Heatwaves have become the serious threat to the comfort and lives of urban residents. The cooling effects of urban tree and lawn through transpiration are regarded as a potential way to address these challenges, but their effects on heatwaves and mechanism remained unclear. Here, using a 10-year observation, we investigate the transpiration responses of urban lawn and a tree to 54 heatwave events in a subtropical city. We hypothesize that urban trees and lawns exhibit distinct transpiration response patterns during heatwaves due to different water use strategies and stomatal regulations. The findings reveal that (1) the lawn maintained high canopy stomatal conductance (Gs) during heatwaves, resulting in a 42.3% increase in transpiration rates (from 2.39 to 3.40 mm day− 1). In contrast, the tree significantly reduced Gs, maintaining relatively stable transpiration rates (slightly decreasing from 51.98 to 48.27 g m− 2 s− 1). (2) the lawn transpiration was highly dependent on soil water content (SWC), with rapid SWC depletion limiting sustained transpiration increases. Conversely, the tree accessed deeper soil water layers, enabling more stable transpiration throughout heatwaves. Urban tree responded to heatwaves much better than that of urban lawn. These results are of great importance for advancing knowledge in urban green space planning and water management.

Heatwaves are an increasing threat to urban health and comfort, and evapotranspiration by urban lawns and trees offers a potential solution. However, their distinct effects and mechanisms remain unclear. Using ten years of observations, we investigate the evapotranspiration responses of urban lawns and trees to 54 heatwave events in a subtropical city. We hypothesize that urban trees and lawns exhibit distinct evapotranspiration response patterns during heatwaves due to different water-use strategies and stomatal regulations. Our results show that (1) lawns, with high canopy stomatal conductance, rapidly increase evapotranspiration (+ 37.65

With rapid urbanization and climate change, water consumption and land-use pattern has dramatically changed, resulting in altered eco-hydrological processes and high ecological water requirements in megacities. However, the water uptake strategies may differ in urban and natural environment, and which remains largely unknown. Therefore, this study investigated the water use patterns of two greening plants species (Ficus concinna and Ligustrum vicaryi) and their responses to rainfall events in a megacity of subtropical China using the stable isotope methods. The results indicated that the two greening plants species showed different water use strategies. F. concinna mainly absorbed water from the shallower soil layer (0-20 cm, 56.29 %) in the wet season and deeper soil water (30-50 cm, 41.13 %) in the dry season, whereas L. vicaryi mainly relied on the shallower soil water (0-20 cm, 48.28 %) throughout the whole year. L. vicaryi absorbed water from the shallower soil layer (0-20 cm) before rainfall events and changed into deeper soil water (30-50 cm) after rainfall events in both dry and wet season; on the contrary, F. concinna did not show these dynamics throughout the year. These results suggested that the water use pattern of F. concinna showed more ecological plasticity, facilitating the adaptation of the plant to seasonal drought and other environment fluctuations in subtropical China urban areas.

The urban green infrastructure such as the low impact development (LID) facility and traditional garden that are relatively small and characterized by decentralized distributions has been proposed as the most effective way to mitigate urban heat through its evaporative cooling effect. Recently, there have been increasing studies on its temperature reduction and evapotranspiration (ET) rate, but few of them correlate ET with external surface temperature reductions. Therefore, this study investigated the evaporative cooling effects, ET rates, and their relationships by the three-temperature (3T) model and ground-based thermal infrared remote sensing. Results show that the cooling effect of both vegetated LID facilities and traditional gardens is significantly stronger than that of non-vegetated LID facilities. Due to a thinner soil layer and lower water connectivity of LID facilities, their ET rates are significantly reduced in the dry period while the evaporative cooling effect of traditional gardens covered by the same vegetation can maintain high. The dependency of their cooling effect can be largely explained by the ET rates. When ET < 0.6 mm h(-1), an increase in ET of 0.1 mm h(-1) can enhance the cooling effect by 3.66 degrees C. When ET exceeds 0.6 mm h(-1), the evaporative cooling effect saturates. Vegetation types and soil water conditions are two main factors that govern evaporative cooling effect. Specifically, shrubs with higher ET rates are more efficient in urban heat mitigation than herbs. The responses of the evaporative cooling effect to soil water availability vary among species, which may require species-specific irrigation regime. These results may have implications on the best management practices for urban heat mitigation by the small widely- distributed green spaces.

