Monthly suspended sediment load prediction using artificial intelligence: testing of a new random subspace method

ABSTRACT

The predictive capability of a new artificial intelligence method, random subspace (RS), for the prediction of suspended sediment load in rivers was compared with commonly used methods: random forest (RF) and two support vector machine (SVM) models using a radial basis function kernel (SVM-RBF) and a normalized polynomial kernel (SVM-NPK). Using river discharge, rainfall and river stage data from the Haraz River, Iran, the results revealed: (a) the RS model provided a superior predictive accuracy (NSE = 0.83) to SVM-RBF (NSE = 0.80), SVM-NPK (NSE = 0.78) and RF (NSE = 0.68), corresponding to very good, good, satisfactory and unsatisfactory accuracies in load prediction; (b) the RBF kernel outperformed the NPK kernel; (c) the predictive capability was most sensitive to gamma and epsilon in SVM models, maximum depth of a tree and the number of features in RF models, classifier type, number of trees and subspace size in RS models; and (d) suspended sediment loads were most closely correlated with river discharge (PCC = 0.76). Overall, the results show that RS models have great potential in data poor watersheds, such as that studied here, to produce strong predictions of suspended load based on monthly records of river discharge, rainfall depth and river stage alone.     

Regional hydromorphological characterization with continuous and automated remote sensing analysis based on VHR imagery and low‐resolution LiDAR data

Recent developments in remote sensing (RS) technologies lead the way in characterizing river morphology at regional scales and inferring potential channel responses to human pressures. In this paper, a unique regional database of continuous hydromorphological variables (HyMo DB) based on areal and topographic data has been generated from RS analysis. Key riverscape units with specific geomorphic meaning have been automatically mapped for 1700 km² of river floodplains from simultaneous very‐high‐resolution (VHR) near‐infrared aerial imagery and low‐resolution LiDAR‐derived products. A multi‐level, geographical object‐based architecture (GEOBIA) was employed to integrate both spectral and topographic information and generate a regional classifier able to automatically map heterogeneous fluvial patterns in different geographical and topographical contexts of the Piedmont Region (Italy). This HyMo‐generated DB offers a unique set of tools for hydromorphologists and can be exploited for different purposes. For the first time, topographic information can be exploited regionally per riverscape unit class, allowing for quantitative analysis of their regional spatial and statistical variability. In this manner, river types can be automatically characterized and classified using objective and repeatable hydromorphological variables. We discuss the potential of quantifying functional links between riverscape units and their driving processes, a valuable source of information to start assessing and highlighting the entity of potential channel adjustments at the regional scale to human pressures. The HyMo DB can also be integrated with historical, field‐based information to better comprehend current fluvial changes at a local scale. In view of future RS acquisitions, the present approach will result in a suitable procedure for quantitative, objective and continuous monitoring of river evolutions over large scales. This type of hydromorphological characterization will allow regional trends and patterns to be highlighted through time and river management strategies to thus be implemented at both regional and local scales. Copyright © 2016 John Wiley & Sons, Ltd.     

LiDAR-based fluvial remote sensing to assess 50–100-year human-driven channel changes at a regional level: The case of the Piedmont Region,Italy

Remote Sensing (RS) technology has recently offered new and promising opportunities to analyze river systems. In this paper, we present a calibration of characteristic Hydraulic Scaling Law (HSL) using a regional database of river geomorphic features. We consistently linked discharge with channel geometry features for estimated Bankfull Channel Depth (eBCD), Active Channel Width (ACW), and Low Flow water Channel Width (LFCW), which are continuously available from RS data along the river course. We then used historical information and external sources of information on channel reaches that were relatively unaffected by human pressure over periods ranging from a few decades to a century (measured in comparable geographical areas) to infer relatively Unaltered HSLs (rUHSLs). Adopting rUHSL validated with available local historical evidence on channel geometry, we were able to assess historical changes in channel geometry consistently over the entire region and within the studied temporal window. The case study was conducted for the Po basin in the Piedmont Region, north-west Italy. From our analysis, it emerges that regionally 74% of the river network has riverbed incisions exceeding 1 m, while 66% of channels have halved their historical widths with a total of 617 ha of land subtracted from the active channel. LFCW is, on average, wider in Alpine rivers compared with those located in the North Apennines. Although it is currently not possible to measure the accuracy of these estimates, the evidence generated is coherent with available historical information, characteristic hydraulic scaling laws, evidence from relatively unaltered reaches and the available literature on local fluvial systems. This methodology provides robust, novel and quantitative information regarding decadal to secular channel changes that have occurred on a regional scale. This new layer of information enriches our ability to rationally address assessments of large-scale past and future channel trajectories. © 2018 John Wiley & Sons, Ltd.     

