Composition,temperature, and thickness of the lithosphere of the Archean Kaapvaal craton

A new model is proposed for the structure of the Kaapvaal craton lithosphere. Based on chemical thermodynamics methods, profiles of the chemical composition, temperature, density, and S wave velocities are constructed for depths of 100–300 km. A solid-state zone of lower velocities is discovered on the S velocity profile in the depth interval 150–260 km. The temperature profiles are obtained from absolute values of P and S velocities, taking into account phase transformations, anharmonicity, and anelastic effects. The examination of the sensitivity of seismic models to the chemical composition showed that relatively small variations in the composition of South African xenoliths result in lateral temperature variations of ～200°C. Inversion of some seismic profiles (including IASP91) with a fixed bulk composition of garnet peridotites (the primitive mantle material) leads to a temperature inversion at depths of 200–250 km, which is physically meaningless. It is supposed that the temperature inversion can be removed by gradual fertilization of the mantle with depth. In this case, the craton lithosphere should be stratified in chemical composition. The depleted lithosphere composed by garnet peridotites exists to depths of 175–200 km. The lithospheric material at depths of 200–250 km is enriched in basaltoid components (FeO, Al₂O₃, and CaO) as compared with the material of garnet peridotites but is depleted in the same components as compared with the fertile substance of the underlying primitive mantle. The material composing the craton root at a depth of ～275 km does not differ in its physical and chemical characteristics from the composition of the normal mantle, and this allows one to estimate the thickness of the lithosphere at 275 km. The results of this work are compared with data of seismology, thermal investigations, and thermobarometry.     

Homologous temperature of olivine: Implications for creep of the upper mantle and fabric transitions in olivine

The homologues temperature of a crystalline material is defined as T/T_m, where T is temperature and T_m is the melting(solidus) temperature in Kelvin. It has been widely used to compare the creep strength of crystalline materials. The melting temperature of olivine system,(Mg,Fe)_2SiO_4, decreases with increasing iron content and water content, and increases with confining pressure. At high pressure, phase transition will lead to a sharp change in the melting curve of olivine. After calibrating previous melting experiments on fayalite(Fe_2SiO_4), the triple point of fayalite-Fe_2SiO_4 spinel-liquid is determined to be at 6.4 GPa and 1793 K. Using the generalized means, the solidus and liquidus of dry olivine are described as a function of iron content and pressure up to 6.4 GPa. The change of T/T_m of olivine with depth allows us to compare the strength of the upper mantle with different thermal states and olivine composition. The transition from semi-brittle to ductile deformation in the upper mantle occurs at a depth where T/T_m of olivine equals 0.5. The lithospheric mantle beneath cratons shows much smaller T/T_m of olivine than orogens and extensional basins until the lithosphere-asthenosphere boundary where T/T_m 0.66, suggesting a stronger lithosphere beneath cratons. In addition, T/T_m is used to analyze deformation experiments on olivine. The results indicate that the effect of water on fabric transitions in olivine is closely related with pressure. The hydrogen-weakening effect and its relationship with T/T_m of olivine need further investigation. Below 6.4 GPa(200 km), T/T_m of olivine controls the transition of dislocation glide from [100] slip to [001] slip. Under the strain rate of 10~(-12)–10~(-15) s~(-1) and low stress in the upper mantle, the [100](010) slip system(A-type fabric) becomes dominant when T/T_m 0.55–0.60. When T/T_m 0.55–0.60, [001] slip is easier and low T/T_m favors the operation of [001](100) slip system(C-type fabric). This is consistent with the widely observed A-type olivine fabric in naturally deformed peridotites, and the C-type olivine fabric in peridotites that experienced deep subduction in ultrahigh-pressure metamorphic terranes. However, the B-type fabric will develop under high stress and relatively low T/T_m. Therefore, the homologues temperature of olivine established a bridge to extrapolate deformation experiments to rheology of the upper mantle. Seismic anisotropy of the upper mantle beneath cratons should be simulated using a four-layer model with the relic A-type fabric in the upper lithospheric mantle, the B-type fabric in the middle layer, the newly formed A- or B-type fabric near the lithosphere-asthenosphere boundary, and the asthenosphere dominated by diffusion creep below the Lehmann discontinuity. Knowledge about transition mechanisms of olivine fabrics is critical for tracing the water distribution and mantle flow from seismic anisotropy.     

