Computational Intelligence Techniques in Earth and Environmental Sciences

Computational intelligence is a group of computational models and tools that encompass elements of learning, adaptation, and/or heuristic optimization. It is used to help study problems that are difficult to solve using conventional computational algorithms. Neural networks, evolutionary computation, and fuzzy systems are the three main pillars of computational intelligence. More recently, emerging areas such as swarm intelligence, artificial immune systems (AIS), support vector machines, rough sets, chaotic systems, and others have been added to the range of computational intelligence techniques. This chapter aims to present an overview of computational intelligence techniques and their applications, focusing on five representative techniques, including neural networks, evolutionary computation, fuzzy systems, swarm intelligence, and AIS.

Temperature is one of the factors of climate variables and understanding its nature is very important because the effect of temperature on climate change is higher than that of other variables. The purpose of the present study was to build an appropriate model to forecast the monthly maximum temperature of Rajshahi district in Bangladesh. The Box–Jenkins modeling strategy was performed using EViews software. This strategy was performed using the Augmented Dickey–Fuller, Phillip–Perron, Kwiatkowski–Phillips–Schmidt–Shin, autocorrelation function, partial autocorrelation function, ordinary least square method, normal P–P plot, Chow’s breakdown test, Chow’s forecast test, and standardized residuals plot. Seasonal variation, cyclical variation, and a slightly upward trend over time were found in the temperature. The temperature was found to be stationary at level after removing the cyclical variation using the resistant smoothing method, 4253H-twice. The SARMA(2, 1)(1, 2)₁₂ model was found to be the most appropriate model for forecasting. The fitted model is also stable with no structural change and thus applicable for forecasting and policy purposes. Finally, this model was used for forecasting maximum temperature from January 2010 to December 2015. The forecasted value divulged that the maximum temperature will be increased by 3 °C during 2010–2015. This is an alarming situation for the environment and should take initiative to control and save our environment of Rajshahi district in Bangladesh.

Estimation of heat stress based on WBGT (wet bulb globe temperature) index is widely accepted as international standard. The purpose of the present study was to provide tolerance limit for people and interventions required to protect individuals from the dangerous consequences of heat. The meteorological data collected from Indian Meteorological Department of Ahmedabad (2001–2011) was used for estimating the WBGT. Multiple regression analysis was used to explore relationship between variables dry bulb temperature (T _a), wet bulb temperature (T _wb), and globe temperature (T _g) across the districts varied widely in two different seasons, i.e., summer and winter months. The linear regression analysis was applied for the purpose of future prediction, with respect to the WBGT index, and heat tolerance limit and visualized using GIS tool. The average tolerance time for 2001–2011 arrived at 82 ± 16 and 159 ± 36 min for the months of summer and winter, respectively. Thus, the WBGT and tolerance limit maps might prevail working population from heat stress fury.

In the semiarid Barind region, episodes of agricultural droughts of varying severity have occurred. The occurrence of these agricultural droughts is associated with rainfall variability and can be reflected by soil moisture deficit that significantly affects crop performance and yield. In the present study, an analysis of long-term (1971–2010) rainfall data of 12 rain monitoring stations in the Barind region was carried out using a Markov chain model which provides a drought index for predicting the spatial and temporal extent of agricultural droughts. Inverse distance weighted interpolation was used to map the spatial extent of drought in a GIS environment. The results indicated that in the Pre-Kharif season drought occurs almost every year in different parts of the study area. Though occurrence of drought is less frequent in the Kharif season the minimum probability of wet weeks leads to reduction in crop yields. Meanwhile, the calculation of 12 months drought suggests that severe to moderate drought is a common phenomenon in this area. Drought index is also found to vary depending on the length of period. The return period analysis suggests that chronic drought is more frequent in the Pre-Kharif season and the frequency of moderate droughts is higher in the Kharif season. On the contrary severe drought is more frequent for a 12-month period.

Bangladesh frequently suffers from tropical cyclones possibly due to its unique location. The funnel-shaped northern part of the Bay of Bengal causes tidal bores when cyclones make landfall. These tropical cyclones can be very devastating and can severely affect the coastline of Bangladesh. In this study we analyzed 135 tropical cyclones occurred in Bangladesh during 1877–2009 considering the physical characteristics of the storm surge process. For analyzing the storm surge process, a Markov renewal model that takes into account both the sojourn times and the transitions between different types of cyclones simultaneously was considered. Exponential distribution for the sojourn times was assumed to derive the probabilities of occurrence of different types of cyclones for various lengths of time intervals. Given the type of the last cyclone occurred probabilities of occurrence of the next cyclone are reported using the fitted Markov renewal model. The mean recurrence times of different type of cyclones were also calculated assuming ergodicity of the Markov renewal process.

An attempt has been made to show whether the recently developed wavelet transformation in forecasting the climatic time series in Bangladesh improves the performance of existing forecasting models, such as ARIMA. These models are applied to forecast the humidity of Rajshahi, Bangladesh. Then the wavelet transformation has been used to decompose the humidity series into a set of better-behaved constitutive series. These decomposed series and inverse wavelet transformation are used as a pre-processing procedure of forecasting humidity series using the same models in two approaches. Finally, the forecasting ability of these two models with and without wavelet transformation is compared using the statistical forecasting accuracy criteria. The results show that the use of wavelet transformation as a pre-processing procedure of forecasting climatic time series improves the performance of forecasting models. The reason is the better behavior of the constitutive series for the filtering effect of the wavelet transform.