Responses of distribution pattern of desert riparian forests to hydrologic process in Ejina oasis

By using the theories and methods of landscape ecology and the technology of GIS and RS, a study has been carried out on the responses of distribution pattern of desert riparian forest to hydrologic process on the basis of the hydrologic data from 1990 to 2000 and the TM image of 2001 year. The results showed that: (1) there appears an even distribution pattern for the relative forest area in oasis, however, the degenerated forest diaplays an increasing tendency from west to east; (2) the desert riparian forest in Ejina is in completely degenerated process at the patch scale; (3) the number of patch is influenced not only by hydrologic process,but also by agricultural activity such as cultivation. The severe deterioration of the degraded vegetation in whole oasis initiates from lower reaches, and gradually impels to upstream; the fragmentation of landscape in the terminal site is more obvious, which is influenced by river shape and decreasing flux of water. It is found that the influence of surface hydrologic process to the ground hydrologic process of desert riparian forest in Ejina oasis is little for the recent ten years. The relative area of the degenerated forest increased with increasing ground water depth in the direction of parallel to river channel. On the contrary, in the direction perpendicular to river channel, there is a decreasing tendency for the average patch area of the forest and the degenerated forest with increasing ground water depth.     

Responses of distribution pattern of desert riparian forests to hydrologic process in Ejina oasis

By using the theories and methods of landscape ecology and the technology of GIS and RS, a study has been carried out on the responses of distribution pattern of desert riparian forest to hydrologic process on the basis of the hydrologic data from 1990 to 2000 and the TM image of 2001 year. The results showed that: (1) there appears an even distribution pattern for the relative forest area in oasis, however, the degenerated forest diaplays an increasing tendency from west to east; (2) the desert riparian forest in Ejina is in completely degenerated process at the patch scale; (3) the number of patch is influenced not only by hydrologic process, but also by agricultural activity such as cultivation. The severe deterioration of the degraded vegetation in whole oasis initiates from lower reaches, and gradually impels to upstream; the fragmentation of landscape in the terminal site is more obvious, which is influenced by river shape and decreasing flux of water. It is found that the influence of surface hydrologic process to the ground hydrologic process of desert riparian forest in Ejina oasis is little for the recent ten years. The relative area of the degenerated forest increased with increasing ground water depth in the direction of parallel to river channel. On the contrary, in the direction perpendicular to river channel, there is a decreasing tendency for the average patch area of the forest and the degenerated forest with increasing ground water depth.     

Assessing the potential of drone‐based thermal infrared imagery for quantifying river temperature heterogeneity

Climate change is altering river temperature regimes, modifying the dynamics of temperature‐sensitive fishes. The ability to map river temperature is therefore important for understanding the impacts of future warming. Thermal infrared (TIR) remote sensing has proven effective for river temperature mapping, but TIR surveys of rivers remain expensive. Recent drone‐based TIR systems present a potential solution to this problem. However, information regarding the utility of these miniaturised systems for surveying rivers is limited. Here, we present the results of several drone‐based TIR surveys conducted with a view to understanding their suitability for characterising river temperature heterogeneity. We find that drone‐based TIR data are able to clearly reveal the location and extent of discrete thermal inputs to rivers, but thermal imagery suffers from temperature drift‐induced bias, which prevents the extraction of accurate temperature data. Statistical analysis of the causes of this drift reveals that drone flight characteristics and environmental conditions at the time of acquisition explain ~66% of the variance in TIR sensor drift. These results shed important light on the factors influencing drone‐based TIR data quality and suggest that further technological development is required to enable the extraction of robust river temperature data. Nonetheless, this technology represents a promising approach for augmenting in situ sensor capabilities and improved quantification of advective inputs to rivers at intermediate spatial scales between point measurements and “conventional” airborne or satellite remote sensing.     