Creepex of Kamchatka shallow earthquakes

Results of investigation of the lithosphere in the Kamchatka seismic focal zone from dynamic characteristics of earthquake records obtained at regional stations are presented. It is assumed that the specificity of the source zone can be estimated by the relation Cr = K _P ? bK _S ? c characterizing relative energies (energy classes, according to [Fedotov, 1972]) of short period transverse and longitudinal waves in the source. Azimuthal, spatial, and temporal variations in Cr and their relation to focal mechanisms are examined. Spatiotemporal variations in this parameter are shown to be caused by the influence of variations in the conditions in the source zone (its substance or process) on the radiation of P and S waves.     

Crustal and upper mantle velocity model along the DOBRE-4 profile from North Dobruja to the central region of the Ukrainian Shield: 1. seismic data

For studying the structure of the lithosphere in southern Ukraine, wide-angle seismic studies that recorded the reflected and refracted waves were carried out under the DOBRE-4 project. The field works were conducted in October 2009. Thirteen chemical shot points spaced 35–50 km apart from each other were implemented with a charge weight varying from 600 to 1000 kg. Overall 230 recording stations with an interval of 2.5 km between them were used. The high quality of the obtained data allowed us to model the velocity section along the profile for P- and S-waves. Seismic modeling was carried out by two methods. Initially, trial-and-error ray tracing using the arrival times of the main reflected and refracted P- and S-phases was conducted. Next, the amplitudes of the recorded phases were analyzed by the finite-difference full waveform method. The resulting velocity model demonstrates a fairly homogeneous structure from the middle to lower crust both in the vertical and horizontal directions. A drastically different situation is observed in the upper crust, where the V _p velocities decrease upwards along the section from 6.35 km/s at a depth of 15–20 km to 5.9–5.8 km/s on the surface of the crystalline basement; in the Neoproterozoic and Paleozoic deposits, it diminishes from 5.15 to 3.80 km/s, and in the Mesozoic layers, it decreases from 2.70 to 2.30 km/s. The subcrustal V _p gradually increases downwards from 6.50 to 6.7–6.8 km/s at the crustal base, which complicates the problem of separating the middle and lower crust. The V _p velocities above 6.80 km/s have not been revealed even in the lowermost part of the crust, in contrast to the similar profiles in the East European Platform. The Moho is clearly delineated by the velocity contrast of 1.3–1.7 km/s. The alternating pattern of the changes in the Moho depths corresponding to Moho undulations with a wavelength of about 150 km and the amplitude reaching 8 to 17 km is a peculiarity of the velocity model.     

Stepwise joint inversion of surface wave dispersion,Rayleigh wave ZH ratio,and receiver function data for 1D crustal shear wave velocity structure

Accurate determination of seismic velocity of the crust is important for understanding regional tectonics and crustal evolution of the Earth. We propose a stepwise joint linearized inversion method using surface wave dispersion, Rayleigh wave ZH ratio (i.e., ellipticity), and receiver function data to better resolve 1D crustal shear wave velocity (v _S) structure. Surface wave dispersion and Rayleigh wave ZH ratio data are more sensitive to absolute variations of shear wave speed at depths, but their sensitivity kernels to shear wave speeds are different and complimentary. However, receiver function data are more sensitive to sharp velocity contrast (e.g., due to the existence of crustal interfaces) and v _P/v _S ratios. The stepwise inversion method takes advantages of the complementary sensitivities of each dataset to better constrain the v _S model in the crust. We firstly invert surface wave dispersion and ZH ratio data to obtain a 1D smooth absolute v _S model and then incorporate receiver function data in the joint inversion to obtain a finer v _S model with better constraints on interface structures. Through synthetic tests, Monte Carlo error analyses, and application to real data, we demonstrate that the proposed joint inversion method can resolve robust crustal v _S structures and with little initial model dependency.     

