On selection of kernel parametes in relevance vector machines for hydrologic applications

Recent advances in statistical learning theory have yielded tools that are improving our capabilities for analyzing large and complex datasets. Among such tools, relevance vector machines (RVMs) are finding increasing applications in hydrology because of (1) their excellent generalization properties, and (2) the probabilistic interpretation associated with this technique that yields prediction uncertainty. RVMs combine the strengths of kernel-based methods and Bayesian theory to establish relationships between a set of input vectors and a desired output. However, a bias–variance analysis of RVM estimates revealed that a careful selection of kernel parameters is of paramount importance for achieving good performance from RVMs. In this study, several analytic methods are presented for selection of kernel parameters. These methods rely on structural properties of the data rather than expensive re-sampling approaches commonly used in RVM applications. An analytical expression for prediction risk in leave-one-out cross validation is derived. For brevity, the effectiveness of the proposed methods is assessed first by data generated from the benchmark sinc function, followed by an example involving estimation of hydraulic conductivity values over a field based on observations. It is shown that a straightforward maximization of likelihood function can lead to misleading results. The proposed methods are found to yield robust estimates of parameters for kernel functions.     

Mapping of Marine Chl-a and Suspended Solid Concentration Using OCM-2 Sensor

Chlorophyll-a (Chl-a) and Suspended Solid Concentration (SSC) shows the productivity of water and their surrounding environment. These parameters can be effectively estimated through several remote sensing techniques. From the recent reports on the Gulf of Thailand, it is found that Chl-a and SSC are increasing in coastal areas due to changing environment caused by variations in the global carbon cycle, climate change and water pollution linking to anthropogenic conditions such as high population density and rapid urbanization in neighbouring coastal areas deteriorating the coastal and marine environment. Various models are evaluated in this study for estimation of marine Chl-a and SSC by employing Ocean Colour Monitor-2 sensor of Oceansat-2 satellite for Northern Gulf of Thailand. The retrieval of Chl-a and SSC by the atmospheric correction of visible bands from 400 to 700 nm to attain normalized water-leaving radiances and then a suitable algorithm is applied. The In-situ reflectance values of sea waters are measured using the ASD spectroradiometer. The reflectance values of the spectroradiometer are correlated for the same day atmospherically corrected satellite reflectance and the analysis offers high correlation R² 0.73. Satellite derived, Chl-a and SSC are correlated with observed in situ Chl-a and SSC. This analysis offered better correlation of R² 0.86 and 0.85 respectively with the algorithms of Chl-a and SSC.     

Impact of LULC change on the runoff,base flow and evapotranspiration dynamics in eastern Indian river basins during 1985–2005 using variable infiltration capacity approach

As a catchment phenomenon, land use and land cover change (LULCC) has a great role in influencing the hydrological cycle. In this study, decadal LULC maps of 1985, 1995, 2005 and predicted-2025 of the Subarnarekha, Brahmani, Baitarani, Mahanadi and Nagavali River basins of eastern India were analyzed in the framework of the variable infiltration capacity (VIC) macro scale hydrologic model to estimate their relative consequences. The model simulation showed a decrease in ET with 0.0276% during 1985–1995, but a slight increase with 0.0097% during 1995–2005. Conversely, runoff and base flow showed an overall increasing trend with 0.0319 and 0.0041% respectively during 1985–1995. In response to the predicted LULC in 2025, the VIC model simulation estimated reduction of ET with 0.0851% with an increase of runoff by 0.051%. Among the vegetation parameters, leaf area index (LAI) emerged as the most sensitive one to alter the simulated water balance. LULC alterations via deforestation, urbanization, cropland expansions led to reduced canopy cover for interception and transpiration that in turn contributed to overall decrease in ET and increase in runoff and base flow. This study reiterates changes in the hydrology due to LULCC, thereby providing useful inputs for integrated water resources management in the principle of sustained ecology.     

Effect of overtaking disturbances on the propagation of spherical shock wave through self-gravitating gas

Effect of overtaking disturbances on the propagation of a spherical shock wave in self gravitating gas has been studied by the technique developed by the first author [Mod. Meas. Cont. B,46(4), 1 (1992)]. The analytical expressions for modified shock velocity and shock strength have been obtained for an initial density distribution0 =r ^–w, where is the density at the axis of symmetry andw is a constant; simultaneously, for the two cases viz.; (i) when the shock is strong and ii) when it is weak. The results accomplished here have been compared with those for freely propagation of shock.It is observed that the conclusions arrived at here agree with experimental observations. Finally, the modified expressions for the pressure, the density and the particle velocity immediately behind the shock have also been derived from, for both cases.     

On the marine atmospheric boundary layer characteristics over Bay of Bengal and Arabian Sea during the Integrated Campaign for Aerosols,gases and Radiation Budget (ICARB)

Detailed measurements were carried out in the Marine Atmospheric Boundary Layer (MABL) during the Integrated Campaign for Aerosols, gases and Radiation Budget (ICARB) which covered both Arabian Sea and Bay of Bengal during March to May 2006. In this paper, we present the meteorological observations made during this campaign. The latitudinal variation of the surface layer turbulent fluxes is also described in detail.     