Remote Sensing-Derived Water Extent and Level to Constrain Hydraulic Flood Forecasting Models: Opportunities and Challenges

Accurate, precise and timely forecasts of flood wave arrival time, depth and velocity at each point of the floodplain are essential to reduce damage and save lives. Current computational capabilities support hydraulic models of increasing complexity over extended catchments. Yet a number of sources of uncertainty (e.g., input and boundary conditions, implementation data) may hinder the delivery of accurate predictions. Field gauging data of water levels and discharge have traditionally been used for hydraulic model calibration, validation and real-time constraint. However, the discrete spatial distribution of field data impedes the testing of the model skill at the two-dimensional scale. The increasing availability of spatially distributed remote sensing (RS) observations of flood extent and water level offers the opportunity for a comprehensive analysis of the predictive capability of hydraulic models. The adequate use of the large amount of information offered by RS observations triggers a series of challenging questions on the resolution, accuracy and frequency of acquisition of RS observations; on RS data processing algorithms; and on calibration, validation and data assimilation protocols. This paper presents a review of the availability of RS observations of flood extent and levels, and their use for calibration, validation and real-time constraint of hydraulic flood forecasting models. A number of conclusions and recommendations for future research are drawn with the aim of harmonising the pace of technological developments and their applications.     

Optical remote mapping of rivers at sub‐meter resolutions and watershed extents

At watershed extents, our understanding of river form, process and function is largely based on locally intensive mapping of river reaches, or on spatially extensive but low density data scattered throughout a watershed (e.g. cross sections). The net effect has been to characterize streams as discontinuous systems. Recent advances in optical remote sensing of rivers indicate that it should now be possible to generate accurate and continuous maps of in‐stream habitats, depths, algae, wood, stream power and other features at sub‐meter resolutions across entire watersheds so long as the water is clear and the aerial view is unobstructed. Such maps would transform river science and management by providing improved data, better models and explanation, and enhanced applications. Obstacles to achieving this vision include variations in optics associated with shadows, water clarity, variable substrates and target–sun angle geometry. Logistical obstacles are primarily due to the reliance of existing ground validation procedures on time‐of‐flight field measurements, which are impossible to accomplish at watershed extents, particularly in large and difficult to access river basins. Philosophical issues must also be addressed that relate to the expectations around accuracy assessment, the need for and utility of physically based models to evaluate remote sensing results and the ethics of revealing information about river resources at fine spatial resolutions. Despite these obstacles and issues, catchment extent remote river mapping is now feasible, as is demonstrated by a proof‐of‐concept example for the Nueces River, Texas, and examples of how different image types (radar, lidar, thermal) could be merged with optical imagery. The greatest obstacle to development and implementation of more remote sensing, catchment scale ‘river observatories’ is the absence of broadly based funding initiatives to support collaborative research by multiple investigators in different river settings. Copyright © 2007 John Wiley & Sons, Ltd.     

On the usefulness of remote sensing input data for spatially distributed hydrological modelling: case of the Tarim River basin in China