Crustal stress field in Yunnan: implication for crust-mantle coupling

We applied the g CAP algorithm to determine239 focal mechanism solutions 3:0 M We 6:0T with records of dense Chin Array stations deployed in Yunnan,and then inverted 686 focal mechanisms(including 447 previous results) for the regional crustal stress field with a damped linear inversion. The results indicate dominantly strike-slip environment in Yunnan as both the maximum(r1) and minimum(r3) principal stress axes are sub-horizontal. We further calculated the horizontal stress orientations(i.e., maximum and minimum horizontal compressive stress axes: S H and S h, respectively) accordingly and found an abrupt change near *26°N. To the north, S H aligns NW-SE to nearly E-W while S h aligns nearly N-S. In contrast, to the south, both S H and S h rotate laterally and show dominantly fan-shaped patterns. The minimum horizontal stress(i.e., maximum strain axis) S h rotates from NW-SE to the west of Tengchong volcano gradually to nearly E-W in west Yunnan, and further toNE-SW in the South China block in the east. The crustal strain field is consistent with the upper mantle strain field indicated by shear-wave splitting observations in Yunnan but not in other regions. Therefore, the crust and upper mantle in Yunnan are coupled and suffering vertically coherent pure-shear deformation in the lithosphere.     

Heat transport variation due to change of North Pacific subtropical gyre interior flow during 1993–2012

Applying segment-wise altimetry-based gravest empirical mode method to expendable bathythermograph temperature, Argo salinity, and altimetric sea surface height data in March, June, and November from San Francisco to near Japan (30^° N, 145^° E) via Honolulu, we estimated the component of the heat transport variation caused by change in the southward interior geostrophic flow of the North Pacific subtropical gyre in the top 700 m layer during 1993–2012. The volume transport-weighted temperature (T_I) is strongly dependent on the season. The anomaly of T_I from the mean seasonal variation, whose standard deviation is 0.14^°C, was revealed to be caused mainly by change in the volume transport in a potential density layer of 25.0?25.5σ_??. The anomaly of T_I was observed to vary on a decadal or shorter, i.e., quasi-decadal (QD), timescale. The QD-scale variation of T_I had peaks in 1998 and 2007, equivalent to the reduction in the net heat transport by 6 and 10 TW, respectively, approximately 1 year before those of sea surface temperature (SST) in the warm pool region, east of the Philippines. This suggests that variation in T_I affects the warm pool SST through modification of the heat balance owing to the entrainment of southward transported water into the mixed layer.     

Testing the reference Moon model in respect of the thermal regime and chemical composition of the mantle: Thermodynamics versus seismology

The VPREMOON seismic reference Moon model (Garcia et al., 2011) has been tested with respect to the thermal regime and chemical composition of the mantle. Based on a self-consistent thermodynamic approach and petrological models of the lunar mantle covering a wide range of concentrations of CaO, Al₂O₃, and FeO, we convert the P- and S-wave velocity profiles to the temperature–depth profiles. The solution procedure relies on the method of the Gibbs free energy minimization and the equations of state for the mantle material which take into account the effects of phase transformations, anharmonicity, and anelasticity. We find that regardless of the chemical composition, the positive P- and S-wave velocity gradient in the lunar mantle leads to a negative temperature gradient, which has no physical basis. For adequate mantle temperatures, the P- and S-wave velocities should remain almost constant or slightly decrease with depth (especially V_S) as a result of the effects of the temperature, which grows faster than pressure. These findings underscore the importance of the relationship of the thermodynamics and physics of minerals with seismology.     

