概率神经网络与BP网络模型在遥感图像分类中的对比研究

通过分析概率神经网络(以下称PNN)的基本结构及其训练算法,建立了卫星图像分类的概率神经网络模型,并通过实例对比分析了概率神经网络与BP网络分类模型的分类效果。实验表明,PNN图像分类方法在分类精度上优于误差反向传播神经网络模型,且分类时间相当,是一种有效的图像分类方法。     

时间动态地图模型

就时间动态地图模型进行了研究，认为动态地图由于表达空间维数的增加，使得地图作为客观现实世界的概念模型更加完善，并且作为物质模型，动态地图的可视化应与空间认知的计算机表达紧密融合。在制作时间动态地图时，不同的视觉变量具有不同的动态表现能力，根据时态现象的特征进行选取。     

武夷山脉主峰黄岗山两种山地草甸土的研究

选取武夷山脉主峰黄岗山典型山地草甸土和黄壤性山地草甸土的代表剖面，研究了土壤形成的条件和特点，推断在气候和植被条件相近的情况下，地形因素的差异引起的水环境的不同，是造成两种土壤理化性质差异的主要原因。     

Mixing of event and pre‐event water in a shallow Entisol in sloping farmland based on isotopic and hydrometric measurements,SW China

Water percolation and flow processes in subsurface geologic media play an important role in determining the water source for plants and the transport of contaminants or nutrients, which is essential for water resource management and the development of measures for pollution mitigation. During June 2013, the dynamics of the rainwater, soil water, subsurface flows and groundwater in a shallow Entisol on sloping farmland were monitored using a hydrometric and isotopic approach. The results showed that effective mixing of rainwater and soil water occurred in hours. The rebound phenomenon of δD profiles in soils showed that most isotope‐depleted rainwater largely bypassed the soil matrix when the water saturation in the soil was high. Preferential‐flow, which was the dominant water movement pattern in the vadose zone, occurred through the whole soil profile, and infrequent piston‐flow was mainly found at 20–40 cm in depth. The interflow in the soil layer, composed of 75.2% rainwater, was only generated when the soil profile had been saturated. Underflow in the fractured mudrock was the dominant flow type in this hillslope, and outflow was dominated by base flow (groundwater flow) with a mean contribution of 76.7%. The generation mechanism of underflow was groundwater ridging, which was superimposed upon preferential‐flow composed mainly of rainwater. The quick mixing process of rainwater and soil water and the rapid movement of the mixture through preferential channels in the study soil, which shows a typical bimodal pore size distribution, can explain the prompt release of pre‐event water in subsurface flow. Water sources of subsurface flows at peak discharge could be affected by the antecedent soil water content, rain characteristics and antecedent groundwater levels. Copyright © 2016 John Wiley & Sons, Ltd.     

Efficient estimation of flood forecast prediction intervals via single‐ and multi‐objective versions of the LUBE method

Prediction intervals (PIs) are commonly used to quantify the accuracy and precision of a forecast. However, traditional ways to construct PIs typically require strong assumptions about data distribution and involve a large computational burden. Here, we improve upon the recent proposed Lower Upper Bound Estimation method and extend it to a multi‐objective framework. The proposed methods are demonstrated using a real‐world flood forecasting case study for the upper Yangtze River Watershed. Results indicate that the proposed methods are able to efficiently construct appropriate PIs, while outperforming other methods including the widely used Generalized Likelihood Uncertainty Estimation approach. Copyright © 2016 John Wiley & Sons, Ltd.     

Discharge‐sediment processes of the Zhadang glacier on the Tibetan Plateau measured with a high frequency data acquisition system

In high elevation cold regions of the Tibetan Plateau, suspended sediment transfer from glacier meltwater erosion is one of the important hydrological components. The Zhadang glacier is a typical valley‐type glacier in the Nyainqentanglha Mountains on the Tibetan Plateau. To make frequent and long period records of meltwater runoff and sediment processes in the very high elevation and isolated regions, an automatic system was installed near the glacier snout (5400 m a.s.l) in August 2013, to measure the transient discharge and sediment processes at 5‐min interval, which is shorter than the time span for the water flow to traverse the catchment from the farthest end to the watershed outlet. Diurnal variations of discharge, and suspended sediment concentration (SSC) were recorded at high frequency for the Zhadang glacier, before suspended sediment load (SSL) was computed. Hourly SSC varied from the range of 0.2 kg/m³ to 0.5 kg/m³ (at 8:00–9:00) to the range of 2.0 kg/m³ to 4.0 kg/m³ (at 17:00–18:00). The daily SSL was 32.24 t during the intense ablation period. Hourly SSC was linearly correlated with discharge (r = 0.885**, n = 18, p < 0.01). A digit‐eight hysteresis loop was observed for the sediment transport in the glacier area. Air temperature fluctuations influence discharge, and then result in the sediment variations. The results of this study provide insight into the responses of suspended sediment delivery processes with a high frequency data in the high elevation cold regions. Copyright © 2016 John Wiley & Sons, Ltd.     