The ecological situation of the Tarim River basin in China seriously declined since the early 1950s, mainly due to a strong increase in water abstraction for irrigation purposes. To restore the ecological system and support sustainable development of the Tarim River basin region in China, more hydrological studies are demanded to properly understand the processes of the watershed and efficiently manage the water resources. Such studies are, however, complicated due to the limited data availability, especially in the mountainous headwater regions of the Tarim River basin. This study investigated the usefulness of remote sensing (RS) data to overcome that lack of data in the spatially distributed hydrological modelling of the basin. Complementary to the conventional station‐based (SB) data, the RS products that are directly used in this study include precipitation, evapotranspiration and leaf area index. They are derived from raw image data of the Chinese Fengyun meteorological satellite and from the Moderate Resolution Imaging Spectroradiometer (MODIS). The MODIS land surface temperature was used to calculate the atmospheric temperature lapse rate to describe the temperature dependency on topographical variations. Moreover, MODIS‐based snow cover images were used to obtain model initial conditions and as validation reference for the snow model component. Comparison of model results based on RS input versus conventional SB input exhibited similar results in terms of high and low river runoff extremes, cumulative runoff volumes in both runoff and snow melting seasons and spatial and temporal variability of snow cover. During summer time, when the snow cover shrinks in the permanent glacier region, it was found that the model resolution influences the model results dramatically, hence, showing the importance of detailed (RS based) spatially distributed input data. Copyright © 2011 John Wiley & Sons, Ltd.     

卫星应用技术在地学研究中的发展现状

卫星应用技术与地学研究很早就存在密切的联系,从最初单一的遥感应用发展到包含遥感(RS)、地理信息技术(GIS)、全球定位系统(GPS)等多种技术在内的空间信息技术,卫星应用技术促进了地学理论研究和定性、定量的分析验证,加速了地球科学的发展.本文综述了近年来地学领域卫星应用技术的主要方向和现状.     

Impacts of river bed gas on the hydraulic and thermal dynamics of the hyporheic zone

Despite the presence of gas in river beds being a well known phenomenon, its potential feedbacks on the hydraulic and thermal dynamics of the hyporheic zone has not been widely studied. This paper explores hypotheses that the presence of accumulated gas impacts the hydraulic and thermal dynamics of a river bed due to changes in specific storage, hydraulic conductivity, effective porosity, and thermal diffusivity. The hypotheses are tested using data analysis and modelling for a study site on the urban River Tame, Birmingham, UK. Gas, predominantly attributed to microbial denitrification, was observed in the river bed up to around 14% by volume, and to at least 0.8 m depth below river bed. Numerical modelling indicates that, by altering the relative hydraulic conductivity distribution, the gas in the river bed leads to an increase of groundwater discharge from the river banks (relative to river bed) by a factor of approximately 2 during river low flow periods. The increased compressible storage of the gas phase in the river bed leads to an increase in the simulated volume of river water invading the river bed within the centre of the channel during storm events. The exchange volume can be more than 30% greater in comparison to that for water saturated conditions. Furthermore, the presence of gas also reduces the water-filled porosity, and so the possible depth of such invading flows may also increase markedly, by more than a factor of 2 in the observed case. Observed diurnal temperature variations within the gaseous river bed at 0.1 and 0.5 m depth are, respectively, around 1.5 and 6 times larger than those predicted for saturated sediments. Annual temperature fluctuations are seen to be enhanced by around 4 to 20% compared to literature values for saturated sediments. The presence of gas may thus alter the bulk thermal properties to such a degree that the use of heat tracer techniques becomes subject to a much greater degree of uncertainty. Although the likely magnitude of thermal and hydraulic changes due to the presence of gas for this site have been demonstrated, further research is needed into the origins of the gas and its spatial and temporal variability to enable quantification of the significance of these changes for chemical attenuation and hyporheic zone biology.     

A reach‐scale biogeochemical model for temporary rivers

The RS‐tempQ Model ( r each– s cale t emporary flow biogeochemical model) is a conceptual model that can describe the hydrologic, sediment transport and biogeochemical processes of temporary rivers at the reach scale. The model takes into account the expansion–contraction of the inundated area of the river. It simulates the sediment transport and the nutrient fluxes that are transferred to the coastal area due to the first flash flood and during extreme rain events. The RS‐tempQ Model simulates the in‐stream processes during the wet and dry cycles as the river corridor expands and contracts. The model was used to assess and quantify the hydrologic and geochemical processes occurring in a temporary river reach (Krathis River) in Greece. Since the conventional gauging techniques cannot measure the flow in rivers that are split into small braided streams, discharge measurements could not be obtained in order to calibrate and verify the model. Other field measurements such as infiltration losses and sediment accumulation were used for model calibration. Copyright © 2008 John Wiley & Sons, Ltd.     