Hyperspectral image (HSI) classification has attracted wide attention in many fields. Applying Graph Neural Network (GNN) to HSI classification is one of the research frontiers, which has improved the HSI classification accuracy greatly. However, GNN-based methods have not been widely applied due to their time-consuming, inefficient information description as well as poor anti-noise robustness. To overcome the deficiencies, a novel multi-scale receptive fields graph attention neural network (MRGAT) is proposed for HSI classification in this paper. In this network, a superpixel segment method is adopted to abstract the original HSI local spatial features. A two-layer one-dimensional convolution neural network (1D CNN) spectral transformer mechanism, is designed to extract the spectral features of superpixels, with which the spectral features can be acquired automatically. Furthermore, graph edges are introduced into Graph Attention Network (GAT) to acquire the local semantic feature of the graph. Moreover, inspired by the transformer network, we design a novel multi-scale receptive field GAT to extract the local-global adjacent node-features and edges-features. Finally, a graph attention network and a softMax function are utilized for multi-receptive feature fusion and pixel-label predicting. On Pavia University, Salinas, and Houston 2013 datasets, the overall accuracies (OAs) of our MRGAT are 71.76%, 82.61%, and 63.82%, respectively. Moreover, the performances with limited labeled samples indicates that the MRGAT contains superior adaptability. Compared with the competitive classifiers, MRGAT achieves high classification efficiency verified by training time comparison experiment.

Heavy metal contamination in soils can pose severe challenges to the safety of geotechnical engineering projects. Loess, which is widely distributed in Northwest China, is a preferred engineering construction material for anti-fouling barriers. Therefore, research on the influence of heavy metal ions on its seepage performance is urgently required. To obtain new insights into the seepage behavior of heavy metal-contaminated loess and its underlying geochemical mechanism, laboratory investigations were performed on the saturated hydraulic conductivity (Ksat), leaching, and microstructural characteristics of loess contaminated with Cu2+ and Zn2+. The results indicate that the hydrolysis of Zn2+ creates an acidic environment, which promotes the dissolution of carbonate minerals in loess, enhances the leaching capacity, and leads to the quantitative transformation of small pores (2–8 μm) to mesopores (8–32 μm). Meanwhile, the alternating adsorption of Zn2+ and its diffuse double-layer effect compresses the diffusion layer, increasing the abundance of free water channels. Thus, the Ksat of Zn-contaminated loess increases by 81.2% during the seepage period. As for Cu-contaminated loess, its seepage behavior is the opposite of that of Zn-contaminated loess, with a Ksat decrease of nearly 50%. The primary factor controlling this phenomenon is the formation and enrichment of Cu2O in the lower part of the soil, which inhibits the enlargement of pores and reduces the effective connectivity of pores. The findings of this work provide insight into the seepage behavior of saturated loess under erosion by heavy metals and the underlying geochemical mechanism thereof.

Wetland evapotranspiration (ET), which involves the land-atmosphere exchange of energy and water, is dynamic and affects the spatiotemporal distribution of water resources. However, due to the variability and complexity of wetlands, accurate estimation of patch-scale ET and its spatial variability remain insufficiently characterized. To overcome this challenge, an advanced unmanned aerial vehicle (UAV) technology was developed by combining the three-temperature (3T) model, which is robust to estimate transpiration and its spatial variability with UAVbased thermal infrared remote sensing, and Penman equation, which is commonly used to estimate open water evaporation. The combined approach was verified using the Bowen ratio system over a subalpine wetland. The results show that the proposed method is simple and applicable for estimating wetland ET and its spatial variability, with a determination coefficient (R2) of 0.93, mean absolute percentage error (MAPE) of 7.90%, root mean squared error (RMSE) of 0.05 mm h-1, and Nash-Sutcliffe efficiency (NSE) of 0.93. It depicts a large spatial variability in wetland ET with respect to surface vegetation characteristics, water regimes, meteorological factors, and larger transpiration rates than open water evaporation. With its limited inputs and no calibration requirements, the proposed method is concluded to be simple and to easily reveal the high temporal and spatial resolution characteristics of patch-scale ET and its components.

