adjusting the ratio of plant growth regulators, dedifferentiation is induced to produce a large number of clustered shoots. Subsequently, these clustered shoots are transferred to a rooting medium to induce the formation of complete root systems, thereby obtaining a large number of tissue-cultured seedlings. After a period of hardening off and adaptation to the natural environment, these tissue-cultured seedlings can be transplanted to a well-isolated dedicated propagation nursery for further cultivation. The offspring population obtained through vegetative propagation thus has a genetic background that is completely consistent with the parent plant and can be directly used as seedlings for field planting.

[0057] This embodiment bridges the gap between hybridization breeding achievements and industrial applications. It transforms individuals with superior genotypes into a large population of asexual lines with pure genetic backgrounds and highly uniform traits. Its beneficial effects lie in its ability to rapidly solidify and amplify the results of hybridization breeding, providing a large number of high-quality, uniform, and healthy seedlings in a very short time. This completely solves the problems of slow propagation and difficulty in promoting superior new germplasm, greatly accelerating the process of breeding iteration and variety renewal, and laying a solid seedling foundation for large-scale, standardized, green, and ecological planting.

[0058] According to one embodiment of the present invention, a green ecological planting method for daylilies is preferably provided. After covering the soil with 3-5cm in step 1), before applying the intermediate layer fertilizer, a layer of isolation and conditioning layer with a thickness of 1-2cm is first laid, which is a mixture of fully decomposed organic fertilizer and bentonite in a weight ratio of 5:1. The intermediate layer fertilizer is applied on the isolation and conditioning layer.

[0059] When applying fertilizer using the layered strip application method, if semi-decomposed intermediate fertilizer that has not yet completed its later aging is applied directly onto the topsoil layer, the incompletely decomposed organic matter in this layer may prematurely mix with the fully decomposed base fertilizer layer below during subsequent irrigation or rainfall. This mixing may disrupt the originally designed nutrient release rhythm of "long-term supply from the bottom layer and slow-release relay from the middle layer." Simultaneously, if the semi-decomposed fertilizer comes into direct contact with the newly formed crop roots, some intermediate decomposition products or the high carbon-to-nitrogen ratio may slightly inhibit or stress the growth of young roots. This is a challenge of interface stability and root protection that needs to be addressed in the practical operation of refined layered fertilization. To solve this problem, this application adds a step of laying an isolation and conditioning layer after the base fertilizer layer is covered with soil and before the application of the intermediate fertilizer. Specifically, after the base fertilizer is applied and covered with approximately four centimeters of soil, fully decomposed organic fertilizer and mineral bentonite are thoroughly mixed at a weight ratio of five to one. Spread this mixture evenly over the topsoil layer to form a continuous thin layer about 1.5 cm thick. Then, apply semi-decomposed intermediate fertilizer over this isolation and conditioning layer, and then continue with the topsoil covering and planting.

[0060] This embodiment adds a functional buffer and control interface to the stratified fertilization system. It optimizes the rhizosphere microenvironment through physical isolation and material properties. Its beneficial effects include more precise control over the release trajectory of nutrients in different fertilizer layers, reduced interlayer interference, and effective implementation of a "fast-slow combination" fertilization design. Simultaneously, it provides a gentler and more stable transition zone for the early downward growth of crop seedling roots, helping to cultivate robust root systems and thus improving fertilizer utilization efficiency and plant growth stability.

[0061] According to one embodiment of the present invention, a green and ecological planting method for daylilies, preferably, the organic fertilizer in step 1) is obtained by the following preparation method: S1. Raw material pretreatment and sorting: The mixture of cow and sheep manure and Asteraceae plant residue is divided into a first raw material part and a second raw material part according to a dry matter weight ratio of 60-70:30-40; the source of Asteraceae plant residue is determined based on the main Asteraceae plant species listed in the official agricultural records, floras or forestry resource survey reports of the county or city administrative division. S2. Asynchronous aerobic fermentation: S2.1 Fermentation of the first raw material: Adjust the moisture content of the first raw material to 55-60% and the carbon-nitrogen ratio (C / N) to 25-30:1, and carry out windrow aerobic composting; when the temperature at the center of the compost pile remains above 55℃ for 10-15 days, it enters the aging stage until it is fully decomposed, and the organic fertilizer used as the bottom fertilizer is obtained; the criteria for determining full decomposition are: the carbon-nitrogen ratio (C / N) is stable in the range of (15-18):1, and the seed germination index (GI) is ≥90%; S2.2 Fermentation of the second raw material: After the first raw material enters the aging stage, the second raw material begins composting fermentation, with the initial fermentation conditions being the same as in S2.1; when the pile of the second raw material experiences a high temperature stage that is continuously above 55°C, and its temperature drops from the peak and stabilizes in the 35-40°C range, active turning is stopped, keeping it in a semi-composted state to obtain the organic fertilizer used as the middle layer fertilizer; the carbon-nitrogen ratio (C / N) of this middle layer fertilizer is 25-30:1; When the first raw material portion reaches complete decomposition, the middle layer fertilizer of the second raw material portion, which is in a semi-decomposed state, is obtained and applied in layers.

