Machine learning algorithm-based risk assessment of riparian wetlands in Padma River Basin of Northwest Bangladesh
02 Mar 2021-Environmental Science and Pollution Research (Springer)-Vol. 28, Iss: 26, pp 34450-34471
TL;DR: In this paper, the authors explored the spatiotemporal dynamics of wetlands, prediction of wetland risk assessment, and showed that wetland areas at present are declining less than one-third of those in 1988 due to the construction of the dam at Farakka, which is situated at the upstream of the Padma River.
Abstract: Wetland risk assessment is a global concern especially in developing countries like Bangladesh. The present study explored the spatiotemporal dynamics of wetlands, prediction of wetland risk assessment. The wetland risk assessment was predicted based on ten selected parameters, such as fragmentation probability, distance to road, and settlement. We used M5P, random forest (RF), reduced error pruning tree (REPTree), and support vector machine (SVM) machine learning techniques for wetland risk assessment. The results showed that wetland areas at present are declining less than one-third of those in 1988 due to the construction of the dam at Farakka, which is situated at the upstream of the Padma River. The distance to the river and built-up area are the two most contributing drivers influencing the wetland risk assessment based on information gain ratio (InGR). The prediction results of machine learning models showed 64.48% of area by M5P, 61.75% of area by RF, 62.18% of area by REPTree, and 55.74% of area by SVM have been predicted as the high and very high-risk zones. The results of accuracy assessment showed that the RF outperformed than other models (area under curve: 0.83), followed by the SVM, M5P, and REPTree. Degradation of wetlands explored in this study demonstrated the negative effects on biodiversity. Therefore, to conserve and protect the wetlands, continuous monitoring of wetlands using high resolution satellite images, feeding with the ecological flow, confining built up area and agricultural expansion towards wetlands, and new wetland creation is essential for wetland management.
Citations
More filters
TL;DR: In this paper, four standalone methods such as additive regression (AR), M5P tree model (M5P), random subspace (RSS), and support vector machine (SVM) were employed to predict WQI based on variable elimination technique.
Abstract: Data-driven models are important to predict groundwater quality which is controlling human health. The water quality index (WQI) has been developed based on the physicochemical parameters of water samples. In this area, water quality is medium to poor and is found in saline zones; very high pH ranges are directly affected on the water quality in this study area. Conventional WQI computation demands more time and is often observed with enormous errors during the calculation of sub-indices. In the present work, four standalone methods such as additive regression (AR), M5P tree model (M5P), random subspace (RSS), and support vector machine (SVM) were employed to predict WQI based on variable elimination technique. The groundwater samples were collected from the Akot basin area, located in the Akola district, Maharashtra, in India. A total of nine different input combinations were developed in this study. The datasets were demarcated into two classes (ratio 80:20) for model construction (training dataset) and model verification (testing dataset) using a fivefold cross-validation approach. The models were assessed using statistical and graphical appraisal metrics. The best input combinations varied among the model, generally, the optimal input variables (EC, pH, TDS, Ca, Mg, and Cl) during the training and validation stages. Results show that AR outperformed the other data-driven models (R2 = 0.9993, MAE = 0.5243, RMSE = 0.0.6356, %RAE = 3.8449, and RRSE% = 3.9925). The AR is proposed as an ideal model with satisfactory results due to enhanced prediction precision with the minimum number of input parameters and can thus act as the reliable and precise method in the prediction of WQI at the Akot basin.
34 citations
TL;DR: In this paper , the authors selected the Google Earth Engine platform to select 100m resolution PROBA-V remote sensing images from 2018 of Zambia, in central Africa, in order to conduct grassland resource surveys for the scientific management of grassland resources.
32 citations
TL;DR: In this article, a hybrid adaptive neuro-fuzzy inference system (ANFIS-WCAMFO) was proposed for predicting reference evapotranspiration (ETo) in south-central Bangladesh.
29 citations
TL;DR: Li et al. as mentioned in this paper proposed a new hybrid model based on the convolutional neural network (CNN) for making effective use of historical datasets and producing a reliable landslide susceptibility map, which is capable of accurately mapping landslide susceptibility and providing a promising method for hazard mitigation and land use planning.
Abstract: This study proposed a new hybrid model based on the convolutional neural network (CNN) for making effective use of historical datasets and producing a reliable landslide susceptibility map. The proposed model consists of two parts; one is the extraction of landslide spatial information using two-dimensional CNN and pixel windows, and the other is to capture the correlated features among the conditioning factors using one-dimensional convolutional operations. To evaluate the validity of the proposed model, two pure CNN models and the previously used methods of random forest and a support vector machine were selected as the benchmark models. A total of 621 earthquake-triggered landslides in Ludian County, China and 14 conditioning factors derived from the topography, geological, hydrological, geophysical, land use and land cover data were used to generate a geospatial dataset. The conditioning factors were then selected and analyzed by a multicollinearity analysis and the frequency ratio method. Finally, the trained model calculated the landslide probability of each pixel in the study area and produced the resultant susceptibility map. The results indicated that the hybrid model benefitted from the features extraction capability of the CNN and achieved high-performance results in terms of the area under the receiver operating characteristic curve (AUC) and statistical indices. Moreover, the proposed model had 6.2% and 3.7% more improvement than the two pure CNN models in terms of the AUC, respectively. Therefore, the proposed model is capable of accurately mapping landslide susceptibility and providing a promising method for hazard mitigation and land use planning. Additionally, it is recommended to be applied to other areas of the world.
22 citations
TL;DR: In this article, the influence of wetland fragmentation due to damming on wetland water richness and the impact of changes in water richness on the ecosystem service value (ESV) of the wetlanddominated rivers of the lower Punarbhaba Basin, India, and Bangladesh, as the case.