Seismicity anomalies before the great earthquake of <Emphasis Type="Italic">M</Emphasis><Subscript>S</Subscript>=8.1 in the Kunlun Pass and its significance to earthquake prediction

A great earthquake of M _S=8.1 took place in the west of Kunlun Pass on November 14, 2001. The epicenter is located at 36.2°N and 90.9°E. The analysis shows that some main precursory seismic patterns appear before the great earthquake, e.g., seismic gap, seismic band, increased activity, seismicity quiet and swarm activity. The evolution of the seismic patterns before the earthquake of M _S=8.1 exhibits a course very similar to that found for earthquake cases with M _S≥7. The difference is that anomalous seismicity before the earthquake of M _S=8.1 involves in the larger area coverage and higher seismic magnitude. This provides an evidence for recognizing precursor and forecasting of very large earthquake. Finally, we review the rough prediction of the great earthquake and discuss some problems related to the prediction of great earthquakes.     

Geothermal data analysis at the high-temperature hydrothermal area in Western Sichuan

The western Sichuan hydrothermal area is located at the northeastern margin of the eastern syntaxis of the Qinghai-Tibet Plateau, which is also the eastern end of the Mediterranean-Himalayan geothermal activity zone. There are 248 warm or hot springs in this area, and 11 have temperatures beyond the local boiling temperature. Most of these hot springs are distributed along the Jinshajiang, Dege-Xiangcheng, Ganzi-Litang, and Xianshuihe faults, forming a NW-SE hydrothermal belt. A geothermal analysis of this high-temperature hydrothermal area is an important basis for understanding the deep geodynamic process of the eastern syntaxis of the Qinghai-Tibet Plateau. In addition, this study offers an a priori view to utilize geothermal resources, which is important in both scientific research and application. We use gravity, magnetic, seismic, and helium isotope data to analyze the crust-mantle heat flow ratio and deep geothermal structure. The results show that the background terrestrial heat flow descends from southwest to northeast. The crustal heat ratio is not more than 60%. The high temperature hydrothermal active is related to crustal dynamics processes. Along the Batang-Litang-Kangding line, the Moho depth increases eastward, which is consistent with the changing Qc/Qm(crustal/mantle heat flow) ratio trend. The geoid in the hydrothermal zone is 4–6 km higher than the surroundings, forming a local "platform". The NW-SE striking local tensile stress zone and uplift structure in the upper and middle crust corresponds with the surface hydrothermal active zone. There is an average Curie Point Depth(CPD) of 19.5–22.5 km in Batang, Litang, and Kangding. The local shear-wave(S-wave) velocity is relatively low in the middle and lower crust. The S-wave shows a low velocity trap(Vs3.2 km s.1) at 15–30 km, which is considered a high-temperature partial melting magma, the crustal source of the hydrothermal active zone. We conclude that the hydrothermal system in this area can be divided into Batang-type and Kangding-type, both of which rely on a crustal heating cycle of atmospheric precipitation and surface water along the fracture zone. The heat is derived from the middle and lower crust: groundwater penetrates the deep faults bringing geothermal energy back to the surface and forming high-temperature springs.     

Characterisation of ground motion recording stations in the Groningen gas field

The seismic hazard and risk analysis for the onshore Groningen gas field requires information about local soil properties, in particular shear-wave velocity (V_S). A fieldwork campaign was conducted at 18 surface accelerograph stations of the monitoring network. The subsurface in the region consists of unconsolidated sediments and is heterogeneous in composition and properties. A range of different methods was applied to acquire in situ V_S values to a target depth of at least 30 m. The techniques include seismic cone penetration tests (SCPT) with varying source offsets, multichannel analysis of surface waves (MASW) on Rayleigh waves with different processing approaches, microtremor array, cross-hole tomography and suspension P-S logging. The offset SCPT, cross-hole tomography and common midpoint cross-correlation (CMPcc) processing of MASW data all revealed lateral variations on length scales of several to tens of metres in this geological setting. SCPTs resulted in very detailed V_S profiles with depth, but represent point measurements in a heterogeneous environment. The MASW results represent V_S information on a larger spatial scale and smooth some of the heterogeneity encountered at the sites. The combination of MASW and SCPT proved to be a powerful and cost-effective approach in determining representative V_S profiles at the accelerograph station sites. The measured V_S profiles correspond well with the modelled profiles and they significantly enhance the ground motion model derivation. The similarity between the theoretical transfer function from the V_S profile and the observed amplification from vertical array stations is also excellent.     