Fashion and phases of late Pleistocene aggradation and incision in the Alaknanda River Valley, western Himalaya, India

We study the aggradation and incision of the Alaknanda River Valley during the late Pleistocene and Holocene. The morphostratigraphy in the river valley at Deoprayag shows the active riverbed, a cut terrace, and a fill terrace. The sedimentary fabric of the fill terrace comprises four lithofacies representing 1) riverbed accretion, 2) locally derived debris fan, 3) the deposits of waning floods and 4) palaeoflood records. The sedimentation style, coupled with geochemical analysis and Optically Stimulated Luminescence (OSL) dating, indicate that this terrace formed in a drier climate and the river valley aggraded in two phases during 21–18 ka and 13–9 ka. During these periods, sediment supply was relatively higher. Incision began after 10 ka in response to a strengthened monsoon and aided by increase of the tectonic gradient. The cut terrace formed at ~ 5 ka during a phase of stable climate and tectonic quiescence. The palaeoflood records suggest wetter climate 200–300 yr ago when the floods originated in the upper catchment of the Higher Himalaya and in the relatively drier climate ~ 1.2 ka when locally derived sediments from the Lesser Himalaya dominated flood deposits. Maximum and minimum limits of bedrock incision rate at Deoprayag are 2.3 mm/a and 1.4 mm/a.     

Experimental and parametric investigation of effects of built-in staircases on the dynamics of RC buildings

Past earthquakes, in many instances, have demonstrated poor performance of commonly used built-in staircase configurations. Codal provisions in India pertaining to staircases present a rather simple approach wherein the effects of built-in staircases on the overall dynamic properties or on the local behavior of structures are not addressed explicitly. Studies in the past have highlighted the scale of such effects, but most of them have relied completely on analytical models of buildings. This study analyzes the adequacy of the codal provisions by investigating two finite element (FE) models calibrated using ambient and forced vibration measurements. The effects of variations in building height, layout of staircase in plan, and presence of masonry infill walls in stairwells are also examined. The codal guidelines regarding empirical estimation of period and provision of enclosure walls around built-in staircases are found to be adequate. However, for the case of built-in staircases without enclosure walls, the force and displacement demands on landing beams are found to be considerably high. Drift-based approaches to estimate these demands are proposed.     

Probabilities for the occurrences of medium to large earthquakes in northeast India and adjoining region

The return periods and occurrence probabilities related to medium and large earthquakes (M _w 4.0–7.0) in four seismic zones in northeast India and adjoining region (20°–32°N and 87°–100°E) have been estimated with the help of well-known extreme value theory using three methods given by Gumbel (1958), Knopoff and Kagan (1977) and Bury (1999). In the present analysis, the return periods, the most probable maximum magnitude in a specified time period and probabilities of occurrences of earthquakes of magnitude M ≥ 4.0 have been computed using a homogeneous and complete earthquake catalogue prepared for the period between 1897 and 2007. The analysis indicates that the most probable largest annual earthquakes are close to 4.6, 5.1, 5.2, 5.5 and 5.8 in the four seismic zones, namely, the Shillong Plateau Zone, the Eastern Syntaxis Zone, the Himalayan Thrusts Zone, the Arakan-Yoma subduction zone and the whole region, respectively. The most probable largest earthquakes that may occur within different time periods have been also estimated and reported. The study reveals that the estimated mean return periods for the earthquake of magnitude M _w 6.5 are about 6–7 years, 9–10 years, 59–78 years, 72–115 years and 88–127 years in the whole region, the Arakan-Yoma subduction zone, the Himalayan Thrusts Zone, the Shillong Plateau Zone and the Eastern Syntaxis Zone, respectively. The study indicates that Arakan-Yoma subduction zone has the lowest mean return periods and high occurrence probability for the same earthquake magnitude in comparison to the other zones. The differences in the hazard parameters from zone to zone reveal the high crustal heterogeneity and seismotectonics complexity in northeast India and adjoining regions.     

Did Tsunami tremor jolt microbial biomass and their activities in soils? A case study in Andaman Island,India

The nature of selected soil-chemical and microbial properties influenced by tsunami affected and non-affected areas along the border areas of the alluvial Andaman Island in India were investigated. Soils of these areas have turned saline and saline-sodic due to the ingression of sea water. The electrical conductivity of the saturation extract of the surface soil varied from 11.2 to 23.8 dS m^?1 in 2005, and it was decreased to 0.8–10.3 dS m^?1 in 2006 due to the heavy rain in the following year after the tsunami. Soil quality indicators, like microbial biomass C, microbial metabolic quotient, microbial respiration quotient and fluorescein diacetate hydrolyzing activity, decreased in the tsunami affected soil in 2005, but slightly increased in 2006. All microbial parameters were significantly negatively correlated with the electrical conductivity, sodium absorption ratio and exchangeable sodium percentage. Suppression of microbial biomass and their activities in the soils due to the increased-salinity is of great agronomic significance and needs suitable intervention for sustainable crop production. Significant differences were found in soil-chemical and microbial characteristics between tsunami affected and non-affected areas. Hierarchical clustering algorithm on the basis of different soil-chemical and microbial characteristics revealed that there is significant difference in grouping between tsunami affected and non-affected zones. From this study, it can be concluded that the sea water ingression detrimentally influenced the microbial properties of tsunami affected soil.