[0062] The organic fertilizer that has not completed the later aging process refers to compost material that has completed the main fermentation stage but has not yet reached all the maturity indicators.

[0063] To implement the strip application method, it is necessary to simultaneously obtain two types of organic fertilizers with different degrees of decomposition: fully decomposed fertilizer and semi-decomposed fertilizer. Using two separate sets of raw materials and fermentation processes would significantly increase management complexity and production costs. Furthermore, attempting to obtain products with different degrees of decomposition from the same batch of fermentation material by removing a portion midway through the fermentation process is difficult to control precisely regarding the timing of removal and the state of the material, leading to unstable product quality and an inability to ensure that the two fertilizers can be obtained simultaneously and in a matching manner when needed. This is an operational challenge faced in the economical and stable preparation of two fertilizer products that meet specific differentiated needs within a single production process. Therefore, this application designs an asynchronous aerobic fermentation preparation method. The specific implementation method is as follows: First, uniformly mixed cow and sheep manure and Asteraceae plant residue are pre-divided into a first raw material part and a second raw material part at a dry matter weight ratio of approximately 65:35. The first batch of raw materials undergoes standard aerobic composting, adjusting its moisture content to approximately 58% and its carbon-to-nitrogen ratio to approximately 28:1. After the compost has experienced a high-temperature period of more than 12 days, it enters the aging stage until fully decomposed, yielding a base fertilizer with a stable carbon-to-nitrogen ratio and a high seed germination index. Crucially, when the first batch of raw materials enters the aging stage where infrequent turning is unnecessary, the composting fermentation of the second batch of raw materials is immediately initiated, with the same initial adjustment parameters as the first batch. Once the second batch of raw materials has also experienced a high-temperature stage, and its temperature stabilizes at approximately 38 degrees Celsius after its peak, active turning is stopped, significantly reducing its biochemical reaction rate and maintaining it in a semi-decomposed state, at which point its carbon-to-nitrogen ratio remains at approximately 28:1. Through this asynchronous initiation and process control, the second batch of raw materials reaches the ideal semi-decomposed state simultaneously with the first batch reaching full decomposition, allowing them to be removed at the same time for stratified fertilization.

[0064] This embodiment provides an efficient and collaborative production process to support the core requirement of stratified fertilization. It integrates the preparation processes of the two fertilizers into a unified whole. Its advantages lie in greatly simplifying production management, reducing the additional costs and operational complexity associated with obtaining the two specialized fertilizers, ensuring a simultaneous supply of fully decomposed and semi-decomposed fertilizers in both quality and quantity, fundamentally guaranteeing the smooth and reliable implementation of the stratified strip application method described in claim 1, and improving the practicality and operability of the entire technical solution.

[0065] According to one embodiment of the present invention, a green ecological planting method for daylilies is preferably provided, in step 2), before laying the black fully biodegradable mulch, a layer of rice husk charcoal or straw fragments with a thickness of 0.5-1.0 cm and a particle size of 2-4 mm is laid on top of the layer of wood ash evenly spread on the ridge surface, and then the mulch is covered. After laying the black fully biodegradable mulch in step 2) and before planting in step 3), add a mulch reinforcement step: evenly cover the surface of the laid mulch with a layer of fine sand and clay mixture with a thickness of 0.5-1.0cm; or, lay a mesh reinforcement layer made of biodegradable plant fibers on the surface of the mulch.

[0066] When laying fully biodegradable mulch in the field, especially in windy areas or on slopes, the thin mulch is easily blown away, shifted, or even torn by the wind, affecting its moisture retention and weed suppression effects. Simultaneously, directly applied wood ash may compact after irrigation, affecting soil aeration, and its fine particles may partially block the micropores on the mulch. These practical problems can weaken the design effectiveness of the core measures and represent engineering stability challenges that need to be overcome to move ecological planting technology from a plan to stable application. Therefore, this application adds two auxiliary steps before and after laying the mulch. Specifically, after applying the wood ash, instead of immediately covering it with mulch, a layer of rice husk charcoal or straw fragments, approximately 0.8 cm thick and with a particle size of about 3 mm, is evenly laid on top of the wood ash layer. This porous carbonaceous material layer acts as a loosening medium. Subsequently, the perforated fully biodegradable mulch is laid on top of this layer. After the mulch film is laid flat, a reinforcement step is performed: a layer of dry soil, a mixture of fine sand and clay, about 0.8 cm thick, is evenly covered on the surface of the mulch film; or, a layer of biodegradable mesh material woven from natural plant fibers such as hemp or palm is laid. The covering dry soil or mesh material can effectively compact and anchor the mulch film to the ridge surface.

