Search

• Solar power prediction has been done by ensemble machine learning (EML) models. • Impact of meteorological parameters on solar power generation is investigated. • A comprehensive test-bed is designed for feature selection of training parameters. • Algorithm performance have been validated by a 10kW p solar power plant data. • Stacking and Voting EML algorithms demonstrates superior prediction accuracy. The challenges in applications of solar energy lies in its intermittency and dependency on meteorological parameters such as; solar radiation, ambient temperature, rainfall, wind-speed etc., and many other physical parameters like dust accumulation etc. Hence, it is important to estimate the amount of solar photovoltaic (PV) power generation for a specific geographical location. Machine learning (ML) models have gained importance and are widely used for prediction of solar power plant performance. In this paper, the impact of weather parameters on solar PV power generation is estimated by several Ensemble ML (EML) models like Bagging, Boosting, Stacking, and Voting for the first time. The performance of chosen ML algorithms is validated by field dataset of a 10kWp solar PV power plant in Eastern India region. Furthermore, a complete test-bed framework has been designed for data mining as well as to select appropriate learning models. It also supports feature selection and reduction for dataset to reduce space and time complexity of the learning models. The results demonstrate greater prediction accuracy of around 96% for Stacking and Voting EML models. The proposed work is a generalized one and can be very useful for predicting the performance of large-scale solar PV power plants also. [ABSTRACT FROM AUTHOR]

Motivated by the growing significance of solar energy, the paper presents the findings of the maiden study on the long-term seasonal performance assessment of three PV module technologies—monocrystalline (m-Si), polycrystalline (p-Si), and amorphous silicon (a-Si) housed in a 1 MW solar plant installed at a local steppe climate in Gandhinagar, Gujarat, India using the key performance indicators (KPIs), daily power generation, final yield (Y f), reference yield (Y r), total energy loss (T EL) and performance ratio (PR). Notably, the am-Si PV modules were found to be the best performing technology, with an average PR of 71.26% and the lowest average total energy loss of 631 h during the monitored period. Moreover, PV systems have become more popular as a source of green, clean energy; yet, they have low base-load energy sustainability. Explainable AI (XAI) aids in understanding predictions made by the machine learning models. The importance of XAI lies in the comprehension and trust which is embedded into the results of machine learning algorithms. Therefore, the present work builds a comprehensive XAI model for predicting KPIs and assesses the performance of several learning algorithms using R-value, MAE, the number of iterations, and execution time. The Levenberg-Marquardt algorithm predicts the KPIs for p-Si, am-Si, and m-Si with prediction accuracies of 98.63%, 98.58%, and 90.09%, respectively. In summary, this study not only advances our understanding of PV module performance in a steppe climate context but also contributes to the broader field by integrating an XAI model for transparent and reliable predictions of the KPIs. [Display omitted] • Comparison of m-Si, p-Si, and a-Si PV modules in a 1 MW solar plant, Gandhinagar, India, showed long-term performance. • a-Si PV modules outperformed with a 71.26% PR and the lowest 631 hours energy loss, surpassing m-Si and p-Si technologies. • Study underscores XAI's role in ML predictions, developing a robust model to predict KPIs and evaluate learning algorithms. • Levenberg-Marquardt showed high accuracy: 98.63% for p-Si, 98.58% for a-Si, and 90.09% for m-Si PV modules. [ABSTRACT FROM AUTHOR]

