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.     

Mobilization of arsenic and iron from Red River floodplain sediments, Vietnam

Sediments from the Red River and from an adjacent floodplain aquifer were investigated with respect to the speciation of Fe and As in the solid phase, to trace the diagenetic changes in the river sediment upon burial into young aquifers, and the related mechanisms of arsenic release to the groundwater. Goethite with subordinate amounts of hematite were, using Mössbauer spectroscopy, identified as the iron oxide minerals present in both types of sediment. The release kinetics of Fe, As, Mn and PO₄ from the sediment were investigated in leaching experiments with HCl and 10 mM ascorbic acid, both at pH 3. From the river sediments, most of the Fe and As was mobilized by reductive dissolution with ascorbic acid while HCl released very little Fe and As. This suggests As to be associated with an Fe-oxide phase. For oxidized aquifer sediment most Fe was mobilized by ascorbic acid but here not much As was released. However, the reduced aquifer sediments contained a large pool of Fe(II) and As that is readily leached by HCl, probably derived from an unidentified authigenic Fe(II)-containing mineral which incorporates As as well. Extraction with ascorbic acid indicates that the river sediments contain both As(V) and As(III), while the reduced aquifer sediment almost exclusively releases As(III). The difference in the amount of Fe(II) leached from river and oxidized aquifer sediments by ascorbic acid and HCl, was attributed to reductive dissolution of Fe(III). The reactivity of this pool of Fe(III) was quantified by a rate law and compared to that of synthetic iron oxides. In the river mud, Fe(III) had a reactivity close to that of ferrihydrite, while the river sand and oxidized aquifer sediment exhibited a reactivity ranging from lepidocrocite or poorly crystalline goethite to hematite. Mineralogy by itself appears to be a poor predictor of the iron oxide reactivity in natural samples using the reactivity of synthetic Fe-oxides as a reference. Sediments were incubated, both unamended and with acetate added, and monitored for up to 2 months. The river mud showed the fastest release of both Fe and As, while the effect of acetate addition was minor. This suggests that the presence of reactive organic carbon is not rate limiting. In the case of the river and aquifer sediments, the release of Fe and As was always stimulated by acetate addition and here reactive organic carbon was clearly the rate limiting factor. The reduced aquifer sediment apparently can sustain slower but prolonged microbially-driven release of As. The highly reactive pools of Fe(III) and As in the river mud could be due to reoxidation of As and Fe contained in the reducing groundwater from the floodplain aquifers that are discharging into the river. Deposition of the suspended mud on the floodplain during high river stages is proposed to be a major flux of As onto the floodplain and into the underlying aquifers.     

Land tenure and REDD+: The good,the bad and the ugly

A number of international donors, national governments and project proponents have begun to lay the groundwork for REDD+, but tenure insecurity – including the potential risks of land grabbing by outsiders and loss of local user rights to forests and forest land – is one of the main reasons that many indigenous and other local peoples have publicly opposed it. Under what conditions is REDD+ a threat to local rights, and under what conditions does it present an opportunity? This article explores these issues based on available data from a global comparative study on REDD+, led by the Center for International Forestry Research, which is studying national policies and processes in 12 countries and 23 REDD+ projects in 6 countries. The article analyses how tenure concerns are being addressed at both national and project level in emerging REDD+ programs. The findings suggest that in most cases REDD+ has clearly provided some new opportunities for securing local tenure rights, but that piecemeal interventions by project proponents at the local level are insufficient in the absence of broader, national programs for land tenure reform. The potential for substantial changes in the status quo appear unlikely, though Brazil – the only one with such a national land tenure reform program – offers useful insights. Land tenure reform – the recognition of customary rights in particular – and a serious commitment to REDD+ both challenge the deep-rooted economic and political interests of ‘business as usual’.     