[0067] This embodiment utilizes auxiliary engineering measures to ensure that the core materials (wood ash and biodegradable mulch film) can continuously and stably perform their intended functions in the complex environment of the field. It enhances the robustness and environmental adaptability of the technical solution. Its beneficial effects include effectively preventing premature failure of the mulch film due to external forces, ensuring its physical integrity throughout the critical growth period; simultaneously, by optimizing the structure of the wood ash layer, it maintains good soil and mulch film aeration, thereby improving the overall stability and reliability of ecological planting field operations, enabling green technology to be truly implemented and withstand the test of actual production conditions.

[0068] To enable those skilled in the art to better understand the technical solution of the present invention, the following embodiments are provided for further explanation: Example 1 In a plain area of ​​East China, the specific steps of a green and ecological method for cultivating daylilies are as follows: 1) Fertilization: Apply organic fertilizer made from local cow and sheep manure and residues of common Asteraceae plants (such as chrysanthemum and mugwort) within the county or city administrative region where the planting area is located, through aerobic composting. The fertilization operation adopts the layered strip application method: In the planting trench dug in the planting ridge, first apply the organic fertilizer that has passed the high-temperature composting period and is fully decomposed as the base fertilizer layer, cover it with 4cm of soil, and then apply the organic fertilizer that has passed the main fermentation stage but has not completed the later aging as the middle layer fertilizer, and then cover it with soil and plant the seedlings.

[0069] 2) Land preparation and laying: After the land is prepared and ridges are formed to form planting ridges, a layer of wood ash with a thickness of 2.5cm is evenly spread on the soil surface of the ridges; then a black fully biodegradable mulch film is laid. The mulch film has micro-ventilation holes arranged in a rectangular array. The pore diameter of the micro-ventilation holes is 0.8mm and the pore density is 400 per square meter.

[0070] 3) Transplanting: Make holes in the plastic film to transplant daylily seedlings.

[0071] Example 2 At a planting base in a mountainous area of ​​Southwest China, the specific steps of a green and ecological planting method for daylilies are as follows: 1) Fertilization: Apply organic fertilizer made from cow and sheep manure and residues of common Asteraceae plants within the county or city administrative region where the planting site is located, through aerobic composting. The fertilization operation adopts a layered strip application method: In the planting trench dug in the planting ridge, first apply organic fertilizer that has passed the high-temperature composting period and is fully decomposed as the base fertilizer layer, cover it with 4cm of soil, then apply organic fertilizer that has passed the main fermentation stage but has not completed the later aging as the middle layer fertilizer, and then cover it with soil and plant the seedlings.

[0072] 2) Land preparation and laying: After the land is prepared and ridges are formed to form planting ridges, a layer of wood ash with a thickness of 2.5cm is evenly spread on the soil surface of the ridges; then a black fully biodegradable mulch film is laid. The mulch film has micro-ventilation holes arranged in a rectangular array. The pore diameter of the micro-ventilation holes is 0.8mm and the pore density is 400 per square meter.

[0073] 3) Transplanting: Make holes in the plastic film to transplant daylily seedlings.

[0074] 4) Deployment and operation of intelligent irrigation systems: a) Multi-source data acquisition and fusion: Real-time soil temperature and humidity data are acquired by IoT sensor nodes pre-embedded in the planting ridges at preset cycles; simultaneously, real-time meteorological data and regional weather forecast data for the next 72 hours are acquired by accessing a meteorological data platform through an application programming interface (API); and multi-source satellite remote sensing data are accessed through a satellite data service interface, including thermal infrared surface temperature data and microwave surface humidity data for inverting regional soil moisture status, as well as normalized vegetation index data for monitoring crop growth.

[0075] b) Irrigation Decision Generation: The multi-source heterogeneous data obtained in step a) is spatiotemporally aligned and fused, and then input into a pre-trained irrigation decision model along with pre-set water requirement threshold parameters matching the current growth stage of daylilies. The model training process specifically includes: collecting and integrating historical soil temperature and humidity time-series data, historical meteorological data, historical satellite remote sensing data (including thermal infrared, microwave, and vegetation index data), spatiotemporally aligned historical irrigation records, and corresponding yield and quality data into a historical dataset; performing data cleaning, interpolation, alignment and fusion, and feature construction; constructing a dual-stream hybrid neural network model consisting of a CNN branch (extracting spatial features) and an LSTM branch (extracting temporal features), using historical irrigation records as action labels and yield and quality data as auxiliary supervision, and employing a composite loss function for end-to-end training and optimization to finally obtain the model. This model is used to output decision instructions containing specific irrigation time, duration, and water volume.