Electrical energy is one of the key components for the development and sustainability of any nation. India is a developing country and blessed with a huge amount of renewable energy resources still there are various remote areas where the grid supply is rarely available. As electrical energy is the basic requirement, therefore it must be taken up on priority to exploit the available renewable energy resources integrated with storage devices like fuel cells and batteries for power generation and help the planners in providing the energy-efficient and alternative solution. This solution will not only meet electricity demand but also helps reduce greenhouse gas emissions as a result the efficient, sustainable and eco-friendly solution can be achieved which would contribute a lot to the smart grid environment. In this paper, a modified grey wolf optimizer approach is utilized to develop a hybrid microgrid based on available renewable energy resources considering modern power grid interactions. The proposed approach would be able to provide a robust and efficient microgrid that utilizes solar photovoltaic technology and wind energy conversion system. This approach integrates renewable resources with the meta-heuristic optimization algorithm for optimal dispatch of energy in grid-connected hybrid microgrid system. The proposed approach is mainly aimed to provide the optimal sizing of renewable energy-based microgrids based on the load profile according to time of use. To validate the proposed approach, a comparative study is also conducted through a case study and shows a significant savings of 30.88% and 49.99% of the rolling cost in comparison with fuzzy logic and mixed integer linear programming-based energy management system respectively. • The article proposes a hybrid MG with renewable energy sources and energy storage systems (battery and fuel cell). • Develop a novel energy management system using modified grey wolf optimizer (MGWO) with multi objective functions. • Analyzes the load demand and the cost of electricity from the power grid for effective designing of MG at test locations. • A significant savings of 30.88% and 49.99% of the rolling cost in comparison with FL and MILP based EMS respectively. [ABSTRACT FROM AUTHOR]

• Landscape ecological vulnerability was assessed using bagging ANN-MLP and RF models. • B-MLP model was found best fit model for LEV assessment based on ROC curve, recall, precision, F1-score and accuracy. • Largest area under very high LEV was recorded in the southern part of the plains. • Rainfall, soil, vegetation and LULC were found influencing factors for LEV. • Plantation, embankments and geotextile technology are suggested to check erosion. Riverbank erosion is one of the most catastrophic hazards that renders floodplains vulnerable across the world vulnerable. It creates a significant negative impact on the environment and socio-economic life. This paper attempts to assess the landscape ecological vulnerability (LEV) to riverbank erosion in the Middle Brahmaputra floodplains of Assam, India by employing two machine learning models, namely artificial neural network-multilayer perceptron (ANN-MLP) and random forest (RF). Bagging ensembles were created for both ANN-MLP and RF (B-MLP and B-RF) to generate LEV zones. A total of eleven site-specific parameters were considered for the study. The receiver operating characteristic (ROC) based area under curve (AUC), accuracy, precision, recall and F1-score were used to validate the models and judge the models' performance. A sensitivity analysis was performed to deduce the most influential LEV parameters. The results revealed that B-MLP was the better-performing model compared to B-RF based on all five validation metrics. The largest area was found under very high vulnerability zone followed by very low, low, high and moderate vulnerability zones, based on both ensemble models. The western part of the floodplains was found to be more vulnerable than the eastern part. Moreover, the southern bank faced more vulnerability in comparison to the northern bank. The factors namely rainfall, soil type, vegetation and land /use land cover (LULC) influenced bank erosion vulnerability. This research provides evidence that the study area is under severe threat to riverbank erosion and urgently requires the implementation of effective mitigation measures. The study might benefit policymakers and local stakeholders to protect the floodplains from bank erosion and reduce vulnerability. [ABSTRACT FROM AUTHOR]

China, India, and the USA are the world's biggest energy consumers and CO 2 emitters. Being the leading contributors to climate change, these economies are also at the core of environmental solutions. This paper investigates the causal relationship among solar and wind energy production, coal consumption, economic growth, and CO 2 emissions for these three countries. To do so, we use an advanced methodology in Machine Learning to verify the predictive causal linkages among variables. The Causal Direction from Dependency (D2C) algorithm set CO 2 emissions as the target variable. The obtained results were disaggregated and estimated in a supervised prediction model. The findings, confirmed by three different Machine Learning procedures, showed an interesting output. While a reduction in overall carbon emissions is predicted in China and the US (resulting from the intensive use of renewable sources of energy), India displays critical predictions of a rise in CO 2 emissions. This indicates that curbing CO 2 emissions cannot be achieved without conducting a comprehensive shift from fossil to renewable resources, although China and the U.S. present a more promising path to sustainability than India. Being an emerging renewable energy leader, India should further enhance the use of low-carbon sources in its power supply and limit its dependence on coal. • Renewable energies, coal consumption, economic growth, and CO 2 we examine. • We investigate the causal predicted relationship on China, India, and the USA. • An advanced methodology in Machine Learning is used. • India has a prediction of an increase in emissions of CO 2. • China and the U.S. present a more promising path to sustainability than India. [ABSTRACT FROM AUTHOR]