Sustainable groundwater development in the coastal Tra Vinh province in Vietnam under saltwater intrusion and climate change

Three-dimensional transient groundwater flow and saltwater transport models were constructed to assess the impacts of groundwater abstraction and climate change on the coastal aquifer of Tra Vinh province (Vietnam). The groundwater flow model was calibrated with groundwater levels (2007–2016) measured in 13 observation wells. The saltwater transport model was compared with the spatial distribution of total dissolved solids. Model performance was evaluated by comparing observed and simulated groundwater levels. The projected rainfalls from two climate models (MIROC5 and CRISO Mk3.6) were subsequently used to simulate possible effects of climate changes. The simulation revealed that groundwater is currently depleted due to overabstraction. Towards the future, groundwater storage will continue to be depleted with the current abstraction regime, further worsening in the north due to saltwater intrusion from inland trapped saltwater and on the coast due to seawater intrusion. Notwithstanding, the impact from climate change may be limited, with the computed groundwater recharge from the two climate models revealing no significant change from 2017 to 2066. Three feasible mitigation scenarios were analyzed: (1) reduced groundwater abstraction by 25, 35 and 50%, (2) increased groundwater recharge by 1.5 and 2 times in the sand dunes through managed aquifer recharge (reduced abstraction will stop groundwater-level decline, while increased recharge will restore depleted storage), and (3) combining 50% abstraction reduction and 1.5 times recharge increase in sand dune areas. The results show that combined interventions of reducing abstraction and increasing recharge are necessary for sustainable groundwater resources development in Tra Vinh province.

     

Lang Trang Cave (Northern Vietnam): Pleistocene Mammalian Locality with Gigantopithecus Finds

Doklady Earth Sciences - Speleological, geological and paleontological characteristics of the Lang Trang cave in northern Vietnam are presented. Primates Gigantopithecus blacki von Koenigswald,...     

Flood vulnerability and buildings’ flood exposure assessment in a densely urbanised city: comparative analysis of three scenarios using a neural network approach

Natural Hazards - Advances in the availability of multi-sensor, remote sensing-derived datasets, and machine learning algorithms can now provide an unprecedented possibility to predict flood events...     

越南西北部Phan Si Pan地区Ngoi Chi岩组的锆石U-Pb年龄和Hf同位素组成

本文对越南西北部Phan Si Pan地区变质地体中的一个Ngoi Chi片麻岩进行了锆石CL内部结构分析、LA-(MC)-ICP-MS锆石U-Pb定年和Hf同位素分析。CL图象和Th/U比值特征显示该片麻岩样品中的锆石主要为岩浆锆石,有少量窄的变质边。岩浆锆石的年龄为~2.9 Ga,表明该样品是越南西北部Phan Si Pan地区的基底岩石。它们的εHf(t)值为–4.70±0.92,二阶段Hf模式年龄为~3.5 Ga,表明其为更古老的(3.5 Ga)冥太古代地壳物质部分熔融作用形成。变质边部锆石给出了~1.8 Ga的年龄,表明变质作用发生在古元古代早期,Phan Si Pan地区在这一时期可能经历了一次重要的构造热事件。     

Torrential rainfall-triggered shallow landslide characteristics and susceptibility assessment using ensemble data-driven models in the Dongjiang Reservoir Watershed,China

Natural Hazards - This study investigated the characteristics of rainfall-triggered landslides during the Typhoon Bilis in the Dongjiang Reservoir Watershed, China. The comparative shallow...     

Ionosphere probing with simultaneous GNSS radio occultations

GPS Solutions - Radio occultation (RO) is a powerful technique for providing vertical profiles of refractivity, temperature, pressure, and water vapor of the neutral atmosphere and electron density...     

Seismic ground-roll noise attenuation using deep learning

We propose to adopt a deep learning based framework using generative adversarial networks for ground-roll attenuation in land seismic data. Accounting for the non-stationary properties of seismic data and the associated ground-roll noise, we create training labels using local time–frequency transform and regularized non-stationary regression. The basic idea is to train the network using a few shot gathers such that the network can learn the weights associated with noise attenuation for the training shot gathers. We then apply the learned weights to test ground-roll attenuation on shot gathers, that are not a part of training input to obtain the desired signal. This approach gives results similar to local time–frequency transform and regularized non-stationary regression but at a significantly reduced computational cost. The proposed approach automates the ground-roll attenuation process without requiring any manual input in picking the parameters for each shot gather other than in the training data. Tests on field-data examples verify the effectiveness of the proposed approach.