[0076] c) Data disaster recovery and degradation decision-making: When real-time soil temperature and humidity data or key satellite remote sensing data are continuously missing for more than 3 acquisition cycles, the irrigation decision model automatically switches to degradation decision-making mode. In this mode, the model generates irrigation decision instructions based on the last valid soil data, historical meteorological data for the same period, and the 24-hour weather forecast obtained through API, combined with the water requirement experience model of the current growth stage of daylilies.

[0077] d) Issuance and execution of instructions: The decision instructions generated in step b) or c) are sent to the irrigation controller deployed in the field via the LoRa wireless IoT protocol, and the controller drives the solenoid valve of the drip irrigation system to perform the quantitative irrigation.

[0078] e) Online optimization of model parameters: After each growing season, the system collects complete actual irrigation execution data, corresponding multi-source monitoring data, and final yield and quality data for that season to form a new set of training samples. These new samples are used to incrementally learn the pre-trained irrigation decision model, focusing on optimizing the water demand threshold parameters so that the model can adapt to the soil water retention characteristics and microclimate environment of a specific field.

[0079] Example 3 The specific steps of a green and ecological cultivation method for daylilies at an ecological farm in South China are as follows: 1) Selection and rapid propagation of superior seedlings: a) Parent selection and hybridization: Wild daylilies growing at altitudes above 850 meters were selected as the female parent, and the superior cultivated variety "April Flower" with strong disease resistance and a large number of inflorescences was selected as the male parent. During the peak flowering period of the female parent, healthy and tender flower buds that opened that day were taken, emasculated, and pollinated with fresh fresh pollen from the male parent, and then bagged for isolation.

[0080] b) Seed treatment and sowing: Harvest the hybrid seeds, disinfect and soak them in warm water, and then sow them in a seedling substrate made by the method described in claim 1, which is a mixture of organic fertilizer and peat in a volume ratio of 1:3.

[0081] c) Seedling selection: When the seedlings have grown to 4 true leaves, select robust individual plants with well-developed root systems and transplant them to the nursery. The following year, based on the plant's disease resistance, number of inflorescences, and bud morphology, select strain No. 1 with excellent overall performance.

[0082] d) Seedling purification and rapid propagation: Rhizome meristems of superior strain No. 1 were selected as explants for in vitro tissue culture to induce the production of clustered shoots, and tissue culture seedlings were obtained through rooting culture. After hardening offspring, the tissue culture seedlings were transplanted to an isolated propagation nursery, and the vegetatively propagated offspring obtained therefrom were used as seedlings for subsequent planting steps.

[0083] 2) Ecological farming: a) Fertilization: Apply organic fertilizer made from cow and sheep manure and residues of common Asteraceae plants (such as wild chrysanthemum and *Chrysanthemum indicum*) within the county or city-level administrative region where the planting site is located, through aerobic composting. The fertilization process uses a layered strip application method: In the planting furrows dug on the planting ridges, first apply fully decomposed organic fertilizer that has undergone high-temperature composting as the base fertilizer layer, cover with 4cm of soil, then apply organic fertilizer that has passed the main fermentation stage but has not yet completed the later aging process as the middle layer fertilizer, and finally cover with soil.

[0084] b) Land preparation and mulching: After the land is prepared and ridges are formed to form planting ridges, a layer of wood ash with a thickness of 2.5 cm is evenly spread on the soil surface of the ridges; then a black fully biodegradable mulch film is laid, which has micro-ventilation holes arranged in a rectangular array. The pore diameter of the micro-ventilation holes is 0.8 mm and the pore density is 400 per square meter.

[0085] c) Transplanting: Using the tissue culture seedling No. 1 obtained in step 1) d), make holes in the plastic film for transplanting.

[0086] 3) Intelligent irrigation management: A smart irrigation system is deployed, collecting information from multiple sources through IoT sensors, meteorological APIs, and satellite remote sensing data interfaces. The spatiotemporally aligned and fused data, along with water requirement threshold parameters for daylily growth stages, is input into a pre-trained irrigation decision model (trained using a hybrid CNN and LSTM algorithm). The model generates decision instructions containing specific irrigation time, duration, and water volume, and drives the drip irrigation system to perform quantitative irrigation via a wireless IoT protocol.

[0087] Example 4 In a high-standard ecological agriculture demonstration park in Northwest China, the specific steps of a green and ecological daylily cultivation method are as follows: 1) Organic fertilizer preparation: S1. Raw material pretreatment and sorting: The mixture of cow and sheep manure and residues of common Asteraceae plants (such as Aster tataricus and Inula japonica) identified according to local agricultural records is divided into the first raw material part and the second raw material part according to a dry matter weight ratio of 65:35.