Plant root and rhizome strength: Are there differences between and within species and rivers?

The strength and architecture of roots and other below-ground organs of riparian and aquatic plants affect plant resistance to uprooting and contribute to reinforcing river bank, bar and bed materials. Therefore, root properties are an important element in models for estimating river bank stability and such models may focus on the role of plants by using root strength–diameter relationships for the particular plant species that are present. Here we explore the degree to which there appear to be significant differences in strength–diameter relationships between and within species-specific data sets obtained for two riparian tree/shrub (Populus nigra, Salix alba) and two emergent aquatic macrophyte (Sparganium erectum, Phalaris arundinacea) species in different European river environments. While the analysed data sets were not specifically collected to answer these research questions, the results are sufficiently compelling to make the case for the collection of a more comprehensive data set and its rigorous analysis. This would allow recommendations to be made on the degree to which (i) species-specific or more general relationships between root/rhizome strength and diameter are appropriate, (ii) such relationships are applicable within and between rivers in different geographical regions and subject to different local environmental conditions, and (iii) further (minimalist) field observations are needed to calibrate such relationships for investigations of new locales or species. © 2018 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd.     

Investigation of meandering in Filyos River by means of satellite sensor data

The most significant morphological property of a river is the meandering process, which is dominated and governed by hydraulic, hydrologic and topographic characteristics of the river and its drainage area. It is possible to obtain reliable data on river morphology in the long term by using remotely sensed data. In this study the Filyos River, located at the Western Black Sea region of Turkey, has been selected as the study area to show the capabilities of remote sensing technology and to define the appropriate techniques for achieving the most reliable information on the river morphology by monitoring the meandering processes. The findings of the study indicate that remotely sensed data can be used successfully in defining some basic characteristics of the meandering process on rivers. Copyright © 2004 John Wiley & Sons, Ltd.     

A Phenomenological Model for Particle Retention in Single,Saturated Fractures

Fractured aquifers are some of the most poorly characterized subsurface environments despite posing one of the highest risks to the protection of potable groundwater. This research was designed to improve the understanding of the factors affecting particle transport through fractures by developing a phenomenological model based on laboratory‐scale transport data. The model presented in this research employed data from over 70 particle tracer tests conducted in single, saturated, variable‐aperture fractures that were obtained from the natural environment and fractured in the laboratory or cast from epoxy in the laboratory. The particles employed were Escherichia coli RS2‐GFP and microspheres. The tracer experiments were conducted in natural (dolomitic limestone and granite) as well as epoxy replicas of the natural fractures. The multiple linear regression analysis revealed that the most important factors influencing particle retention in fractures are the ratio of the ionic strength of solution to collector charge, the ratio of particle to collector charge, and the ratio of advective to diffusive forces as described by the Peclet number. The model was able to reasonably (R² = 0.64) predict the fraction of particles retained; however, it is evident that some factors not accounted for in the model also contributed to retention. This research presents a novel approach to understanding particle transport in fractures, and illustrates the relative importance of various factors affecting the transport mechanisms. The utility of this model lies in the increased understanding of particle transport in fractures, which is extremely useful for directing future research.     

Integration of GRACE mass variations into a global hydrological model

Time-variable gravity data of the GRACE (Gravity Recovery And Climate Experiment) satellite mission provide global information on temporal variations of continental water storage. In this study, we incorporate GRACE data for the first time directly into the tuning process of a global hydrological model to improve simulations of the continental water cycle. For the WaterGAP Global Hydrology Model (WGHM), we adopt a multi-objective calibration framework to constrain model predictions by both measured river discharge and water storage variations from GRACE and illustrate it on the example of three large river basins: Amazon, Mississippi and Congo. The approach leads to improved simulation results with regard to both objectives. In case of monthly total water storage variations we obtained a RMSE reduction of about 25 mm for the Amazon, 6 mm for the Mississippi and 1 mm for the Congo river basin. The results highlight the valuable nature of GRACE data when merged into large-scale hydrological modeling. Furthermore, they reveal the utility of the multi-objective calibration framework for the integration of remote sensing data into hydrological models.     