Features of Radiation and Propagation of Seismic Waves in the Northern Caucasus: Manifestations in Regional Coda

Based on the Anapa (ANN) seismic station records of ~40 earthquakes (M_W > 3.9) that occurred within ~300 km of the station since 2002 up to the present time, the source parameters and quality factor of the Earth’s crust (Q(f)) and upper mantle are estimated for the S-waves in the 1–8 Hz frequency band. The regional coda analysis techniques which allow separating the effects associated with seismic source (source effects) and with the propagation path of seismic waves (path effects) are employed. The Q-factor estimates are obtained in the form Q(f) = 90 × f 0.7 for the epicentral distances r < 120 km and in the form Q(f) = 90 × f1.0 for r > 120 km. The established Q(f) and source parameters are close to the estimates for Central Japan, which is probably due to the similar tectonic structure of the regions. The shapes of the source parameters are found to be independent of the magnitude of the earthquakes in the magnitude range 3.9–5.6; however, the radiation of the high-frequency components (f > 4–5 Hz) is enhanced with the depth of the source (down to h ~ 60 km). The estimates Q(f) of the quality factor determined from the records by the Sochi, Anapa, and Kislovodsk seismic stations allowed a more accurate determination of the seismic moments and magnitudes of the Caucasian earthquakes. The studies will be continued for obtaining the Q(f) estimates, geometrical spreading functions, and frequency-dependent amplification of seismic waves in the Earth’s crust in the other regions of the Northern Caucasus.     

The seismic wave absorption in the crust and upper mantle in the vicinity of the Kislovodsk seismic station

The Q-factor estimates of the Earth’s crust and upper mantle as the functions of frequency (Q(f)) are obtained for the seismic S-waves at frequencies up to ~35 Hz. The estimates are based on the data for ~40 earthquakes recorded by the Kislovodsk seismic station since 2000. The magnitudes of these events are M_W > 3.8, the sources are located in the depth interval from 1 to 165 km, and the epicentral distances range from ~100 to 300 km. The Q-factor estimates are obtained by the methods developed by Aki and Rautian et al., which employ the suppression of the effects of the source radiation spectrum and local site responses in the S-wave spectra by the coda waves measured at a fixed lapse time (time from the first arrival). The radiation pattern effects are cancelled by averaging over many events whose sources are distributed in a wide azimuthal sector centered at the receiving site. The geometrical spreading was specified in the form of a piecewise-continuous function of distance which behaves as 1/R at the distances from 1 to 50 km from the source, has a plateau at 1/50 in the interval from 50–70 km to 130–150 km, and decays as \({\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 {\sqrt R }}}\right.\kern-\nulldelimiterspace} \!\lower0.7ex\hbox{${\sqrt R }$}}\) beyond 130–150 km. For this geometrical spreading model and some of its modifications, the following Q-factor estimates are obtained: Q(f) ~ 85f^0.9 at the frequencies ranging from ~1 to 20 Hz and Q(f) ~ 75f^1.0 at the frequencies ranging from ~1 to 35 Hz.     

Precursor times of abnormal <Emphasis Type="Italic">b</Emphasis>-values prior to mainshocks

Seismic observations exhibit the presence of abnormal b-values prior to numerous earthquakes. The time interval from the appearance of abnormal b-values to the occurrence of mainshock is called the precursor time. There are two kinds of precursor times in use: the first one denoted by T is the time interval from the moment when the b-value starts to increase from the normal one to the abnormal one to the occurrence time of the forthcoming mainshock, and the second one denoted by T _p is the time interval from the moment when the abnormal b-value reaches the peak one to the occurrence time of the forthcoming mainshock. Let T* be the waiting time from the moment when the abnormal b-value returned to the normal one to the occurrence time of the forthcoming mainshock. The precursor time, T (usually in days), has been found to be related to the magnitude, M, of the mainshock expected in a linear form as log(T)?=?q?+?rM where q and r are the coefficient and slope, respectively. In this study, the values of T, T _p , and T* of 45 earthquakes with 3?≤?M?≤?9 occurred in various tectonic regions are compiled from or measured from the temporal variations in b-values given in numerous source materials. The relationships of T and T _p , respectively, versus M are inferred from compiled data. The difference between the values of T and T _p decreases with increasing M. In addition, the plots of T*/T versus M, T* versus T, and T* versus T-T* will be made and related equations between two quantities will be inferred from given data.     