[0088] S2. Asynchronous aerobic fermentation: S2.1 Fermentation of the first raw material: Adjust the moisture content of the first raw material to 58% and the carbon-to-nitrogen ratio (C / N) to 28:1, and carry out windrow aerobic composting. After the temperature at the center of the compost pile remains above 55°C for 12 days, it enters the aging stage until it is fully decomposed (the carbon-to-nitrogen ratio is stable at 16:1, and the seed germination index is ≥90%), thus obtaining the organic fertilizer used as the bottom fertilizer.

[0089] S2.2 Fermentation of the second raw material: After the first raw material enters the aging stage, the second raw material begins composting fermentation, with the initial fermentation conditions being the same as in S2.1. After the pile has experienced a sustained high-temperature stage, once its temperature stabilizes at 38°C, active turning is stopped, allowing it to remain in a semi-composted state (carbon-nitrogen ratio approximately 28:1), thus obtaining the organic fertilizer used as the middle layer fertilizer.

[0090] 2) Selection and rapid propagation of superior seedlings: Wild daylilies growing at altitudes above 850 meters were selected as the female parent, and superior cultivated varieties with strong disease resistance and numerous inflorescences were selected as the male parent for artificial hybridization. After harvesting and treating the seeds, they were sown in a seedling substrate prepared in step 1) by mixing organic fertilizer and peat moss in a 1:3 volume ratio. Vigorous plants with well-developed root systems were selected, and the following year, superior strains were selected based on resistance, number of inflorescences, and flower bud morphology. Subsequently, rhizome meristems of these superior strains were used for in vitro tissue culture to induce shoot clusters, which were then rooted to obtain tissue culture seedlings. These seedlings were hardened off and transplanted to an isolated propagation nursery, and the resulting vegetatively propagated offspring were used as seedlings for planting.

[0091] 3) Ecological farming: a) Fertilization: Use the layered strip application method. In the planting furrow dug in the planting ridge, first apply the fully decomposed organic fertilizer prepared in step 1) as the base fertilizer layer, and cover it with 4cm of soil.

[0092] b) Laying the isolation and conditioning layer: On the topsoil layer, first lay a 1.5cm thick isolation and conditioning layer made of fully decomposed organic fertilizer and bentonite mixed in a 5:1 weight ratio.

[0093] c) Apply intermediate fertilizer: Apply the semi-decomposed intermediate fertilizer prepared in step 1) onto the isolation conditioning layer, and then cover it with soil.

[0094] d) Land preparation and mulching: After ridging, first evenly spread a 2.5cm thick layer of wood ash on the ridge surface; then, lay a 0.8cm thick layer of rice husk charcoal with a particle size of 3mm on top of the wood ash layer. Then, lay a black, fully biodegradable mulch film with micro-ventilation holes arranged in a rectangular array, with a hole diameter of 0.8mm and a hole density of 400 holes per square meter.

[0095] e) Mulch film reinforcement: On the surface of the laid mulch film, evenly cover it with a layer of fine sand and clay mixture with a thickness of 0.8cm.

[0096] f) Transplanting: Using the tissue culture seedlings offspring obtained in step 2), make holes in the mulch film for transplanting.

[0097] 4) Deployment and optimized operation of intelligent irrigation systems: a) System deployment and data fusion: IoT sensor nodes are pre-installed and connected to the meteorological data platform and satellite data service interface via API to obtain real-time soil data, meteorological data, 72-hour forecasts, and satellite remote sensing data including thermal infrared surface temperature, microwave soil moisture and normalized vegetation index.

[0098] b) Irrigation Decision Generation and Execution: The aforementioned multi-source heterogeneous data, after spatiotemporal alignment and fusion, is input along with the water requirement threshold parameters for the current growth stage of daylilies into a pre-trained irrigation decision model. This model is trained using a historical dataset, which, after cleaning, alignment, and feature construction, is trained and optimized using a hybrid neural network consisting of a CNN branch (extracting spatial features) and an LSTM branch (extracting temporal features). Historical irrigation records serve as action labels, and yield and quality data serve as auxiliary supervision. The model output includes decision instructions with specific time, duration, and water volume, which drive the drip irrigation system via a wireless protocol.

[0099] c) Data disaster recovery: When critical data is missing for more than 3 consecutive periods, the model automatically switches to degrade mode and generates a decision based on the last valid soil data, historical meteorological data for the same period and the weather forecast for the next 24 hours, combined with the water demand experience model.