This chapter presents a data adaptive filtering technique to extract annual cycles and the analysis of inter-annual climate variability based on different climate signals using discrete wavelet transform (DWT). The annual cycle is considered as higher energy trend in a climate signal and separated by implementing a threshold-driven filtering technique. The fractional Gaussian noise (fGn) is used here as a reference signal to determine adaptive threshold without any prior training constraint. The climate signal and fGn are decomposed into a finite number of subband signals using the DWT. The subband energy of the fGn and its confidence intervals are computed. The upper bound of the confidence interval is set as the threshold level. The energy of individual subband of a climate signal is compared with the threshold. The lowest order subband of which the energy is greater than the threshold level is selected yielding the upper frequency limit of the trend representing annual cycle. All the lower frequency subbands starting from the selected one are used to reconstruct the annual cycle of the corresponding climate signal. The distance between adjacent peaks in the extracted cycles refers to the inter-annual variation of the climate condition. The experimental results illustrate the efficiency of the proposed data adaptive approach to separate the annual cycle and the quantitative analysis of climate variability.

This chapter proposes fuzzy genetic approach so as to predict suspended sediment concentration (SSC) carried in natural rivers for a given stream cross section. Fuzzy genetic models are improved by combining two methods, fuzzy logic and genetic algorithms. The accuracy of fuzzy genetic models was compared with those of the adaptive network-based fuzzy inference system, multilayer perceptrons, and sediment rating curve models. The daily streamflow and suspended sediment data belonging to two stations, Muddy Creek near Vaughn (Station No: 06088300) and Muddy Creek at Vaughn (Station No: 06088500), operated by the US Geological Survey were used as case studies. The root mean square errors and determination coefficient statistics were used for evaluating the accuracy of the models. The comparison results revealed that the fuzzy genetic approach performed better than the other models in the estimation of the SSC.

Bed sill local scour is an important issue in environmental and water resources engineering in order to prevent degradation of river bed and save the stability of grade-control structures. This chapter presents genetic algorithms (GA), gene expression programming, and M5 decision tree model as an alternative approaches to predict scour depth downstream of bed sills. Published data were compiled from the literature for the scour depth downstream of sills. The proposed GA approach gives satisfactory results (R ² = 0.96 and RMSE = 0.442) compared to existing predictors.

The inherent complexities in hydrologic phenomena have been turned into a barrier to get accurate prediction by conventional linear methods. Therefore, there is an increasing interest toward data-driven black box models. In recent decades artificial neural network (ANN) as a branch of artificial intelligence method has proved its efficiency in providing accurate results to model hydrologic processes, which subsequently leads to provide important information for the urban and environmental planning, land use, flood, and water resources management. The efficiency of any data-driven model (e.g., ANN) largely depends on quantity and quality of available data; furthermore, the occult noises in data may impact the performance of the model. Although ANN can capture the underlying complexity and nonlinear relationship between input and output parameters, there might be a need to preprocess data. In this way, noise reduction of data using an appropriate de-noising scheme may lead to a better performance in the application of the data-driven ANN model. Thereupon, in this chapter, the ANN-based hydrological models (i.e., stream-flow and sediment) were developed by focusing on wavelet-based global soft thresholding method to de-noise hydrological time series on the daily scale. The appropriate selection of decomposition level and mother wavelet type is effective in thresholding results, so that sensitivity analysis was performed over levels and several Daubechies group mother wavelets (Haar, db2, db3, db4, and db5) to choose the proper variables. In this way, de-noised time series were imposed into an ANN model to forecast flow discharge and sediment values. The comparison of obtained results for both single ANN-based and de-noised-based (i.e., preprocessed) approaches revealed that the outcomes have been improved for the later model. Furthermore, the consequences indicated that the wavelet de-noising was significantly dependent on the chosen mother wavelet whereas forecasting results varied obviously with the alteration of mother wavelets. Eventually, it was resulted that after a specific threshold, no eminent progress in results was obtained unlike the reduction occurred. Overall, the wavelet-based de-noising approach, as a preprocessing method, can be a promising idea to improve the ANN-based hydrological models.

Conventional error-based statistical parameters like the Nash–Sutcliffe efficiency index are popular among hydrologists to check the accuracy of hydrological models and to compare the relative performance of alternative models in a particular modelling scenario. A major drawback of those traditional indices is that they are based on only one modelling attribute, i.e. the modelling error. This study has identified an overall model utility index as an effective error-sensitivity-uncertainty procedure which could serve as a useful quality indicator of data-based modelling. This study has also made an attempt to answer the question—should the increasing complexity of the existing model add any benefit to the model users? The study evaluates the utility of some popular and widely used data-based models in hydrological modelling such as local linear regression, artificial neural networks (ANNs), Adaptive neuro fuzzy inference system (ANFIS) and support vector machines (SVMs) along with relatively complex wavelet hybrid forms of ANN, ANFIS and SVM in the context of daily rainfall–runoff modelling. The study has used traditional error-based statistical indices to confirm capabilities of model utility index values in identifying better model for rainfall–runoff modelling. The implication of this study is that a modeller may use utility values to select the best model instead of using both calibration and validation processes in the case of data scarcity. The study comprehensively analysed the modelling capabilities of SVM and its waveform in the context of rainfall–runoff modelling.