Global Navigation satellite systems (GNSS) are predominantly affected by Ionospheric space weather. The GNSS signal delays due to ionospheric Total Electron Content (TEC) can be forecasted using advances in the emerging mathematics tools and algorithms. Machine learning algorithms such as Gaussian Process Regression (GPR) is considered in the present paper to implement in the forecasting of low-latitude ionospheric conditions. The GPS receiver data is obtained for 8 years (2009–2016) during 24th solar cycle from International GNSS Services (IGS) station located in Bengaluru (Geographical latitude: 12.97° N, Geographical longitude: 77.59° E), India. The performance of GPR model is validated using statistical parameters such as Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Mean Square Error (MSE), Root Mean Square Error (RMSE), and correlation coefficient. The results of the proposed GPR model were compared with the Auto Regressive Moving Average (ARMA) model and Artificial Neural Networks (ANN) model during solar maximum period and descending phase of 24th solar cycle. The experiment results are evident that GPR model is significantly providing the promising results in forecasting the ionospheric time delays for GNSS signals. The outcome of the work can be useful to develop web based Ionospheric TEC forecasting system to alert the GNSS users. • The GPS-TEC variability is investigated during solar maximum period and descending phase of 24th solar cycle. • The performance of ARMA, ANN and GPR models are evaluated in forecasting the ionospheric delays for GPS signals. • GPR model has better predicted TEC than the existing ARMA and ANN models during geomagnetic quiet and storm periods. • Kernel-based approach of GPR model is well suitable to forecast the ionospheric TEC variations over low latitudes. [ABSTRACT FROM AUTHOR]

Identifying current and future informal regions within cities remains a crucial issue for policymakers and governments in developing countries. The delineation process of identifying such regions in cities requires a lot of resources. While there are various studies that identify informal settlements based on satellite image classification, relying on both supervised or unsupervised machine learning approaches, these models either require multiple input data to function or need further development with regards to precision. In this paper, we introduce a novel method for identifying and informal settlements using street intersection data. With such minimal input data, we attempt to provide planners and policy-makers with a pragmatic tool that can aid in identifying informal zones in cities. The algorithm of the model is based on spatial statistics and Artificial Neural Networks (ANN). The proposed model relies on defining informal settlements based on two ubiquitous characteristics of informal settlements. We applied the model in five major cities in Egypt and India that have spatial structures in which informality is present. These cities are Greater Cairo, Alexandria, Hurghada and Minya in Egypt, and Mumbai in India. The predictSLUMS model shows high validity and accuracy for identifying and predicting informality within the same city the model was trained on or in different ones of a similar context. To enable this model to be used as a pragmatic tool for future prediction or further research concerning informal settlements and slums, we have made the python code for the pre-trained ANN models of the five studied cities. • A modelling framework is developed for predicting informal settlements. • The model only requires street intersection data as input, making it broadly applicable. • The model is applied to five cities; Greater Cairo, Alexandria, Hurghada and Minya in Egypt, and Mumbai in India. • By training the model using data from one city, the model can predict informal settlements in another city. • Using ground truth data of planning status for validation, the model achieves accuracy of between 82.1 and 97.6%. [ABSTRACT FROM AUTHOR]