[0100] d) Online optimization: After each growing season, new samples are formed using the complete execution, monitoring and yield quality data of that season. The model is then incrementally learned and fine-tuned, with a focus on optimizing the water requirement threshold parameter to achieve model adaptation to specific fields.

[0101] Scale settings Comparative Example 1 The specific steps of a green and ecological daylily cultivation method are as follows: The conventional local daylily cultivation model is adopted. After land preparation, compound fertilizer (N-P2O5-K2O ratio of 15-15-15) is applied once as base fertilizer in the planting furrow, at a rate of 50 kg per mu (approximately 0.067 hectares). After ridging, ordinary polyethylene (PE) black mulch film is directly covered, with no air holes. Conventional tillers from the same source are planted. During the growing season, topdressing (mainly using urea) and chemical pest and disease control are carried out based on experience. Irrigation relies on manual judgment and is conducted using furrow irrigation. This model represents traditional practice without employing any of the ecological cultivation techniques described in this application.

[0102] Comparative Example 2 The specific steps of a green and ecological daylily cultivation method are as follows: The method is carried out in fields with drip irrigation infrastructure, and the soil conditions are similar to those in Example 2. Fertilization, mulching, and seedlings are all the same as in Comparative Example 1, i.e., chemical fertilizers and PE mulch are used. The difference lies in the irrigation method: a drip irrigation plan based on a fixed schedule is adopted, i.e., irrigation is initiated every 5 days, with each irrigation lasting a fixed 2 hours (approximately 25 cubic meters of water per mu), regardless of weather changes or actual soil moisture. This model represents a conventional upgrade practice that only automates the irrigation facilities but lacks intelligent decision-making.

[0103] The implementation effect of Example 1 contrasts directly and sharply with that of Comparative Example 1. In terms of yield and quality, Example 5 provided stable and long-lasting nutrients through localized organic fertilizer and stratified strip application. Combined with the rhizosphere environment created by wood ash and breathable biodegradable mulch, this resulted in a yield increase of over 15% for daylilies, and a significant increase in the content of beneficial substances in the flower buds. Regarding environmental and soil impacts, the soil organic matter content in Comparative Example 1 did not improve due to continuous application of chemical fertilizers, and PE mulch residue accumulated year by year. In contrast, Example 1 completely eliminated the conventional input of chemical fertilizers and pesticides, resulting in a 0.6 percentage point increase in soil organic matter content over three years. The mulch was essentially degraded by the end of the season, and the soil ecology was effectively restored. This demonstrates that even without introducing intelligent systems, simply adopting the core ecological planting method of this application can achieve a significant reversal of ecological benefits while improving economic efficiency.

[0104] Regarding water resource utilization efficiency, the intelligent model in Example 2 can sensitively respond to real-time changes in soil and weather. Its irrigation decisions are more precise and flexible than the fixed-cycle irrigation of Comparative Example 2, reducing irrigation water consumption by more than 23% while ensuring crop needs are met, thus significantly improving water use efficiency. In terms of system stability and adaptability, the data disaster recovery mechanism designed in Example 2 ensures basic operation under extreme conditions, while the fixed-cycle model of Comparative Example 2 is prone to insufficient or excessive irrigation in the event of abnormal weather. Compared to Comparative Example 1, Example 2, while achieving similar soil ecological benefits as Example 1, further optimizes the crop growth environment through precision irrigation, resulting in more stable yield and quality performance, demonstrating the multiplier effect of combining smart agricultural management with ecological agricultural technology.

[0105] Compared to a control group using the same ecological planting method but with conventional seedlings, the No. 1 seedling obtained through hybridization and rapid propagation via tissue culture in Example 3 fully translated its genetic potential for disease resistance and high yield into field performance. The number of inflorescences per plant increased by approximately one-third, ultimately leading to a significant additional increase in yield per acre on top of the ecological planting method. This demonstrates that the ecological planting method of this application provides a platform for superior seedlings to fully express their genetic advantages, while the high-quality seedlings, in turn, amplify the output benefits of ecological planting, creating a virtuous cycle.

[0106] Example 4 achieved the highest level among all examples, thanks to the seamless integration of specialized organic fertilizer, purified superior seedlings, refined soil management, and intelligent irrigation, forming an "optimal environment-optimal seedling-optimal management" system. In terms of ecological benefits, a near-closed-loop material cycle was achieved, from localized raw materials and self-prepared fertilizer to complete degradation of the mulch film. Regarding system intelligence, the irrigation model not only possesses high-precision decision-making and disaster recovery capabilities but can also continuously adapt to specific field conditions through online optimization, enabling the system to achieve long-term self-improvement.