Evapotranspiration (ET) cooling of urban spaces is an effective and economical way to improve the urban thermal environment. However, the distribution of urban ET rate is typically unknown owing to the high heterogeneity of urban land covers, which limits the application of many conventional techniques for measuring ET, such as ground-based observations and satellite remote sensing. In this study, an improved approach called "UAV + IRs + 3T", combining unmanned aerial vehicle (UAV), thermal infrared remote sensing, and a threetemperature model (3T), was developed for estimating urban ET and validated by Bowen ratio method. Results showed that the proposed method could accurately measure urban LE with R2 = 0.95, MAE = 21.98 W m- 2, RMSE = 30.33 W m- 2, RRMSE =19.65%. The proposed method could obtain urban ET with an ultra-high spatial resolution (approximately 15.5 cm) and temporal resolution (once per hour). Furthermore, 9 plant species distributed across the 18 sample plots showed significant differences in mean intra-day ET rates. Even for the same plant species at different sites, such as Ficus concinna and Zoysia matrella, their average intra-day ET rates differed by 50% and 400%, respectively. These large differences could be attributed to artificial pavement and infrastructure, different artificial irrigation methods, and difference in artificial and natural shade. In conclusion, there is spatio-temporal variability in urban ET rates, which can be precisely revealed by the proposed method. Therefore, the "UAV + IRs + 3T" method has the potential for a wide range of applications in urban environmental planning.

In response to a decline in pan evaporation over the last 60 years under global warming of water bodies, we designed an experiment with water bodies heated naturally to different temperatures to investigate the physics of pan evaporation and explore the effect of water temperature. In this study, we developed a new aerodynamic model for pan evaporation by combining a free convention sub‐model that couples Fick's First Law of Diffusion with boundary layer theory and a forced convection sub‐model based on convection mass transfer. Both the improved aerodynamic model and the two sub‐models have a higher accuracy and stability (|PBIAS| < 6%, root mean squared Error [RMSE] < 0.65 mm d −1 , and Nash‐Sutcliffe efficiency [NSE] > 0.8). Sensitivity analysis shows that water temperature is the most sensitive parameter to evaporation ( S 1 = 0.58, S T = 0.78). The mechanism of rising water temperature on evaporation is not only due to the strengthening in mass diffusion (under windless conditions), but also in the promotion in mass convection (under windy conditions). The integrated effects of mass diffusion and mass convection could result in an increase in evaporation of 0.8 mm d −1 , as mean water temperature rises by 1°C. These results would be useful for evaporation estimation of the warming global water.

利用一对设置在相同气象条件下的黑白标准A型蒸发器皿,观测两者气象要素、水温及蒸发量等动态特征,并根据现有的6个蒸发模型,探究水温对器皿蒸发量的影响.结果表明:1)设计的观测方法可以用于研究水温对器皿蒸发的影响.黑皿与白皿之间的水温和蒸发速率呈现明显的差异,在50天的观测期内,二者平均水温差为0.4℃,平均日蒸发量差为1.1 mm/d;2)在太阳辐射及其他气象要素相同的条件下,黑白皿水温差每升高1.0℃,由水温驱动的日蒸发量差异为0.808 mm/(d·℃);3)在水温升高条件下,没有考虑水温的经典蒸发模型的估值小于观测值,并且估值的误差也有所增加.

With ongoing climate change and rapid urbanization, the influence of extreme weather conditions on long-term nocturnal sap flow (Qn) dynamics in subtropical urban tree species is poorly understood despite the importance of Qn for the water budgets and development plantation. We continuously measured nighttime sap flow in Ficus concinna over multiple years (2014–2020) in a subtropical megacity, Shenzhen, to explore the environmental controls on Qn and dynamics in plant water consumption at different timescales. Nocturnally, Qn was shown to be positively driven by the air temperature (Ta), vapor pressure deficit (VPD), and canopy conductance (expressed as a ratio of transpiration to VPD), yet negatively regulated by relative humidity (RH). Seasonally, variations in Qn were determined by VPD in fast growth, Ta, T/VPD, and meteoric water input to soils in middle growth, and RH in the terminal growth stages of the trees. Annual mean Qn varied from 2.87 to 6.30 kg d−1 with an interannual mean of 4.39 ± 1.43 kg d−1 (± standard deviation). Interannually, the key regulatory parameters of Qn were found to be Ta, T/VPD, and precipitation (P)-induced-soil moisture content (SMC), which individually explained 69, 63, 83, and 76% of the variation, respectively. The proportion of the nocturnal to the total 24-h sap flow (i.e., Qn/Q24-h × 100) ranged from 0.18 to 17.39%, with an interannual mean of 8.87%. It is suggested that high temperatures could increase transpirational demand and, hence, water losses during the night. Our findings can potentially assist in sustainable water management in subtropical areas and urban planning under increasing urban heat islands expected with future climate change.