The effect of expansion ratio on the critical Reynolds number in single fracture flow with sudden expansion

Flow in a single fracture (SF) is an important research subject in groundwater hydrology, hydraulic engineering, radioactive nuclear waste repository and geotechnical engineering. An abruptly changing aperture is a unique type of SF. This study discusses the relation between the values of the critical Reynolds number (Re_c) for the onset of symmetry breaking of flow and the expansion ratio (E) of SF, which is defined as the ratio between the outlet (D) and inlet (d) apertures. This study also investigates the effect of inlet aperture d on Re_c for flow in an SF with abruptly changing apertures (SF‐ACA) using the finite volume method. Earlier numerical and experimental results showed that flow is symmetric in respect to the central plane of the SF‐ACA at small Reynolds number (Re) but becomes asymmetric when Re is sufficiently large. Our simulations show that the value of Re_c decreases with the increasing E, and the relationship between the logarithm of Re_c and E can be described accurately using either a quadratic polynomial function or a logarithmic function. However, the relationship of Re_c and d for a given E value is vague, and Re_c becomes even less sensitive to d when E increases. This study also reveals that the hydraulic gradient (J) and flow velocity (v) follow a super‐linear relationship that can be fitted almost perfectly by the Forchheimer equation. The inertial component (J_i) of J increases monotonically with Re, whereas the viscous component (J_v) of J decreases monotonically with Re. The Re value corresponding to equal inertial and viscous components of J (named as the transitional point Re) decreases when E increases, and such a transitional point Re should be closely related to the critical Reynolds number Re_c, although a rigorous theoretical proof is not yet available. Copyright © 2015 John Wiley & Sons, Ltd.     

Assessing the effectiveness of imperviousness on stormwater runoff in micro urban catchments by model simulation

Imperviousness, considered as a critical indicator of the hydrologic impacts of urbanization, has gained increasing attention both in the research field and in practice. However, the effectiveness of imperviousness on rainfall–runoff dynamics has not been fully determined in a fine spatiotemporal scale. In this study, 69 drainage subareas <1 ha of a typical residential catchment in Beijing were selected to evaluate the hydrologic impacts of imperviousness, under a typical storm event with a 3‐year return period. Two metrics, total impervious area (TIA) and effective impervious area (EIA), were identified to represent the impervious characteristics of the selected subareas. Three runoff variables, total runoff depth (TR), peak runoff depth (PR), and lag time (LT), were simulated by using a validated hydrologic model. Regression analyses were developed to explore the quantitative associations between imperviousness and runoff variables. Then, three scenarios were established to test the applicability of the results in considering the different infiltration conditions. Our results showed that runoff variables are significantly related to imperviousness. However, the hydrologic performances of TIA and EIA were scale dependent. Specifically, with finer spatial scale and the condition heavy rainfall, TIA rather than EIA was found to contribute more to TR and PR. EIA tended to have a greater impact on LT and showed a negative relationship. Moreover, the relative significance of TIA and EIA was maintained under the different infiltration conditions. These findings may provide potential implications for landscape and drainage design in urban areas, which help to mitigate the runoff risk. Copyright © 2015 John Wiley & Sons, Ltd.     

Size and spatial distribution of loess slides on the Chinese Loess Plateau

ABSTRACT

The size and spatial distribution of loess slides are important for estimating the yield of eroded materials and determining the landslide risk. While previous studies have investigated landslide size distributions, the spatial distribution pattern of landslides at different spatial scales is poorly understood. The results indicate that the loess slide distribution exhibits a power-law scaling across a range of the size distribution. The mean landslide size and size distribution in the different geomorphic types are different. The double Pareto and inverse gamma functions can coincide well with the empirical probability distribution of the loess slide areas and can quantitatively reveal the rollover location, maximum probability, and scaling exponents. The frequency of loess slides increases with mean monthly precipitation. Moreover, point distance analysis showed that > 80% of landslides are located < 3 km from other loess slides. We found that the loess slides at the two study sites (Zhidan and Luochuan County) in northern Shaanxi Province, China show a significant clustered distribution. Furthermore, analysis results of the correlated fractal dimension show that the landslides exhibit a dispersed distribution at smaller spatial scales and a clustered distribution at larger spatial scales.