[0107] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A green and ecological method for cultivating daylilies, characterized in that, Includes the following steps: 1) Fertilization: Apply organic fertilizer made from cow and sheep manure and Asteraceae plant residues through aerobic composting; the Asteraceae plant residues are derived from common Asteraceae plants within the county or city administrative division where the planting site is located; 2) Land preparation and mulching: After preparing the land and forming planting ridges, first spread a layer of wood ash with a thickness of 2-3cm evenly on the soil surface of the ridges; then lay a black fully biodegradable mulch film with micro-ventilation holes arranged in a rectangular array on the mulch film. The pore diameter of the micro-ventilation holes is 0.5-1.0mm and the pore density is 300-500 per square meter. 3) Transplanting: Make holes in the plastic film to transplant daylily seedlings; In step 1), the fertilization operation adopts the layered strip application method, which is as follows: In the planting trench dug in the planting ridge, first apply organic fertilizer that has passed the high temperature decomposition period and is fully decomposed as the base fertilizer layer, cover it with soil 3-5cm, then apply organic fertilizer that has passed the main fermentation stage but has not completed the later aging as the middle layer fertilizer, and then cover it with soil and plant the seedlings.

2. The green and ecological planting method for daylilies as described in claim 1, characterized in that, It also includes field management, where irrigation steps are controlled by a pre-trained irrigation decision model, specifically including the following operational steps: a) Multi-source data acquisition and fusion: real-time soil temperature and humidity data collected by IoT sensor nodes pre-embedded in the planting ridges; simultaneously, real-time meteorological data and regional weather forecast data for the next 72 hours are acquired by accessing the meteorological data platform through the application programming interface; and multi-source satellite remote sensing data are accessed through the satellite data service interface, including at least thermal infrared surface temperature data and microwave surface humidity data for inverting regional soil moisture status, as well as normalized vegetation index data for monitoring crop growth. b) Irrigation decision generation: The multi-source heterogeneous data obtained in step a) is spatiotemporally aligned and fused, and then input into the irrigation decision model along with the pre-set water requirement threshold parameters that match the current growth stage of daylily. The model is trained using a hybrid deep learning algorithm that combines convolutional neural networks and long short-term memory networks, with historical irrigation data, crop growth data and corresponding yield and quality data as labels. It is used to output decision instructions that include specific irrigation time, duration and water volume. c) Command issuance and execution: The decision command generated in step b) is sent to the irrigation controller deployed in the field via a wireless Internet of Things protocol. The controller then drives the solenoid valve of the drip irrigation system to perform quantitative irrigation.

3. The green and ecological planting method for daylilies as described in any one of claims 1 or 2, characterized in that, The organic fertilizer in step 1) is obtained through the following preparation method: S1. Raw material pretreatment and sorting: The mixture of cow and sheep manure and Asteraceae plant residue is divided into a first raw material part and a second raw material part according to a dry matter weight ratio of 60-70:30-40; the source of Asteraceae plant residue is determined based on the main Asteraceae plant species listed in the official agricultural records, floras or forestry resource survey reports of the county and city administrative divisions. S2. Asynchronous aerobic fermentation: S2.1 Fermentation of the first raw material: Adjust the moisture content of the first raw material to 55-60% and the carbon-nitrogen ratio to 25-30:1, and carry out windrow aerobic composting; when the temperature at the center of the compost pile remains above 55℃ for 10-15 days, it enters the aging stage until it is fully decomposed, yielding organic fertilizer as the bottom fertilizer; the criteria for determining full decomposition are: the carbon-nitrogen ratio is stable within the range of 15-18:1, and the seed germination index is ≥90%; S2.2 Fermentation of the second raw material: After the first raw material enters the aging stage, the second raw material begins composting fermentation, with the initial fermentation conditions being the same as in S2.1; when the pile of the second raw material experiences a high temperature stage that is continuously above 55℃, and its temperature drops from the peak and stabilizes in the 35-40℃ range, active turning is stopped, keeping it in a semi-composted state to obtain organic fertilizer as a middle layer fertilizer; the carbon-nitrogen ratio of this middle layer fertilizer is 25-30:1; When the first raw material part is fully decomposed, the middle layer fertilizer of the second raw material part, which is in a semi-decomposed state, is obtained and applied in layers.

