Spatial Variability of Rock Depth in Bangalore Using Geostatistical, Neural Network and Support Vector Machine Models

Geospatial technology is increasing in demand for many applications in geosciences. Spatial variability of the bed/hard rock is vital for many applications in geotechnical and earthquake engineering problems such as design of deep foundations, site amplification, ground response studies, liquefaction, microzonation etc. In this paper, reduced level of rock at Bangalore, India is arrived from the 652 boreholes data in the area covering 220 km². In the context of prediction of reduced level of rock in the subsurface of Bangalore and to study the spatial variability of the rock depth, Geostatistical model based on Ordinary Kriging technique, Artificial Neural Network (ANN) and Support Vector Machine (SVM) models have been developed. In Ordinary Kriging, the knowledge of the semi-variogram of the reduced level of rock from 652 points in Bangalore is used to predict the reduced level of rock at any point in the subsurface of the Bangalore, where field measurements are not available. A new type of cross-validation analysis developed proves the robustness of the Ordinary Kriging model. ANN model based on multi layer perceptrons (MLPs) that are trained with Levenberg–Marquardt backpropagation algorithm has been adopted to train the model with 90% of the data available. The SVM is a novel type of learning machine based on statistical learning theory, uses regression technique by introducing loss function has been used to predict the reduced level of rock from a large set of data. In this study, a comparative study of three numerical models to predict reduced level of rock has been presented and discussed.     

Experimental study on loss factor for corrugated plates by bandwidth method

The material loss factor for technically orthotropic plates was measured by half-power bandwidth method. Rectangular and trapezoidal corrugated plates of steel were considered. A simple isotropic steel plate was also considered for comparison of the results. The concept of single degree of freedom system was adopted. The tests were undertaken at very low frequency range (0–100) Hz. The plate models were suspended freely with two wires to minimize or prevent excessive extraneous energy dissipation. Out of plane point force, random in nature was applied to the top middle of the plates and the responses were measured from the middle point of the plates by FFT analyzer using miniature small mass accelerometer as sensor. The aim of these tests is to investigate the effects of bending rigidity and mode orders over material loss factor. The values of estimated modal damping loss factors are compared and tabulated for the plates models considered. Natural frequencies of some of the initial modes of the plates are also presented.It is observed that the higher the value of bending rigidity of the plates, the larger the values of loss factor of it. There was a significant increase in value of loss factor in corrugated plates to that of the isotropic plate.     

Impact of horizontal resolution on prediction of tropical cyclones over Bay of Bengal using a regional weather prediction model

The present study is carried out to examine the performance of a regional atmospheric model in forecasting tropical cyclones over the Bay of Bengal and its sensitivity to horizontal resolution. Two cyclones, which formed over the Bay of Bengal during the years 1995 and 1997, are simulated using a regional weather prediction model with two horizontal resolutions of 165 km and 55 km. The model is found to perform reasonably well towards simulation of the storms. The structure, intensity and track of the cyclones are found to be better simulated by finer resolution of the model as compared to the coarse resolution. Rainfall amount and its distribution are also found to be sensitive to the model horizontal resolution. Other important fields, viz., vertical velocity, horizontal divergence and horizontal moisture flux are also found to be sensitive to model horizontal resolution and are better simulated by the model with finer horizontal grids.     

Kinematics of the Gondwana basins of peninsular India

The Gondwana successions (1–4 km thick) of peninsular India accumulated in a number of discrete basins during Permo-Triassic period. The basins are typically bounded by faults that developed along Precambrian lineaments during deposition, as well as affected by intrabasinal faults indicating fault-controlled synsedimentary subsidence. The patterns of the intrabasinal faults and their relationships with the respective basin-bounding faults represent both extensional and strike-slip regimes. Field evidence suggests that preferential subsidence in locales of differently oriented discontinuities in the Precambrian basement led to development of Gondwana basins with varying, but mutually compatible, kinematics during a bulk motion, grossly along the present-day E–W direction. The kinematic disparity of the individual basins resulted due to different relative orientations of the basement discontinuities and is illustrated with the help of a simple sandbox model. The regional E–W motion was accommodated by strike-slip motion on the transcontinental fault in the north.     

Simulation of Orissa Super Cyclone (1999) using PSU/NCAR Mesoscale Model

In this study a non-hydrostatic version of Penn State University (PSU) -- NationalCenter for Atmospheric Research (NCAR) mesoscale model is used to simulate thesuper cyclonic storm that crossed Orissa coast on 29 October 1999. The model isintegrated up to 123 h for producing 5-day forecast of the storm. Several importantfields including sea level pressure, horizontal wind and rainfall are compared with theverification analysis/observation to examine the performance of the model. The modelsimulated track of the cyclone is compared with the best-fit track obtained from IndiaMeteorological Department (IMD) and the track obtained from NCEP/NCAR reanalysis. The model is found to perform reasonably well in simulating the track and in particular, the intensity of the storm.     