How far the de-linking and morphological and hydrological degradation of tie channels connecting river to the wetland are caused for the areal and hydrological transformation of a wetland was not received enough attention in previous literature. This paper tried to explain this about the confluence reach of the Dwarka and Brahmni rivers in Moribund deltaic India. In order to explain the linkages 16 tie channels' change depicting variables and seven wetland area and hydrological transformation (consistency, hydro-period, water depth, water richness etc.) related variables were taken. Ordinary least square (OLS) regression was applied for explain the linkage. Machine learning approaches were applied for water richness mapping. Tie channel evolution was digitized from the Survey of India (SOI) toposheet (1974) and historical Google Earth images. The result revealed that, the tie channel witnessed morphological and hydrological degradations like channel constriction, channel clogging, flow lowering etc. and these were further identified as some vectors of areal shrinkage and growing hydrological insecurity in the linked wetlands. The total wetland area declined from 44.89 km2 to 16.86 km2 from 1991 to 2021. The rate of areal loss, shallowing WD, growing inconsistency of water presence, narrowing HP, and weakening WR was found high in recent times due to de-linking and degradations of tie channels. This approach of explanation of hydrological insecurity of wetland in relation to tie channel degradations, de-linking and findings are quite unique. So, re-linking and morphological correction of the tie channels would be a good policy for wetland restoration. • The floodplain wetland area was reduced by 37.55% between 1991 and 2021. • Tie channels were either lost or degraded losing sufficient water-carrying capacity. • Consistency, hydro-period, wetland depth declined in relation to tie channel loss. • De-linking of tie channel deteriorated water richness zone in wetlands. • Re-linking of tie channel could improve the hydrological condition of the wetland. [ABSTRACT FROM AUTHOR]

There have been research attempts to automate the process of lithology identification from the well-logs using machine learning algorithms. However, a single algorithm cannot be applied on a universal scale due to the nonstationary nature of geological data sets. The notion of changes in the well-log signal must be considered to resolve the issue. This paper presents a multistage method using multivariate change detection and statistical machine learning to (a) identify the significant changes in the well-logs, (b) map multivariate change in log signal with the lithofacies classes, (c) enhance the data set by removing the highly populated class, (d) using statistical machine learning tools, i.e., Support Vector Machine (SVM) to classify target classes. We apply the algorithm to the data collected from a mostly shale-contained subsurface environment in the Krishna-Godavari (KG) basin, India. The primary objective of the research is to locate thin layers of hydrocarbon-contained sand in the subsurface while dealing with challenges such as intra-well and inter-well heterogeneity and a severe class imbalance. The algorithm satisfactorily performs in lithofacies classification with significantly better performance than the standard machine learning algorithm. • Lithofacies identification using a multi-stage change detection approach • Lithofacies identification in a non-stationary subsurface • Lithofacies classification in an unbalanced condition [ABSTRACT FROM AUTHOR]

Large-scale tree planting has been advocated for decades in many countries as a cost-effective strategy to mitigate carbon, rejuvenate degraded landscapes, and support local livelihoods. Recent studies, however, suggest limited ecological and livelihood impacts of tree planting programs, indicating possible limits to tree planting as a natural climate solution. In this paper, we explore these carrying capacity-like limits by evaluating the site suitability of forestry landscapes of the northern Indian state of Himachal Pradesh using theory/expert-based rules and machine learning algorithms. We find that the state can only achieve a maximum of 31.54% forest cover due to socio-ecological and biophysical constraints, much less than India's goal of bringing 66% of the total geographical area of Himalayan states into forest cover. The low availability of suitable lands for growing trees, together with poor plantation site selection by forest range officers limit the use of tree planting as a sole climate mitigation strategy. To approach theoretical planting limits, we propose an ePSA (e-Plantation Site Assistant) recommendation system based on our site-suitability results to assist forest rangers in selecting suitable sites for planting trees. An initial deployment of the recommender system suggests its potential utility in shaping the long-term success of tree plantations as an effective carbon strategy in northern India and beyond. Overall, we argue that there is a need to realign over-ambitious national and international tree planting targets with actual limits from site characteristics to avoid massive wastage of funds and to obtain feasible carbon mitigation outcomes. [ABSTRACT FROM AUTHOR]

