Protected agriculture plays a critical role in ensuring global food security by enabling stable crop production under controlled environmental conditions. However, climate change–driven extreme weather events have increased flood-related risks, posing growing threats to the safety, productivity, and sustainability of solar greenhouse cultivation. In practical greenhouse production, timely and accurate monitoring of waterlogging is essential for reducing crop losses, improving environmental regulation, and supporting scientific management decisions. Nevertheless, accurate segmentation of greenhouse waterlogging remains challenging because of limited monitoring visibility, uneven illumination, strong specular reflections, and the pronounced scale diversity of inundated areas. These adverse conditions often result in blurred boundaries, ambiguous textures, and unstable visual characteristics between flooded and non-flooded regions. In addition, satellite- and drone-based remote sensing is often unsuitable for greenhouse-scale applications because of resolution limitations and operational constraints. To address these challenges, this study developed an end-to-end image-based segmentation framework, termed the Residual Multiscale Cascade Encoder–Decoder (RMCED) network, for delineating waterlogging areas inside solar greenhouses. The proposed architecture adopted the middle four stages of ResNet-50 as the encoder to balance network depth and computational cost while extracting multi-level semantic features. This encoder design enabled the model to effectively capture both low-level texture patterns and high-level contextual representations, thereby improving discrimination between waterlogged areas and surrounding background regions. The decoder was designed as a multiscale cascaded attention decoder, which progressively restored spatial resolution through the hierarchical fusion of low- and high-level features. This design enhanced the network’s ability to identify fine spatial boundaries and accurately localize small-scale flooded regions that are often overlooked by conventional models. To ensure effective information flow, long skip connections were introduced between the encoder and decoder, thereby reducing feature degradation and semantic loss during feature transmission and improving the preservation of structural details. A key contribution of this work was the introduction of a Lighting-Aware Convolutional Block Attention Module (LA-CBAM). Compared with the original Convolutional Block Attention Module, LA-CBAM incorporated multi-kernel convolution filters and Sobel-gradient operations to improve both illumination robustness and edge sensitivity. The multi-kernel design enabled the model to capture contextual cues across different receptive fields and to better adapt to the diverse spatial scales of greenhouse waterlogging patterns. Meanwhile, the Sobel operation extracted gradient-based texture transitions and boundary-related information, strengthening boundary awareness under varying light intensities and reflective interference. As a result, the proposed network maintained strong discrimination capability even in complex greenhouse environments with reflections and uneven brightness, and it showed better adaptability to irregular and fragmented flooded regions. The model was trained and validated on a self-constructed dataset of 571 pixel-wise annotated images collected in a solar greenhouse under natural operating conditions. All images were resized and normalized using ImageNet statistics to match the pretrained ResNet-50 encoder. Comparative experiments with mainstream segmentation models, including U-Net, DeepLabV3+, and PSPNet, showed that RMCED outperformed representative baselines on the proposed dataset, achieving a precision of 93.22%. Ablation studies further demonstrated that LA-CBAM enhanced the model’s perception of key features in waterlogged areas, while the multiscale cascaded decoder enhanced edge continuity and object completeness. Overall, this study presented an efficient and robust framework for greenhouse waterlogging monitoring, providing a promising technical basis for flood risk assessment, waterlogging management, and related image-based monitoring tasks in protected agricultural systems.