4. The green and ecological planting method for daylilies as described in claim 2, characterized in that, The pre-training method for the irrigation decision model specifically includes the following steps: Historical datasets were collected and integrated, including: historical multi-source heterogeneous data, which includes historical soil temperature and humidity time-series data, historical meteorological data, and historical satellite remote sensing data, with the satellite remote sensing data including at least historical thermal infrared band surface temperature data, microwave band soil surface humidity data, and normalized vegetation index data; historical irrigation operation records aligned with the historical multi-source heterogeneous data in time and space; and historical daylily yield and quality index data corresponding to the historical irrigation operation records. The constructed historical dataset is cleaned, missing values ​​are imputed, and outliers are handled. Then, spatiotemporal alignment and fusion are performed to unify data from different sources and formats into the same spatiotemporal grid. Finally, features are extracted and constructed to generate model input feature vectors containing spatiotemporal features and crop growth stage identifiers. A two-stream hybrid neural network model is constructed, where the first stream is a convolutional neural network branch used to extract spatial distribution features and texture features from the fused spatialized data; the second stream is a long short-term memory network branch used to extract time-series dependent features from historical soil temperature and humidity time-series data and historical meteorological time-series data; the high-level features output by the CNN branch and the LSTM branch are concatenated and fused, and then mapped to the irrigation decision action space through a fully connected layer; The preprocessed training dataset is divided into training, validation, and test sets. Historical irrigation operation records are used as action labels, and historical yield and quality index data are used as auxiliary supervision signals. A composite loss function, including mean squared error loss and a multi-objective reward function based on yield-water use efficiency, is used to perform end-to-end supervised training and optimization of the hybrid deep learning model. Early shutdown and hyperparameter tuning are performed through the validation set to finally obtain the pre-trained irrigation decision model.

5. The green and ecological planting method for daylilies as described in claim 2, characterized in that, Step b) Irrigation decision generation also includes a data disaster recovery and degradation decision module: When real-time soil temperature and humidity data or key satellite remote sensing data are continuously missing for more than a preset threshold, the irrigation decision model automatically switches to a degraded decision mode. In the downgraded decision-making mode, the model generates irrigation decision instructions based on the last effective soil data, historical meteorological data for the same period, and the 24-hour weather forecast obtained through the application programming interface, combined with the water requirement experience model of the current growth stage of daylilies.

6. The green and ecological planting method for daylilies as described in claim 2, characterized in that, It also includes an online optimization step for irrigation decision model parameters: At the end of each growing season, the system collects complete actual irrigation execution data, corresponding multi-source monitoring data, and final yield and quality data for that season to form a new set of training samples. These new samples are used to incrementally learn or fine-tune the pre-trained irrigation decision model, focusing on optimizing the water demand threshold parameter so that the model can adapt to the soil water retention characteristics and microclimate environment of specific fields.

7. The green and ecological planting method for daylilies as described in claim 1, characterized in that, After covering the soil with 3-5cm of soil in step 1), before applying the intermediate layer fertilizer, first lay a 1-2cm thick isolation and conditioning layer made of fully decomposed organic fertilizer and bentonite mixed in a 5:1 weight ratio; apply the intermediate layer fertilizer on top of this isolation and conditioning layer.

8. The green and ecological planting method for daylilies as described in claim 1, characterized in that, Before laying the black fully biodegradable mulch in step 2), on top of the layer of wood ash evenly spread on the ridge surface, lay a layer of rice husk charcoal or straw fragments with a thickness of 0.5-1.0cm and a particle size of 2-4mm, and then cover it with mulch. After laying the black fully biodegradable mulch in step 2) and before planting in step 3), add a mulch reinforcement step: evenly cover the surface of the laid mulch with a layer of fine sand and clay mixture with a thickness of 0.5-1.0cm; or, lay a mesh reinforcement layer made of biodegradable plant fibers on the surface of the mulch.

9. The green and ecological planting method for daylilies as described in claim 1, characterized in that, Daylily seedlings were obtained through the following hybridization and breeding methods: 3.1) Parent selection: Wild daylilies growing at altitudes above 800 meters were selected as the female parent, and superior cultivated varieties with strong disease resistance and a large number of inflorescences were selected as the male parent; 3.2) Artificial hybridization: During the peak flowering period of the female parent, take the healthy and tender flower buds that open on the same day, remove the male parts, pollinate them with the new fresh pollen of the male parent, and then isolate them by bagging. 3.3) Seed treatment and sowing: Harvest the hybrid seeds, disinfect and soak them in warm water, and then sow them in the seedling substrate made of organic fertilizer and peat in claim 1 at a volume ratio of 1:3; 3.4) Seedling selection and transplanting: When the seedlings grow to 4-5 true leaves, select strong and well-developed individual plants and transplant them to the nursery. The following year, based on the plant's resistance, number of inflorescences and bud morphology, select the best-performing strains as seedlings for the planting process.

10. The green and ecological planting method for daylilies as described in claim 9, characterized in that, After selecting strains with excellent overall performance in step 3.4), an additional step of seedling purification and rapid propagation is added: Rhizome meristems of superior strains were selected as explants for in vitro tissue culture to induce the formation of clustered shoots, and tissue culture seedlings were obtained through rooting culture. After hardening offspring, tissue culture seedlings are transplanted to an isolated propagation nursery, and the resulting vegetatively propagated offspring are used as seedlings for the planting process.