• Feature engineering based framework for location-sensitive disaster event detection. • Augmentation of context-free and context-based features with place of occurrence. • Evaluation with unsupervised machine learning algorithm with 09 performance metrics. • AHP-TOPSIS based selection of an efficient machine learning algorithm and feature set. • Location-augmented context-based features outperformed traditional textual features. Disaster event detection aims to identify events like terrorist attacks, fire incidents, stampede incidents, building collapse, etc., reported in the online news articles or social media. Place of occurrence of disaster event is a significant feature associated with events for location-sensitive disaster event detection. Efficient feature selection and their augmentation with location information can contribute towards the evolution of traditional approaches and their adoption for location-sensitive disaster event detection leading to improvement in the overall process as a whole. Since the evaluation of event detection techniques deliberates various intrinsic and extrinsic performance metrics, the decision-making for the selection of feature sets is treated as a Multiple-Criteria Decision Making (MCDM) problem. This paper proposes a framework, GeoClust , that is based on feature engineering of traditional textual features in order to enhance their capability for improved location-sensitive disaster event detection. The framework augments context-free and context-based textual feature sets with feature sets of place of occurrence of the events and evaluates their performance using unsupervised machine learning algorithms for various performance metrics. Finally, the best feature set is selected using AHP-TOPSIS technique of MCDM in order to tune the system for automatic and efficient location-sensitive disaster event detection in real-time. Extensive set of experiments have been performed in order to evaluate the framework on a dataset of online news articles reporting disaster events about terrorist attacks, fire incidents, stampede incidents, building collapse and maoist attacks happened at different locations in India. The results show that the location-augmented feature sets significantly improve performance of location-sensitive disaster event detection as compared with traditional feature sets. The results also demonstrate that the context-based feature sets with location-augmentation are ranked higher than the context-free feature sets in MCDM analysis. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic has dramatically altered global energy consumption, particularly affecting investment in renewable energy projects. In India, strict shelter-in-place orders enforced during March 2020 have since led to a considerable change in public and private sector investments in planned renewable energy installations. In this paper, we attempt to highlight trends in energy consumption and installed renewable energy capacity noted in India during a period concurrent with the shelter-in-place orders. We discuss recent policy measures and additions to installed renewable energy capacity, and propose key policy recommendations that may help the sector adopt a growth trajectory similar to one noted pre-pandemic. This paper is organized into four main parts. In the first section, we draw focus to India's renewable energy policies and pay special emphasis on recent interventions and campaigns targeted towards achieving high growth rates in the sector. We briefly discuss the need for effective public-private partnerships in order to meet these targets. In the second part, we quantitatively characterise the growth of renewables in India. We present an overview of several mechanisms and missions the government has launched in line with their policy to mitigate the environmental impact of India's energy mix. In the third part, we analyse the decrease in electricity demand in India from 24 March to 30 June 2020, a period concurrent with shelter-at-home orders issued by the Government. We also characterise changes in installed renewable energy capacity between March to December 2016–2020 to provide causal evidence of the effect of the pandemic on the growth of renewables. In this section, we also compile and analyse data on state-wise stressed assets across renewable energy generators in the country. Lastly, in the fourth and final portion of this paper, we highlight policy recommendations that may help the sector overcome logistical and financial bottlenecks in the short-term. We do this with the hope of outlining key measures that decision makers may employ to achieve pre-COVID sectoral growth in the long term. Our recommendations cover three different policy instruments: investment subsidies, operational subsidies, and recommendations for DISCOMs. • The increase in Indian renewable energy capacity installations between February to May 2020 hasseen an increment of 0.75%. • Daily electricity demand has decreased by 18.8% after the announcement of the shelter at home orders. • Immediate effects of the pandemic on the Indian renewable energy sector have been quantified. • A 3 tier system may be adopted to regulate incentives among renewable energy power producers. • Key policy recommendations to improve the fitness of the renewable energy sector have been discussed. [ABSTRACT FROM AUTHOR]