The middle Heihe River Basin (MHRB) in arid areas in China faces the challenge of sustainability, but frequent over-irrigation deteriorated the situation. The water balance and energy budget have not been widely used for the quantifying evaluation of over-irrigation. Although some studies have measured the energy budget and water balance in cropland, the energy and water balance in pepper fields is still unclear. An experiment was conducted using an eddy covariance system, SmartView fluke and other measurements in MHRB. The leaf area index (LAI) for pepper increased in the first two months, then decreased thereafter. After Aug. 4, the daily evapotranspiration (ET) decreased during the maturation stage. The soil moisture was higher than the field capacity frequently, indicating that over-irrigation had occurred. Soil water content was relatively high at 100 cm depth, causing high percolation. The water inputs were 1.5 times ET, and the deep percolations accounted for 74% of the total irrigation inputs. Irrigation water could be conserved 301 mm, in case that percolation was avoided. Therefore, the irrigation amount (per time and total) should be reduced to achieve high water use efficiency (WUE). The latent heat flux (LE) consumed 80% of the net radiation (Rn). In July, the LE/Rn was the highest among all months, whereas H/Rn was the lowest due to the highest value of LAI in July. The net radiation had a significant positive linear correlation with the latent heat flux because of the high water supply. Our study showed that there was considerable room for water conservation in MHRB. Keywords: Sensible heat flux; Soil water content; Evapotranspiration; Latent heat flux; Leaf area index

Climate models predict rising temperatures and more frequent and prolonged urban heat islands (UHI) in southern China. The urban vegetations have become a focal point to mitigate the detrimental effects of UHI and to provide cooling through transpiration. However, transpiration in trees such as Ficus concinna in relation to extreme long-term conditions of UHI is scarcely documented. Here, we investigated the transpiration dynamics and its cooling effects in F. concinna induced by changes in site environmental variables in a subtropical megacity, Shenzhen over five consecutive years (2015-2019) based on continuous sap flow measurements. Seasonally, the transpiration (T-r) and its cooling effect (i.e., heat energy absorbed (Q) and temperature reduction (Delta T) by T-r) were highest in summer, peaking in the month of July with the mean values of 1.98 mm d(-1), 4.91 MJ m(-2) d(-1) and 3.93 degrees C m(-2) d(-1), respectively. The highest cooling effect was shown during warmer and wet years. Daily Tr had a positive linear relationship with shortwave radiation (R-s) in May-June, air temperature (T-a) and volumetric soil water content (SWC30) in July-August and vapor pressure deficit (VPD) in September, respectively. Furthermore, the main regulatory variables of T-r within each season were found to be the spring R-s, summer T-a and SWC30, and autumn Rs and VPD, which explain 49, 82 and 74% of the variation, respectively. The influence of these variables (i.e., T-a, R-s and VPD) on T-r was modified by the effect of SWC30. Interannually, T-a, SWC30 and precipitation (PPT) were responsible for most of the observed variation in T-r, individually explaining 72, 81, and 66% of the variation. Furthermore, multiple regression model indicated that together T-a, SWC30 and PPT explained 89% of the variation in T-r. The diminished sensitivity of T-r to environmental variables and enhanced sensitivity to SWC30 during dry years point to species acclimatization to soil dryness. Our findings clearly indicate the temporal dynamics in T-r and its cooling effectiveness for F. concinna over longer timescales. It is suggested that F. concinna could be better suited in response to increasing temperature in subtropical urban areas, as the species may be capable to provide efficient cooling based on its high T-r rate and has the potential to mitigate the UHI effect.

Drug repurposing aims at identifying new indications for approved drugs that are outside the scope of the original indications. Understanding the acting mechanism among drugs, protein targets, and diseases, especially the interdependent and indecomposable relationships, is a critical step. However, most existing methods rely on pairwise relationships. To model the biological interactions between the three types of entities, which are likely ignored by the pairwise paradigm, we propose an end-to-end Event-Graph Neural Network (EGNN) to predict multivariate relationships of drugs, targets, and diseases for drug repurposing. Specifically, we introduce the event to describe the interdependence of drug-target-disease as a complete semantic unit and design the Event-Graph to model the multivariate relationships. To predict the potential relationships, we perform the representation learning on the Event-Graph by a bidirectional aggregating operation and an event-level attention mechanism. Experimental results on real-world datasets demonstrate the effectiveness and promising performance of EGNN compared with several competitive methods.

This paper extends the existing MapReduce framework to allow the user programmer to control data locality and reduce communication costs of the shuffle operations in iterative in-memory computation. The programming extension is fully consistent with the style of MapReduce and allows straightforward fast implementation.

