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.     

Adaptive neuro-fuzzy inference system coupled with shuffled frog leaping algorithm for predicting river streamflow time series

ABSTRACT

Accurate runoff forecasting plays a key role in catchment water management and water resources system planning. To improve the prediction accuracy, one needs to strive to develop a reliable and accurate forecasting model for streamflow. In this study, the novel combination of the adaptive neuro-fuzzy inference system (ANFIS) model with the shuffled frog-leaping algorithm (SFLA) is proposed. Historical streamflow data of two different rivers were collected to examine the performance of the proposed model. To evaluate the performance of the proposed ANFIS-SFLA model, six different scenarios for the model input–output architecture were investigated. The results show that the proposed ANFIS-SFLA model (R² = 0.88; NS = 0.88; RMSE = 142.30 (m³/s); MAE = 88.94 (m³/s); MAPE = 35.19%) significantly improved the forecasting accuracy and outperformed the classic ANFIS model (R² = 0.83; NS = 0.83; RMSE = 167.81; MAE = 115.83 (m³/s); MAPE = 45.97%). The proposed model could be generalized and applied in different rivers worldwide.     

Synchronous Transformations of Mineral and Organic Constituents of Sedimentary Rocks in Geological Structure with an Initial Extension and Subsequent Compression Tectonic Regime

Sedimentary rocks of the section in the Red River fold zone of northern Viet Nam are considered. It is shown that secondary mineral parageneses formed in two stages. The first stage (35–17 Ma ago) corresponded to the period of structure extension and sediment subsidence to a depth of about 6 km. This period and subsequent 10 Ma were marked by the formation of a usual dia- and catagenetic zoning of metasedimentary rocks. The second stage (5–7 Ma ago) corresponded to processes of compression that were responsible for the deformation of rocks into gentle folds and 1.5 to 2.2 times contraction of the section thickness in different places. The sequential–mineralogical zoning was disturbed at this stage. Smectites and mixed-layer minerals were replaced by chlorites and hydromicas. Organic material also responded to compression simultaneously with inorganic components. The bituminous component was released from humic matter and rocks became enriched in hydrocarbons.     

Improving Indigenous Technologies for Sustainable Land Use in Northern Mountainous Areas of Vietnam

More than 30 ethnic groups are now living in northern mountainous regions, Vietnam, mainly relying on shifting cultivation with the fallow period being shortened from time to time. Naturally, soil fertility reduces from cycle to cycle, entailing the reduction of productivity. Large areas of moderately sloping lands suitable for upland agriculture have become bare after many cultivation-fallow cycles. The soils there have been severely degraded with more toxicity, low porosity, low water retention capacity and poor floral diversity. Normally, these lands cannot be used for food crop cultivation. So farmers in uplands have to rely on slash-and-burn practices for their livelihood. As there is no more forest with good soil in medium slopes, farmers go to cut forests in watershed, high slope lands and old forests up to the mountains‘ top. There are ecologically and environmentally very sensitive areas, so their destruction will inevitably cause hazardous consequences in the whole basin. Meanwhile,cultivation in these areas has low economic efficiency and sustainability because the crop yield may decrease very fast due to severe erosion as the higher the slope, the more serious erosion. Consequently living standards of highland farmers remain low and unstable. Sustainable farming on these lands in the perspective of a seriously deteriorated ecology and environmental is not an easy task. There have been many projects trying to help mountainous farmers get out of their vicious circle. However, due to different reasons, the results gained are low, and in some cases,things ceased to move after the projects phased out. During past few years, based on the farmer experiences, the Vietnam Agricultural Science Institute has cooperated with local and international partners to implement different projects in order to solve the problems by developing simple, easy and cheap cultivation technologies, which can be accepted and applied by local poor farmers for sustainable agricultural production. The first results of our activities offered good opportunities for sustain food production, improve soil health, recharge of aquifers,and enhanced household income for better rural lively hoods in the upland eco-regions of northern Vietnam.     

Non‐local damage modelling of concrete: a procedure for the determination of model parameters

This paper presents a procedure for the determination of parameters of non‐local damage models. This is to assure a consistent response of a non‐local damage model, as choice of the internal length and other parameters of the model are varied. Correlations between the internal length and other parameters governing the local constitutive behaviour of the model are addressed and exploited. Focus is put on the relationship between the internal length of the non‐local model and the width of the fracture process zone. Numerical examples are used to demonstrate the rigour of the proposed method. Copyright © 2006 John Wiley & Sons, Ltd.     

Spread-F during the magnetic storm of 22 January 2004 at low latitudes: Effect of IMF-Bz in relation to local sunset time

The paper describes the results of spread-F at low latitude stations around the world during the magnetic storm starting at 0130 UT on 22 January 2004. The storm can be divided into two phases, first phase up to 1000 UT when interplanetary magnetic field IMF-Bz was highly fluctuating around a small positive value and the second phase after a sudden large southward turning of IMF-Bz at 1030 UT. The first phase produced strong spread-F at Jicamarca, Sao Luis, and Ascension Island and caused complete inhibition of spread-F at Thumba and Waltair in India. It generated weak spread-F at Ho Chi Minh City in Vietnam and strong spread-F at Hainan and Chung Li. The strong spread-F at Hainan and Chung Li were caused by the positive IMF-Bz during the first phase of the storm and not by the negative pulse of IMF-Bz at 1000 UT.     

New Methodology to Characterize Shear Behavior of Joints by Combination of Direct Shear Box Testing and Numerical Simulations

A new procedure is presented, which combines big shear box tests on rocks and corresponding numerical simulations with explicit consideration of joint roughness to get deeper insight into the shear behavior of rock joints. The procedure consists of three parts: (1) constant normal load- or CNS-shear box tests with registration of shear- and normal-components of stress and displacements and deduction of basis rock mechanical parameters; (2) high resolution 3D-scanning of joint surface to deduce joint topography; and (3) set-up, run and evaluation of 3-dimensional numerical model with explicit duplication of joint roughness as back-analysis of shear box tests. The numerical back-analysis provides deeper insight into the joint behavior at the micro-scale. Several parameters can be deduced, like micro-slope angle distribution, aperture size distribution, local normal stress distribution and detailed analysis of dilation in relation to shear direction. The potential of the new procedure is illustrated exemplary by shear box tests on slate.     

Statistical data fusion of multi-sensor AOD over the Continental United States

This article illustrates two techniques for merging daily aerosol optical depth (AOD) measurements from satellite and ground-based data sources to achieve optimal data quality and spatial coverage. The first technique is a traditional Universal Kriging (UK) approach employed to predict AOD from multi-sensor aerosol products that are aggregated on a reference grid with AERONET as ground truth. The second technique is spatial statistical data fusion (SSDF); a method designed for massive satellite data interpolation. Traditional kriging has computational complexity O(N³), making it impractical for large datasets. Our version of UK accommodates massive data inputs by performing kriging locally, while SSDF accommodates massive data inputs by modelling their covariance structure with a low-rank linear model. In this study, we use aerosol data products from two satellite instruments: the moderate resolution imaging spectrometer and the geostationary operational environmental satellite, covering the Continental United States.