• Urban freight flow estimation for developing urban settings like India. • Integration of direct flow estimation models with origin destination synthesis model. • Spatio-temporal Kriging and machine learning direct flow estimation models performance comparison. • Demonstration of road link speed estimation using map application programming interfaces. This paper presents an innovative methodological framework for determining urban freight flows using a multimodal origin–destination synthesis (ODS) model, which we ran with a novel combination of primary and secondary data. Primary data included classified traffic volume counts at limited locations, whereas secondary data, enriched with certain techniques, came from (a) map application programming interfaces used to extract real-time speeds and calibrate a disaggregated speed–volume relationship and (b) advanced techniques for estimating direct flow models (e.g. spatio-temporal Kriging and machine learning models). The paper also discusses the sensitivity of multimodal ODS models to variations in base–seed matrices. [ABSTRACT FROM AUTHOR]

• Shot-term reservoir inflow prediction accounting for hydrological and climate indices. • ML-based Ensemble methods to quantify forecasting uncertainty. • The Ensemble model performed better than the standalone ML models. • Two reservoirs of India and USA were chosen to test the robustness of ML models. Short and long range reservoir inflow forecast is essential for efficient real time operational planning, scheduling of hydroelectric power system and management of water resources. Large-scale climate phenomenon indices have a strong influence on hydrological processes under complex weather conditions, and it should be considered to forecast reservoir inflow for efficient dam operation strategies. This study aims to explore the relevance of large-scale climate phenomenon indices in improving the reservoir inflow prediction at short-term time scales. This paper presents a simple and effective framework to combine various data-driven machine learning (ML) algorithms for short-range reservoir inflow forecasting. Random Forest (RF), Gradient Boosting Regressor (GBR), K-Nearest Neighbors Regressor (KNN), and Long Short-Term Memory (LSTM) were employed for predicting daily reservoir inflows considering various climate phenomenon indices (e.g., Arctic Oscillation, North Atlantic Oscillation, and Southern Oscillation Index) and hydroclimatic variables (precipitation), accounting for time-lag effects. After training the individual ML algorithm, a framework was developed to create an ensemble model using a robust weighted voting ensemble method to quantify forecasting uncertainty and to improve the model performance by combining the inflow results of the single ML model and the highest vote is chosen based on the weights assigned to the single ML model. The developed framework was examined in two distinct reservoirs located in India and California, USA. The ensemble model consistently outperformed the standalone RF, GBR, KNN, and LSTM in predicting high (flood control and monsoon seasons) and low flows (runoff and non-monsoon seasons) of both study reservoirs. The demonstrated short-term reservoir forecasting model allows reservoir operators to adapt and add regional hydrological and large-scale climate indices in real-time decision-making. The presented framework can be applied for any reservoir inflow forecasting. [ABSTRACT FROM AUTHOR]