Abstract: Evaluation of the importance of ecosystem services (ES) of various wetlands is well reported with global and regional level research, but the degree to which spatial-temporal variations in water richness (availability of water) have had an effect on ES has not yet been examined. The present work is intended to investigate the influence of wetland fragmentation due to damming on wetland water richness and the impact of changes in water richness on the ecosystem service value (ESV) of the wetland-dominated rivers of the lower Punarbhaba Basin, India, and Bangladesh, as the case. Water richness models of pre- and post-dam periods have been constructed based on four hydro-ecological parameters (hydro-period, depth of water, consistency of water appearance, and wetland size) following the semi-quantitative analytic hierarchy process (AHP). ESV of different wetland types, with and without considering water richness effect, has been computed. The result indicates that the overall wetland area decreased from 73,563 to 52,123 km2 during the post-dam period. Approximately 53.8% of the high water-rich region is decreased. Total wetland ESV has been lowered by 63.4% from 1989 to 2019, with an average reduction rate of 2%. This is mainly due to the squeezing of the wetland area during the post-dam period. If the impact of water richness on ESV is considered, the scenario is found to be very distinct. Total ESV of various ESV areas amounted to $33 million during the pre-dam period and is reduced to $19.71 million during the post-dam period. If compared to the total ESV of the wetland without considering the effect of water richness, the calculated ESV gap was $105 million in pre-dam and $38 million in post-dam period indicating a widening of the gap. Maintaining the ES of wetland hydrological management, specifically the flow maintenance of river and riparian wetlands, is essential.
15 citations
References
More filters
[...]
TL;DR: Internal estimates monitor error, strength, and correlation and these are used to show the response to increasing the number of features used in the forest, and are also applicable to regression.
Abstract: Random forests are a combination of tree predictors such that each tree depends on the values of a random vector sampled independently and with the same distribution for all trees in the forest. The generalization error for forests converges a.s. to a limit as the number of trees in the forest becomes large. The generalization error of a forest of tree classifiers depends on the strength of the individual trees in the forest and the correlation between them. Using a random selection of features to split each node yields error rates that compare favorably to Adaboost (Y. Freund & R. Schapire, Machine Learning: Proceedings of the Thirteenth International conference, aaa, 148–156), but are more robust with respect to noise. Internal estimates monitor error, strength, and correlation and these are used to show the response to increasing the number of features used in the splitting. Internal estimates are also used to measure variable importance. These ideas are also applicable to regression.
79,257 citations
TL;DR: High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated and the performance of the support- vector network is compared to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Abstract: The support-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.
High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
37,861 citations
TL;DR: In this paper, an approach to synthesizing decision trees that has been used in a variety of systems, and it describes one such system, ID3, in detail, is described, and a reported shortcoming of the basic algorithm is discussed.
Abstract: The technology for building knowledge-based systems by inductive inference from examples has been demonstrated successfully in several practical applications. This paper summarizes an approach to synthesizing decision trees that has been used in a variety of systems, and it describes one such system, ID3, in detail. Results from recent studies show ways in which the methodology can be modified to deal with information that is noisy and/or incomplete. A reported shortcoming of the basic algorithm is discussed and two means of overcoming it are compared. The paper concludes with illustrations of current research directions.
17,177 citations
01 Jan 2007
TL;DR: random forests are proposed, which add an additional layer of randomness to bagging and are robust against overfitting, and the randomForest package provides an R interface to the Fortran programs by Breiman and Cutler.
Abstract: Recently there has been a lot of interest in “ensemble learning” — methods that generate many classifiers and aggregate their results. Two well-known methods are boosting (see, e.g., Shapire et al., 1998) and bagging Breiman (1996) of classification trees. In boosting, successive trees give extra weight to points incorrectly predicted by earlier predictors. In the end, a weighted vote is taken for prediction. In bagging, successive trees do not depend on earlier trees — each is independently constructed using a bootstrap sample of the data set. In the end, a simple majority vote is taken for prediction. Breiman (2001) proposed random forests, which add an additional layer of randomness to bagging. In addition to constructing each tree using a different bootstrap sample of the data, random forests change how the classification or regression trees are constructed. In standard trees, each node is split using the best split among all variables. In a random forest, each node is split using the best among a subset of predictors randomly chosen at that node. This somewhat counterintuitive strategy turns out to perform very well compared to many other classifiers, including discriminant analysis, support vector machines and neural networks, and is robust against overfitting (Breiman, 2001). In addition, it is very user-friendly in the sense that it has only two parameters (the number of variables in the random subset at each node and the number of trees in the forest), and is usually not very sensitive to their values. The randomForest package provides an R interface to the Fortran programs by Breiman and Cutler (available at http://www.stat.berkeley.edu/ users/breiman/). This article provides a brief introduction to the usage and features of the R functions.
14,830 citations
Book•
01 Dec 1995
TL;DR: Introductory Digital Image Processing: A Remote Sensing Perspective focuses on digital image processing of aircraft- and satellite-derived, remotely sensed data for Earth resource management applications.
Abstract: For junior/graduate-level courses in Remote Sensing in Geography, Geology, Forestry, and Biology. Introductory Digital Image Processing: A Remote Sensing Perspective focuses on digital image processing of aircraft- and satellite-derived, remotely sensed data for Earth resource management applications. Extensively illustrated, it explains how to extract biophysical information from remote sensor data for almost all multidisciplinary land-based environmental projects. Part of the Pearson Series Geographic Information Science. Now in full color, the Fourth Edition provides up-to-date information on analytical methods used to analyze digital remote sensing data. Each chapter contains a substantive reference list that can be used by students and scientists as a starting place for their digital image processing project or research. A new appendix provides sources of imagery and other geospatial information.
5,478 citations