Geodynamic and Seismotectonic Activity in Eastern Tibet in the 21st Century

At the beginning of the 21st century, a series of great earthquakes were recorded in northeastern Tibet, along the periphery of the Bayan Hara lithospheric block. An earthquake with M_S = 8.1 occurred within the East Kunlun fault zone in the Kunlun Mountains, which caused an extended surface rupture with left-lateral strike slip. An earthquake with M_S = 8 occurred in Wenchuan (China) on May 12, 2008, giving rise to an extended overthrust along the Lunmanshan fault zone. An earthquake with M_S = 7.1 occurred in Yushu (China) on April 14, 2010; its epicenter was on the Grazze–Yushu–Funchuoshan fault; a left-lateral strikeslip offset was observed on the surface. An earthquake with M_S = 7 occurred in the vicinity of Lushan on April 20, 2013; its epicenter was within the Lunmanshan fault zone, 103 km southwest of the zone of the catastrophic Wenchuan earthquake. An earthquake with M_S = 8.2 occurred in Nepal on April 25, 2015. Based on the CSN seismic catalog, the energy of all earthquakes in eastern Tibet at the end of the 20th and beginning of the 21st centuries was estimated. It was found that Tibet was seismically quiet from 1980 to 2000. The beginning of the 21st century has been marked by seismic activation with earthquake sources migrating southward to surround the Bayan Hara lithospheric block from every quarter. Therefore, this block can be regarded as one of the most seismically active regions of China.     

Neural network estimate of seismic velocities and resistivity of rocks from electromagnetic and seismic sounding data

The neural network estimates of seismic P- and S-wave velocities from electrical resistivity of the rocks and, vice versa, resistivity estimates from seismic velocities are presented. It is shown that, depending on the ratio between the volumes of the known data and the data to be reconstructed, the accuracy of the estimates of the P- and S-wave velocities ranges within 1–4 and 4–6%, respectively. The logarithmic resistivity is estimated from seismic P- and S-velocities as accurately as up to 15–17%. In all cases, the biggest errors are obtained when the estimates are based on correlated data.     

Near-source attenuation of high-frequency body waves beneath the New Madrid Seismic Zone

Attenuation characteristics in the New Madrid Seismic Zone (NMSZ) are estimated from 157 local seismograph recordings out of 46 earthquakes of 2.6?≤?M?≤?4.1 with hypocentral distances up to 60 km and focal depths down to 25 km. Digital waveform seismograms were obtained from local earthquakes in the NMSZ recorded by the Center for Earthquake Research and Information (CERI) at the University of Memphis. Using the coda normalization method, we tried to determine Q values and geometrical spreading exponents at 13 center frequencies. The scatter of the data and trade-off between the geometrical spreading and the quality factor did not allow us to simultaneously derive both these parameters from inversion. Assuming 1/R ^1.0 as the geometrical spreading function in the NMSZ, the Q _P and Q _S estimates increase with increasing frequency from 354 and 426 at 4 Hz to 729 and 1091 at 24 Hz, respectively. Fitting a power law equation to the Q estimates, we found the attenuation models for the P waves and S waves in the frequency range of 4 to 24 Hz as Q _P?=?(115.80?±?1.36) f ^{(0.495?±?0.129)} and Q _S?=?(161.34?±?1.73) f ^{(0.613?±?0.067)}, respectively. We did not consider Q estimates from the coda normalization method for frequencies less than 4 Hz in the regression analysis since the decay of coda amplitude was not observed at most bandpass filtered seismograms for these frequencies. Q _S/Q _P?>?1, for 4?≤?f?≤?24 Hz as well as strong intrinsic attenuation, suggest that the crust beneath the NMSZ is partially fluid-saturated. Further, high scattering attenuation indicates the presence of a high level of small-scale heterogeneities inside the crust in this region.     