Stable carbon and nitrogen isotopic characteristics of PM2.5 and PM10 in Delhi,India

This study presents the chemical composition (carbonaceous and nitrogenous components) of aerosols (PM_2.5 and PM₁₀) along with stable isotopic composition (δ¹³C and δ¹⁵N) collected during winter and the summer months of 2015–16 to explore the possible sources of aerosols in megacity Delhi, India. The mean concentrations (mean?±?standard deviation at 1σ) of PM_2.5 and PM₁₀ were 223?±?69 µg m^?3 and 328?±?65 µg m^?3, respectively during winter season whereas the mean concentrations of PM_2.5 and PM₁₀ were 147?±?22 µg m^?3 and 236?±?61 µg m^?3, respectively during summer season. The mean value of δ¹³C (range: ??26.4 to ??23.4‰) and δ¹⁵N (range: 3.3 to 14.4‰) of PM_2.5 were ??25.3?±?0.5‰ and 8.9?±?2.1‰, respectively during winter season whereas the mean value of δ¹³C (range: ??26.7 to ??25.3‰) and δ¹⁵N (range: 2.8 to 11.5‰) of PM_2.5 were ??26.1?±?0.4‰ and 6.4?±?2.5‰, respectively during the summer season. Comparison of stable C and N isotopic fingerprints of major identical sources suggested that major portion of PM_2.5 and PM₁₀ at Delhi were mainly from fossil fuel combustion (FFC), biomass burning (BB) (C-3 and C-4 type vegitation), secondary aerosols (SAs) and road dust (SD). The correlation analysis of δ¹³C with other C (OC, TC, OC/EC and OC/WSOC) components and δ¹⁵N with other N components (TN, NH₄⁺ and NO₃^?) are also support the source identification of isotopic signatures.

     

Field based index of flood vulnerability(IFV): A new validation technique for flood susceptible models

The flood hazard management is one of the major challenges in the floodplain regions worldwide. With the rise in population growth and the spread of infrastructural development, the level of risk has increased over time.Therefore, the prediction of flood susceptible area is a key challenge for the adoption of management plans.Flood susceptibility modeling is technically a common work, but it is still a very tough job to validate flood susceptible models in a very rigorous and scientific manner. Therefore, the present work in the Atreyee River Basin of India and Bangladesh was planned to establish artificial neural network(ANN), radial basis function(RBF), random forest(RF) and their ensemble-based flood susceptibility models. The flood susceptible models were constructed based on nine flood conditioning parameters. The flood susceptibility models were validated in a conventional way using the receiver operating curve(ROC). To validate the flood-susceptible models, a two dimensional(2 D) hydraulic flood simulation model was developed. Also, the index of flood vulnerability model was developed and applied for validating the flood susceptible models, which was a very unique way to validate the predictive models. Friedman test and Wilcoxon Signed rank test were employed to compare the generated flood susceptible models. Results showed that 11.95%–12.99% of the entire basin area(10188.4 km2) comes under very high flood-susceptible zones. Accuracy evaluation results have shown that the performance of ensemble flood susceptible models outperforms other standalone machine learning models. The flood simulation model and IFV model were also spatially adjusted with the flood susceptibility models. Therefore, the present study recommended for the ensemble flood susceptibility prediction and IFV based validation along with conventional ways.     

Applying deep learning and benchmark machine learning algorithms for landslide susceptibility modelling in Rorachu river basin of Sikkim Himalaya,India

Landslide is considered as one of the most severe threats to human life and property in the hilly areas of the world. The number of landslides and the level of damage across the globe has been increasing over time. Therefore, landslide management is essential to maintain the natural and socio-economic dynamics of the hilly region. Rorachu river basin is one of the most landslide-prone areas of the Sikkim selected for the present study. The prime goal of the study is to prepare landslide susceptibility maps(LSMs) using computer-based advanced machine learning techniques and compare the performance of the models.To properly understand the existing spatial relation with the landslide, twenty factors, including triggering and causative factors, were selected. A deep learning algorithm viz. convolutional neural network model(CNN) and three popular machine learning techniques, i.e., random forest model(RF), artificial neural network model(ANN), and bagging model, were employed to prepare the LSMs. Two separate datasets including training and validation were designed by randomly taken landslide and nonlandslide points. A ratio of 70:30 was considered for the selection of both training and validation points.Multicollinearity was assessed by tolerance and variance inflation factor, and the role of individual conditioning factors was estimated using information gain ratio. The result reveals that there is no severe multicollinearity among the landslide conditioning factors, and the triggering factor rainfall appeared as the leading cause of the landslide. Based on the final prediction values of each model, LSM was constructed and successfully portioned into five distinct classes, like very low, low, moderate, high, and very high susceptibility. The susceptibility class-wise distribution of landslides shows that more than 90% of the landslide area falls under higher landslide susceptibility grades. The precision of models was examined using the area under the curve(AUC) of the receiver operating characteristics(ROC) curve and statistical methods like root mean square error(RMSE) and mean absolute error(MAE). In both datasets(training and validation), the CNN model achieved the maximum AUC value of 0.903 and 0.939, respectively. The lowest value of RMSE and MAE also reveals the better performance of the CNN model. So, it can be concluded that all the models have performed well, but the CNN model has outperformed the other models in terms of precision.