Improving hydrodynamic modeling of river networks by incorporating data assimilation using a particle filter

Numerical modeling is a well-recognized method for studying the hydrodynamic processes in river networks. Multi-source measurements also offer abundant information on the patterns and mechanisms within the processes. Therefore, improving hydrodynamic modeling of river networks through the use of data assimilation techniques has become a hot research topic in recent years. The particle filter (PF) is a commonly used data assimilation method and has been proven to be applicable to various nonlinear and non-Gaussian models. In the current study, an improved numerical hydrodynamic model for large-scale river networks is established by incorporating the advanced PF algorithm. Furthermore, the PF method based on the Gaussian likelihood function (GLF) and the method based on the Cauchy likelihood function (CLF) are compared for a complex river network scenario. The feasibility of the PF-based methods was evaluated through application to the Yangtze-Dongting River-lake Network (YDRN) by assimilating water stage data collected at six hydrometric stations during the entire hydrodynamic process in 2003. Additionally, the parameters used in the likelihood function, which affect the assimilation performance, also were explored in the current study. The study results found that the accuracy of the model-derived water stage data was improved when the PF-based methods are utilized, with improvement not only at the data assimilation (calibration) sites but also at three hydrometric stations not used in the data assimilation (i.e., verification sites). The highest average Nash-Sutcliffe Efficiency result for the six assimilation sites were 0.98 while the lowest summed root-mean-square-error result was 1.801 m. The comparison results also indicated that the CLF-based PF outperformed the GLF-based PF when high-accuracy observed data are available. Specifically, the CLF can effectively resolve the filtering failure problem and the dispersion problem of PFs, and further improve the accuracy of the filtering results for a river network scenario. In summary, the CLF-based PF method along with high-accuracy observation data shows promise to provide reliable reference and technical support for hydrodynamic modeling of large-scale river networks.     

基于ASTER GDEM的河南西部伏牛山区断裂构造特征解译

采用ASTER GDEM数据及LANDSAT影像,运用GIS地形分析技术,通过山体阴影提取、坡度、坡向、水系走向、夷平面以及三维可视化等分析方法,对河南西部伏牛山地区的断裂构造特征及力学性质进行提取和分析,验证并总结了DEM与遥感、GIS相结合的手段对断裂构造解译的可靠性。     

Incorporating synoptic-scale climate signals for streamflow modelling over the Mediterranean region using machine learning models

ABSTRACT

Understanding streamflow patterns by incorporating climate signal information can contribute remarkably to the knowledge of future local environmental flows. Three machine learning models, the multivariate adaptive regression splines (MARS), the M5 Model Tree and the least squares support vector machine (LSSVM) are established to predict the streamflow pattern over the Mediterranean region of Turkey (Besiri and Baykan stations). The structure of the predictive models is built using synoptic-scale climate signal information and river flow data from antecedent records. The predictive models are evaluated and assessed using quantitative and graphical statistics. The correlation analysis demonstrates that the North Pacific (NP) and the East Central Tropical Pacific Sea Surface Temperature (Niño3.4) indices have a substantial influence on the streamflow patterns, in addition to the historical information obtained from the river flow data. The model results reveal the utility of the LSSVM model over the other models through incorporating climate signal information for modelling streamflow.     

地震应急中制约遥感应用因素分析

遥感在汶川、玉树地震中的有效应用,突出了重大自然灾害对遥感技术的需求。针对地震灾害快速评估的特殊性,详细分析了地震发生在不同的地理环境下,产生不同的地震灾害,地震应急中监测目标的不同,需要采用最佳遥感数据进行灾情判别,从而更好地得到地震灾情的分布情况;对现有遥感数据源和地震应急实际案例进行综合分析,得到遥感在地震应急中的应用现状及制约因素,制约因素主要有两点:时间分辨率和图像处理技术;最后提出了遥感技术在地震应急应用中存在的一些问题和解决的方法。