This study aims to investigate the effect of rated flow rate and lateral distance on the risk of clogging in the button-type emitter during drip irrigation using high-sediment sediment-loaded water. Three rated flow rates (2, 4, and 8 L/h) of button-type emitters were subjected to the muddy water clogging tests, corresponding to the different lateral positions (lengths of branch pipe at the inlet of the lateral w, 2w, and 3w, respectively, with lateral spacing w=204 mm; referred to the inside, the middle, and the outside, in that order). A two-factor full-scale test was then carried out with a total of nine combinations. Three combinations were tested at one time, and three tests were conducted as one round, where a total of two rounds of tests were conducted to ensure the reliability. The results show that the rated flow rate of the emitter and the lateral distance posed the great impacts on the full cross-section mean flow rate in the lateral and the branch pipe, leading to the initiation of sediment deposition in the pipe, even in the process of emitter clogging. The emitter with the rated flow rate of 4 L/h presented the slowest decrease in the average relative flow rate and the coefficient of irrigation uniformity, indicating the best anti-clogging performance, and the highest number of effective irrigation times. The average service life increased by 11.84% and 49.11%, respectively, compared with the emitters with a rated flow rate of 2 and 8 L/h, respectively. The smaller the rated flow rate of the emitter was, the more significant the impact of the lateral distance on the service life of the emitter was. The lower the rated flow rate of the emitter was, the greater the mass of sediment retained in the lateral was, the greater the proportion of large-particle sediment deposited was, and the greater the tendency for the large-particle sediment to be retained in the lateral relative to small-particle sediment was. The emitter with the rated flow rates of 8, and 2 L/h shared the highest and lowest number of particles in the initiating motion of sediment deposited in the lateral, as well as the largest and smallest upper limit of particle size, respectively. The higher the rated flow rate of the emitter and the closer to the front of the installation on a single lateral were, the larger the sediment particle size it discharged. The initiation of sediment deposits in the lateral was the main cause of faster clogging of high-flow emitters. The amount of sediment entered the emitter, due to the too too-large rated flow rate of the emitter, the large velocity of water flow in the lateral, and the strong sand holding. The sediment particles then failed to discharge from the emitter in time to be easily deposited inside the emitter or flocculation and sedimentation by collision, leading to the emitter more susceptible to clogging. The clogging was easier to be washed away, due to the rated flow rate of the emitter for the 8 L/h of the flow channel cross-section dimensions (1.60 mm×1.08 mm) and the flow velocity of the muddy water within the channel is maximum. Repeated clogging was more likely to occur in the 8 L/h emitter than in the 4 and 2 L/h ones. The finding can provide a strong reference for the selection of a rated flow rate to prevent the emitter clogging in drip irrigation.

Y-type mesh filters have been widely used in the micro-irrigation systems. A better hydraulic performance is highly required for the stable operation during irrigation. In this study, the standard k-model was adopted to simulate the internal flow field of different filters using the computational fluid dynamics with multi-angle analysis. A series of simulation tests were carried out under three types of filter screen shapes (square, circular and diamond) and three angles of cylinder arc (0°, 15°, and 30°). Specifically, a systematic analysis was implemented on the hydraulic characteristics of the internal pressure drop coefficient, the flow distribution on the surface of the filter element, the internal flow field, and the pressure distribution. Physical tests were also conducted to verify the numerical simulation. The results show that: There was 9% average difference in the head loss coefficient between the physical test and the numerical simulation, indicating the better reliability of the numerical simulation. The head loss of the filter was concentrated on the outlet side of the screen, which was accounted for 85% of the total head loss. A stagnant zone of water flow was formed inside the plug, where the velocity was very low without the water flowing back. The smallest pressure was found in the center of the filter chamber from the center to the surrounding area. There was also an increase in the minimal pressure in a stepwise manner from the center to the periphery. Among them, the circular mesh filter shared the largest head loss coefficient, followed by the square mesh filter, and the smallest was found in the diamond mesh filter. The pressure dropped at the mesh, and the total pressure dropped to change, as the shape of the mesh changed. But there was no variation in the value and distribution of the maximum and minimum pressure in the chamber. It infers that the shape of the mesh posed a greater influence on the distribution of the overflow rate on the mesh surface of the filter. The highest proportion of medium-rate overflow area was 47.5% in the square mesh filters, followed by the circular shape, and the smallest medium-rate overflow area of diamond shape was only 26.5%. The head loss of the filter gradually decreased with the increase of the arc angle of the cartridge. The pressure drop coefficient at 35° decreased by 73.15%, compared with 0°. There was also the much more uniform distribution of the flow rate on the mesh surface with the increase of the angle, in which the area of the medium speed overflow area at 35° increased by 71.48%, compared with 0°, indicating the outstandingly improved hydraulic performance. There was the significant decrease in the internal and external pressure difference at the middle and upper section of the screen on the outlet side with the increase of the arc angle of the cartridge , particularly for the differences between 35° and 0°. The difference of the pressure drop between 35° and 0° was 2.97 times. Therefore, an optimal filter can be selected with a square cylinder arc angle of 30°in the actual micro-irrigation system, in order to improve the hydraulic performance and service life of the filter with the gentle internal flow field and uniform flow distribution on the mesh surface.