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.     

A minimalistic approach for evapotranspiration estimation using the Prophet model

ABSTRACT

This study aimed to evaluate the potential of the recently introduced Prophet model for estimating reference evapotranspiration (ETo). A comparative study was conducted for benchmarking the model results with support vector regression (SVR) and temperature-based empirical models (Thornthwaite and Hargreaves) in southern Japan. The performance of the Prophet, SVR and temperature-based empirical models was evaluated by Nash–Sutcliffe efficiency (NSE) and coefficient of determination (R²). The results indicate that temperature-based Prophet and SVR models have greater accuracy than the empirical models. The Prophet model with sole input of relative humidity, sunshine hours or windspeed showed acceptable accuracy (NSE > 0.80; R² > 0.80), while SVR models with similar inputs showed greater errors. Accuracy improved with increasing number of input parameters, giving excellent performance (NSE > 0.95; R² > 0.95) with all input parameters. Hence, the Prophet model is a new promising approach for modelling ETo with limited input variables.     

Formation Of Coronal Mhd Shock Waves – I. The Basic Mechanism

The formation and evolution of a large amplitude MHD perturbation propagating perpendicular to the magnetic field in a perfectly conducting low plasma is studied. The perturbation is generated by an abrupt expansion of the source region. Explicit expressions for the time and the distance needed for the transformation of the perturbation's leading edge into a shock wave are derived. The results are applied to coronal conditions and the dynamic spectra of the radio emission excited by the shock are synthesized, reproducing metric and kilometric type II bursts. The features corresponding to the metric type II burst precursor and the moving type IV burst in the case of kilometric type II bursts are identified. A specific radio signature that is sometimes observed at the onset of a metric type II burst is found to appear immediately before the shock wave formation due to the associated growth of the magnetic field gradient. Time delays and starting frequencies of bursts' onsets are calculated and presented as a function of the impulsiveness of the source-region expansion, using different values of the ambient Alfvén velocity and various time profiles of the expansion velocity. The results are confronted with the observations of metric and kilometric type II solar radio bursts.     

Seeding agent dispersion within convective cloud as simulated by a 3-D numerical model

Summary The three-dimensional mesoscale cloud-resolving model ARPS (Advanced Regional Prediction System) was used to investigate the dispersal of an inert seeding agent within a cumulonimbus (Cb) cloud developing from two different initial states. In this paper, we stress the influence vortices in the cloud have on seeding agent dispersion. If a strong directional ambient wind shear is present in the lowest layer, a vortex pair formed at the flanks of the simulated cloud. Following the velocity field, a considerable amount of the injected seeding agent would be thrown out to the rear of the cloud, where both updrafts associated with vortices and downdrafts occurred. After a short time the agent was present only in the cloud periphery. If the Cb cloud developed under conditions where directional ambient wind shear did not exist, seeding agent dispersion would be quite different. In this case, almost all the seeding agent was transported into the main updraft, while the residence time of the agent within the cloud was longer due to the weaker cloud dynamics. Therefore, we must pay attention to whether or not the cloud contains vortices when we make the decision where to seed. This is necessary in order to minimize the loss of seeding material.     

Comparison of three mathematical models for the estimation of 10-day drain discharge

Summary This paper presents an the evaluation of three mathematical models for the estimation of 10-day drain discharge using weather, pedological and phenological parameters. One model uses modified Palmer’s procedure for the calculation of water balance components and the second two use empirical multiple linear regression. The models have been applied to data of an amelioration experimental field in central Croatia (south-western part of the Pannonian Lowland), where a set of average 10-day air temperature and relative humidity data, average wind speed, cumulative precipitation and drain discharge amounts, insolation and soil moisture data and vegetation indices have been collected. The experiment has been performed during the period 1990–1992. The models have been tested by comparing estimated and observed 10-day drain discharge amounts. All models mainly give acceptable cumulative results although their accuracy depends on the season.     

Spectral analysis of the “Koshava” wind

Summary “Koshava” is a gusty wind of changeable intensity, blowing from a south-easterly direction, over Serbia, Romania and Bulgaria. It is caused by the interaction between the synoptic circulation and the orography of the Carpathian and the Balkan mountains. This paper analyzes wind data measured at the Belgrade-Observatory during the longest period of consecutive days of “Koshava” which occurred from 14 January to 13 February 1972. Mean hourly wind speed data has been examined using spectral analysis. The power spectra are calculated using autocorrelation spectral analysis, the multi-taper method and wavelet transform. The maximum of which is about 122 h (5 days) corresponds to the time span of synoptic processes.     