Coal exploration in the Indian scenario is challenging due to plenty of carbon contents and dirt bands within a coal seam. Manual interpretation of geophysical logging data in such conditions is time-consuming and tedious due to the non-linear behaviors of well log signals. This paper intends to apply supervised machine learning techniques such as Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Multi-Layer Perceptron (MLP), and Extreme Gradient Boosting technique (XGBoost) for meticulously interpreting such banded coal seams from geophysical logs. To investigate the efficacy of the above-mentioned five ML techniques in the Indian scenario, we have considered Gamma-ray, Density, and Resistivity logs from four boreholes drilled in Talcher coalfield, Eastern India. ML model training results indicate that all the prediction models have scored more than 88% accuracy scores in classifying carbonaceous and non-coal lithofacies. Receiver Operating Characteristics (ROC) curves of different ML models suggest that the obtained area under the curve (AUC) is positive and above the main diagonal for all lithotypes. Finalized ML models with appropriate hyper-parameters are also applied on the nearby drilled wells, and the outputs are validated against geological core data. It is found that all ML models have acquired an overall accuracy score greater than 80% in all three test wells, conveying that ML techniques are a potential solution for dealing with banded coal seam problems in the Indian scenario. • Lithology prediction from Well logs using Machine learning (ML) algorithms. • All ML models have secured accuracy scores more than 88% in the training phase. • Predicted lithology from ML algorithms is validated with geological core data. • ML models have obtained more than 80% accuracy scores in all three test wells. • ML techniques are the potential solutions for tackling banded coal seam problems. [ABSTRACT FROM AUTHOR]

The agricultural sector in India accounts for a significant part of the country's GDP and is the primary income source for many farmers in rural areas. While it creates employment opportunities and offers food security for the entire nation, the lack of infrastructure and resources might be limiting its potential to thrive further. One of the aspects addressed in this paper is low yield production. With the aid of a sensor-based irrigation model, data is collected and analyzed in the cloud to enable real-time monitoring. It is then integrated with an Android application for displaying results in an user-friendly interface. Through the application, farmers can control the farm manually, or with a timer in minutes. The Machine Learning model predicts the suitable crops, in accordance with varying weather parameters. The application has a classified portal for farmers and customers to buy/sell directly, eliminating any involvement of mediators. One of the novelties in this research includes monitoring/controlling farm equipment and predicting field crops from a locally installed LCD display and keypad present in farmer's respective homes. The proposed work aims to create an energy-efficient, user-friendly framework for the agricultural workforce, yielding better crop production, improving farmers' living standards, and contributing effectively to the nation's economic growth. The prototype shows a reduction of water usage in fields by more than 60%. In order to incorporate the model with the best behavior in Android Application, different Machine Learning algorithms have been studied, among which Random Forest has been selected with a test accuracy of 91.59%. [ABSTRACT FROM AUTHOR]

• Stuying DSM products for spatializing the STICS model (simulated yield and drainage) • Comparisons of STICS outputs using observed vs DSM estimated STICS soil parameters. • DSM products outperformed the soil map for providing the STICS soil parameters. • Few yearly and the all-year-aggregated simulations were impacted by DSM errors. • Coupling DSM with a crop model: an interesting solution for Digital Soil Assessment. Digital Soil Mapping (DSM) can be an alternative data source for spatializing crop models over large areas. The objective of the paper was to evaluate the impact of DSM products and their uncertainties on a crop model's outputs in an 80 km2 catchment in south India. We used a crop model called STICS and evaluated two essential soil functions: the biomass production (through simulated yield) and water regulation (via calculated drainage). The simulation was conducted at 217 sites using soil parameters obtained from a DSM approach using either Random Forest or Random Forest Kriging. We first analysed the individual STICS simulations, i.e., at two cropping seasons for 14 individual years, and then pooled the simulations across years, per site and crop season. The results show that i) DSM products outperformed a classical soil map in providing spatial estimates of STICS soil parameters, ii) although each soil parameters were estimated separately, the correlations between soil parameters were globally preserved, ii) Errors on STICS' yearly outputs induced by DSM estimations of soil parameters were globally low but were important for the few years with high impacts of soil variations, iii) The statistics of the STICS simulations across years were also affected by DSM errors with the same order of magnitude as the errors on soil inputs and iv) The impact of DSM errors was variable across the studied soil parameters. These results demonstrated that coupling DSM with a crop model could be a better alternative to the classical Digital Soil Assessment techniques. As such, it will deserve more work in the future. [ABSTRACT FROM AUTHOR]