A novel approach for classification of earthquake ground-motion records

This paper presents a new clustering procedure based on K-means and self-organizing map (SOM) network algorithms for classification of earthquake ground-motion records. Six scalar indicators are used in data analysis for describing the frequency content features of earthquake ground motions, named as the average spectral period (T _avg ), the mean period (T _m ), the smoothed spectral predominant period (T ₀), the characteristic period (T _4.3), the predominant period based on velocity spectrum (T _gSv ), and the shape factor (Ω). Different clustering validity indexes were applied to determine the best estimates of the number of clusters on real and synthetic data. Results showed the high performance of proposed procedure to reveal salient features of complex seismic data. The comparison between the results of clustering analyses recommend the smoothed spectral predominant period as an effective indicator to describe ground-motion classes. The results also showed that K-means algorithm has better performance than SOM algorithm in identification and classification procedure of ground-motion records.     

Determination of source parameters of strong earthquakes of Kamchatka,the Kurile Islands,and Japan from aftershock sequences

Aftershock sequences of some strong earthquakes of Kamchatka, the Kurile Islands, and Japan are examined. Such source parameters as the length L, along-dip width W, motion on fault D, and stress drop Δσ are determined from the aftershock sequences considered. The values of these parameters were obtained by the formal estimation of linear source parameters (lower bound estimates) and visually (upper bound estimates). The correlation dependences of the obtained parameters on the surface wave (M _S ) and seismic moment (M _W ) magnitudes are calculated.     

Estimation of monthly evaporative loss using relevance vector machine,extreme learning machine and multivariate adaptive regression spline models

The forecasting of evaporative loss (E) is vital for water resource management and understanding of hydrological process for farming practices, ecosystem management and hydrologic engineering. This study has developed three machine learning algorithms, namely the relevance vector machine (RVM), extreme learning machine (ELM) and multivariate adaptive regression spline (MARS) for the prediction of E using five predictor variables, incident solar radiation (S), maximum temperature (T _max), minimum temperature (T _min), atmospheric vapor pressure (VP) and precipitation (P). The RVM model is based on the Bayesian formulation of a linear model with appropriate prior that results in sparse representations. The ELM model is computationally efficient algorithm based on Single Layer Feedforward Neural Network with hidden neurons that randomly choose input weights and the MARS model is built on flexible regression algorithm that generally divides solution space into intervals of predictor variables and fits splines (basis functions) to each interval. By utilizing random sampling process, the predictor data were partitioned into the training phase (70 % of data) and testing phase (remainder 30 %). The equations for the prediction of monthly E were formulated. The RVM model was devised using the radial basis function, while the ELM model comprised of 5 inputs and 10 hidden neurons and used the radial basis activation function, and the MARS model utilized 15 basis functions. The decomposition of variance among the predictor dataset of the MARS model yielded the largest magnitude of the Generalized Cross Validation statistic (≈0.03) when the T _max was used as an input, followed by the relatively lower value (≈0.028, 0.019) for inputs defined by the S and VP. This confirmed that the prediction of E utilized the largest contributions of the predictive features from the T _max, verified emphatically by sensitivity analysis test. The model performance statistics yielded correlation coefficients of 0.979 (RVM), 0.977 (ELM) and 0.974 (MARS), Root-Mean-Square-Errors of 9.306, 9.714 and 10.457 and Mean-Absolute-Error of 0.034, 0.035 and 0.038. Despite the small differences in the overall prediction skill, the RVM model appeared to be more accurate in prediction of E. It is therefore advocated that the RVM model can be employed as a promising machine learning tool for the prediction of evaporative loss.