Urban shrubs are one of the main components of urban vegetation and play a significant role in mitigating the urban heat island (UHI) through transpiration. As there are no suitable methods to measure the transpiration rate of urban shrubs, many issues related with urban transpiration rate and UHI are still not properly understood. In this study, the two-dimensional (2D) three-temperature (3T) model is extended to three-dimensional (3D) scale and applied to estimate the transpiration rate of typical urban shrubs from May to December in 2018. Three common species of shrubs are observed in this study, namely Lagerstroemia indica cv "Bush", Podocarpus macrophyllus var. angustifolius, and Ligustrum vicaryi. Estimated transpiration rates are proofed by isotope method. Results show that (1) the modified 3D-3T model is a reliable method for estimating urban shrub transpiration (R-2 = 0.61, MAE = 0.10 mm.h(-1)); (2) urban shrubs have a quite high transpiration rate, ranging from 0.2 to 0.6 mm.h(-1) in the daytime; (3) the rate of transpiration varies greatly in different directions of the shrub canopy (top, south, north, east, and west); (4) the 3D-3T model is able to better reflect transpiration characteristics under the heterogeneous condition in urban environments. To the best of our knowledge, the improved 3D-3T model is one of the very few methods that can measuring the transpiration rate of urban shrubs. The results from this study provide meaningful information and a useful tool for measuring urban evapotranspiration and urban planning.

水中氢氧稳定同位素(D、18O)是示踪水分过程的重要方法,但不同设备D、18O测量结果一致性有待探索.基于内蒙古乌拉特荒漠草原的同批植被、土壤样本,经抽提获得水样后,分别使用两种同位素仪(LGR和PICARRO)测量氢氧同位素.结果表明:1)LGR测量精度更高;2)D的两组测量结果差异不显著(R2= 0.98)、一致性高;3)18O的两组测量结果一致性较差(R2=0.55)、差异显著(ANOVA,P＜0.05);4)设备稳定性和水样汽化浓度导致测量误差.

The urban heat island (UHI) effect is a widespread phenomenon because of increased urbanization, making the urban thermal environment less comfortable. The UHI effect may worsen during heat waves (HWs), with projected increases in extreme climatic events in the future due to global warming. Researchers have revealed interactions between the UHI effect and HWs using weather station data and proposed mitigation schemes at a city scale. However, the UHI effect in urban areas with different land use/land cover (LULC) types should respond differently to HWs, which has drawn little attention. Hence, this study conducted a mobile transect experiment in the subtropical megacity of Shenzhen and obtained high spatial resolution data. The results showed that the UHI effect was significantly amplified during HWs. The UHI intensity (UHII) of the transect increased from 0.56 +/- 0.50 degrees C under non-heat wave (NHW) conditions to 0.68 +/- 0.65 degrees C during HWs. LULC types had a significant influence on this interaction. The UHII in more urbanized areas increased during HWs, whereas less urbanized areas had improved cooling effects. These interactions were more evident at nighttime. Increasing the natural underlying surface coverage mitigated the intensity and warming potential of the UHI effect. With a 10% increase in the natural underlying surface coverage, the nighttime UHII decreased by 0.38 degrees C and 0.39 degrees C during NHWs and HWs, respectively. These cooling effects were attributed to the increased latent heat consumption during HWs by vegetation. Therefore, different measures should be taken in other areas to mitigate UHII amplification during HWs.

The binding affinity between drugs and proteins is a substantial part of the drug discovery process. Graph neural networks (GNNs) have shown great promise in graph-related structure by learning the representations of graphs, which are suitable for tasks such as binding affinity prediction. However, most of the existing GNN architectures only pay attention to the information flow on a single graph, while the interaction between two graphs is unconcerned. In this paper, we propose an attention-enhanced graph cross-convolution network (GCAT) to explore binding affinity on pure 3D atomistic geometry. It consists of two components: cross-convolution and self-attention pooling. Specifically, cross-convolution performs an aggregate-update mechanism to simulate the interaction between the protein and the drug, then self-attention pooling is adopted to capture global interactions and get graph-level representations. Extensive experiments conducted on the PDB-Bind dataset demonstrate the effectiveness of our GCAT.

This work analyses different communication modes in applications of supercomputing, proposes a communication dynamic performance model based on topology awareness, and realizes the prototype system of all-to-all communication and stencil communication optimization based on this model. Basic tests on the optimization of all-to-all communication and stencil communication were carried out on the Sunway TaihuLight System, and this achieved obvious optimization results. Several applications, including molecular dynamics simulation and turbulence simulation, have been optimized and tested. The average performance has been improved obviously. It can be expected that, for other large-scale applications, this optimization method can also be used to obtain significant improvement in communication performance.