Land Use Land Cover (LULC) change assessment and prediction are essential for optimised water resources planning and management. This paper attempts to intercompare the different LULC change modelling techniques (two-hybrid and two traditional models) to predict the future LULC change. These include Multi-Layer Perceptron Markov Chain (MLP_MC), Logistic Regression-Markov Model (LR-MC), and two hybrid models, namely, Multi-Layer Perceptron Markov Chain Cellular Automata (MLP_MC_CA), Logistic Regression Markov Model Cellular Automata (LR_MC_CA) models. These models were tested on Nagavali River Basin (NRB), a river basin in Southern India, which has seen significant land-use changes over the past two decades. Over the past two decades, the study region experienced dominant changes on the order of 17.42% and 15.22% decrease in scrubland and forest, respectively. At the same time, the agricultural land cover is increased by 35.28%. For the LULC prediction, the model was initially trained using the relevant driver variables and LULC maps of 2010 and 2015. The calibrated model was validated using the 2020 LULC map. The statistical results in terms of Kappa values and chi-square results reveal that the hybrid model MLP-MC-CA has a better agreement (Kappa coefficient 0.902) compared to the other models. Further, it is also observed that the CA-based models have a better ability to capture spatial connections. After combining the MLP_MC model with the Cellular Automata, the former model was improved by 10.8% in terms of the overall Kappa coefficient. The best model for LULC prediction over the next decade (LULC map of 2030) showed that the forest area would decrease by 9.02%, and the agricultural land would increase by 8.74%. Further, the results from the study indicate that the hybrid machine learning models provide a promising alternative for land-use change prediction. • Intercomparison of different Machine Learning Methods for Land Use Land Cover Prediction. • Hybrid model based on MLP_MC_CA outperforms other non-hybrid and traditional methods. • LULC 2030 for NRB showed that the forest area would decrease by 9.02%, and the agricultural land would increase by 8.74%. [ABSTRACT FROM AUTHOR]

Tropical cyclone prediction is essential to limit death toll and damage caused by them. In this paper, a model has been formulated to classify the cyclone as no cyclone, minimal, moderate, extensive, extreme and catastrophic in view of attributes such as Wind Speed(10 m), Rainfall, Wind Direction(10 m), Sea Surface Temperature(2 m), Sea Level Pressure(2 m) and Relative humidity(2 m). The models have been trained using the meteorological data from MERRA-2 Web service which provides time-series data with spatial resolution of approximately 50 km. The cyclone data is taken from RSMC — New Delhi for Tropical cyclone disturbances over Indian Oceans. Initially, the models are trained using deep learning networks like MLP, LSTM, GRU, RNN, BI-LSTM and CNN. Since CNN gives better results, the CNN model is chosen for further analysis. The hyper parameters of the CNN model are optimized using genetic algorithm. The values drawn from genetic algorithm appear to be promising than the values which were chosen manually in random. The model is then modified by removing the fully connected layer which operates as a classifier in CNN network. The conventional machine learning classifiers like — Decision Tree, K-Nearest neighbor, logistic regression, Naive Bayes, Random Forest, SVM and XGBoost are used as a classifier in the place of a fully connected layer in CNN. Further to increase the prediction accuracy, C4.5 Decision tree algorithm is modified to be used as a classifier in CNN. Classification is performed by considering the Spatio-temporal data of various important cities in India. The model was tested to classify the category for 5 different cyclones and was also compared with Saffir–Simpson's scale to validate the correctness of the model. The proposed model gives a better performance compared to the conventional machine learning and deep learning classifiers in terms of time complexity, accuracy, precision and recall and this can supplement the cyclone prediction process of current NWP models. • Cyclone Prediction is proposed using CNN, modified C4.5 and genetic algorithms. • CNN is used for feature extraction and genetic algorithm for hyper parameter tuning. • Modified C4.5 is proposed to solve the limitations of C4.5 decision tree algorithm. • Accuracy of the proposed model is significantly improved compared to the existing works. [ABSTRACT FROM AUTHOR]