A Photometric Investigation of the Eclipsing System OO Aql

The paper is concerning with the light curve analysis of an active binary system of OO Aql. Basic orbital and physical parameters of this system were estimated by analysing photometric observations by Lafta and Grainger (1985) and Essam et al. (1992). Both groups of observations gave quite different light curves with a conspicuous asymmetry arising from unequal height of successive maxima. A slightly changed programme by Djuraevi (1992a) was applied. The programme was based on the Roche model and it included an option of overcontact configuration and active (bright or dark) spotted regions on the system's components. An inverse-problem method by Djuraevi (1992b) was used to estimate the parameters of the system and active regions. The light-curve analysis suggested a significant change in the system. For Lafta and Grainger's light curves our analysis showed the presence of a dark spotted area near the polar region of the primary component. This dark area covered about 22% of the primary's hemisphere. Essam et al.'s light curves had very complex shape. In this case the mechanism of the mass transfer and the exchange of the thermal energy between the components could produce a bright-spot (bs) area on the secondary, near the Lagrange equilibrium point L₁ around the neck region between the stars. To explain a sharp increase of the system's brightness in the B filter immediately after the secondary minimum, we had to introduce also a hot-spot (hs) area of a high temperature contrast on the secondary. With the increase of the brightness in the V-filter being of much lower intensity, this hot spot region could be taken as a local eruption. The behaviour of the B–V colour index with the orbital phase is in favour of this hypothesis also. The obtained results show that the light-curve changes are due to development and moving of active spot regions on the components.     

Age of amphibolites associated with alpine peridotites in the Dinaride ophiolite zone,Yugoslavia

Amphibolites associated with alpine peridotites in the Central Ophiolite zone in Yugoslavia have K-Ar ages of 160–170 m.y. These amphibolites and associated peridotites underwent deep-seated metamorphism prior to tectonic emplacement into the sedimentary-volcanic assemblage of the Dinarides. The alpine peridotites and associated local rocks of the ophiolite suite are interpreted as Jurassic oceanic crust and upper mantle.     

On forecasting of cyclogenesis in the West Mediterranean and other areas bounded by mountain ranges by baroclinic model

Summary Changes occurring in the thermal structure of air masses in the Alps area and the West Mediterranean region, during cyclonic developments in the Gulf of Genoa, have been considered. The effect of the height and the form of the mountain ranges as well as that of the nonadiabatic processes, on the deformation of the thickness pattern, have been discussed. These factors have been further taken into consideration for a modification of the equations used inEstoque's method for the baroclinic model, and a new method has been developed for forecasting cyclogenesis in regions bounded by mountain ranges.

---

Zusasmmenfassung Es werden die Änderungen der thermischen Struktur der Luftmassen im Alpengebiet und im westlichen Mittelmeerraum bei der Entwicklung von Zyklonen im Golf von Genua aufgezeigt und die Einflüsse von Höhe und Form der Gebirgszüge sowie von nichtadiabatischen Vorgängen auf die Deformation der relativen Topographie besprochen. Diese Faktoren werden auch zur Modifikation der in der Methode vonEstoque für ein baroklines Modell verwendeten Gleichungen in Betracht gezogen und damit wird eine neue Methode zur Vorhersage der Zyklogenese unter dem Einfluß von Gebirgszügen entwickelt.

---

Résumé On expose dans ce mémoire à quelles modifications la structure thermique des masses d'air est soumise dans la région des Alpes et dans le bassin occidental de la Méditerranée lors du développement de dépressions sur le Golfe de Gênes. On y discute en outre l'influence de la hauteur et de la forme des chaînes de montagnes ainsi que celle des processus non-adiabatiques sur des déformations de la topographie relative. Tous ces éléments sont pris en considération pour leur adapter les équations utilisées dans le modèle baroclined'Estoque. Par là, on accède à une nouvelle méthode de prévision de la cyclogénèse, méthode qui tient compte de l'influence des chaînes de montagnes.

---

---

With 5 Figures     

Zur Revision der GattungGloeocapsa Kützing (Cyanophyta)

Ohne Zusammenfassung Die vorliegende Arbeit entstand dank der Unterstützung der Alexander-von-Humboldt-Stiftung, Bad Godesberg.