S Angornai; H Memarian; M Shariat Panahi; M.J Bolourchi
Abstract
Land subsidence is an environmental phenomenon that involves gradual or sudden settlement of the land surface because of compaction of underground material. Groundwater withdrawal, which occurs due to excessive use of water resources, is among the most important reasons for this phenomenon. Therefore, ...
Read More
Land subsidence is an environmental phenomenon that involves gradual or sudden settlement of the land surface because of compaction of underground material. Groundwater withdrawal, which occurs due to excessive use of water resources, is among the most important reasons for this phenomenon. Therefore, land subsidence can lead to destructive results in residential, industrial and agricultural areas. As a result, subsidence caused by excessive use of groundwater resources has occurred in many countries in the world. Tehran metropolitan plain in Iran is one of the most obvious examples, where land subsidence is happening. Although the relationship between land subsidence, groundwater level decline and changes in the physical properties of subsurface material is broadly understood, a comprehensive and precise model to predict land subsidence remains unconstrained. Land subsidence modeling is a complicated matter in geological engineering but can help to better understand subsidence and possibly prevent damages. The commonly used numerical methods for modeling land subsidence are generally based on simple assumptions, which make the model results to be associated with some errors. In this study, artificial intelligent methods such as Artificial Neural Networks (ANN) were used to propose a new method to predict land subsidence. The efficiency of this method was then tested in the South Tehran plain as a case study. We have used hydrological, geotechnical, remote sensing and ambient vibrations for site effect investigations. First, the collected data was studied statistically. Then, the delay between groundwater withdrawal and subsidence was computed by genetic algorithms using available hydrographs and GPS data in a period of 27 months. Model input parameters include changes in groundwater level, natural frequency of soil, alluvial thickness, defined geographic coordinates and time. The model output was an estimated subsidence measured by radar interferometry method. The model was built in 15 time steps using a set of data having 4 months of time difference with the data used to create the model. The comparison between the predicted (modeled) and real (measured by remote sensing) subsidence shows a good correlation, which makes the proposed model reliable.
o Memarian Sorkhabi; Y Djamour
Abstract
In order to study the crustal movements in Iran, establishment of several campaign GPS networks in 1998 seriously initiated geodynamical activities. After that in 2005, a network of ~120 permanent GPS stations named Iranian Permanent GPS Network (IPGN) has been installed to complete the campaign GPS ...
Read More
In order to study the crustal movements in Iran, establishment of several campaign GPS networks in 1998 seriously initiated geodynamical activities. After that in 2005, a network of ~120 permanent GPS stations named Iranian Permanent GPS Network (IPGN) has been installed to complete the campaign GPS networks already existing in Iran. Thanks to all campaign and continuous GPS sites, there are many geodetic velocity vectors indicating kinematic behavior of the crust at their positions. Now, the main question is about geodetic velocity for any other arbitrary station. Evidently, the best reliable solution is installing more GPS stations and recording satellite signals, which need considerable cost and time. Another solution, which could be an appropriate alternative, is applying some modern and smart estimation methods such as “Artificial Neural Networks (ANN)”. The main advantages of ANN method are capability learning of networks, parallel processing and computation flexibility. Based on 42 GPS velocity vectors existing in NW Iran, we estimated new velocity vectors for some arbitrary positions in study area by using two estimation methods: “Back Propagation Artificial Neural Networks (BPANN)” and “Collocation”. This estimation was run in 2 models including 2 different reference stations but the same check points. The results from model 1 (with fewer reference points) showed BPANN’s RMSE in E and N components is ±2 mm and ±3.5 mm respectively, which is less than Collocation’s RMSE. The results from model 2 (with more reference points) showed BPANN’s RMSE in E and N components increased to ±1 mm and ±1.5 mm respectively. Therefore, it seems BPANN method could be considered as a good alternative to estimate geodetic velocity field relative to other classical estimation methods.
A. Hezarkhani; P. Tahmasbi; O. Asghari
Abstract
Separation of alteration zones is one of the important processes in evaluation and identification of mining activities that provide great help to have better view of the region and its mineralization. Most of the alteration separation is based on petrological investigations and the other methods are ...
Read More
Separation of alteration zones is one of the important processes in evaluation and identification of mining activities that provide great help to have better view of the region and its mineralization. Most of the alteration separation is based on petrological investigations and the other methods are less applied. Therefore, in this research, there is an attempt by applying RBPNN (Radial Basis Probabilistic Neural Network) to separate these alteration zones. Because of the special structure and easy designing of these networks, they are usually capable to solve the classification problem. The input data were 28 element analyses related to 45 geochemical samples and its outputs were classified alteration zones (potassic, transition, phyllic) that was coding for every inputs data. After selection the training and testing data, the network has been prepared for training and then the data were inputs and the results were outputs. According to the results, the network could distinguish the difficult spatial relation between the inputs, with 28 spatial variables and classify those correctly. The calculated MSE (Mean Square Error) is 0.0163, which shows the good performance of network in this field.
A. Moradzadeh; F. Tahmasbi; M. Fateh
Abstract
The magnetotelluric (MT) method is a natural source electromagnetic geophysical technique, which is used mainly in petroleum, mineral and geothermal exploration. As in this method, the quantity of the measured data is bulky and have a complex structure, their modeling, ...
Read More
The magnetotelluric (MT) method is a natural source electromagnetic geophysical technique, which is used mainly in petroleum, mineral and geothermal exploration. As in this method, the quantity of the measured data is bulky and have a complex structure, their modeling, compared with the modeling of the other electrical data, is a very complex task or even impossible in some instances.
The main objective of this paper is to use the ability of the artificial neural networks (ANN) to find a solution for two-dimensional (2D) joint TE (transverse electric) and TM (transverse magnetic) modes inverse modeling of MT data. To achieve the goal, a multilayer perceptron (MLP) network with back propagation (BP) learning algorithm is used. In order to learn the designed network, many synthetic 2D models with the same category, have been created and their responses have been calculated for each polarization mode by forward modeling. Synthetic data include apparent resistivity and impedance phase in 9 stations and 11 frequencies in two polarization modes. After a comprehensive study, a perceptron with 3 layers and architecture of 396-9-9 has been designed and used to model the data.
This study show that the designed network is capable enough to produce an acceptable 2D underground model so that the correspondence mean relative modeling error is 3.9% and 6.9 % respectively for noise free data and 5 percent randomly added noisy data. This indicates that if ANN is designed and trained properly, then it would be capable enough to perform 2D inverse modeling of MT data. It has also shown that once the designed network has been trained properly it is able to perform the inverse modeling precisely in a short time. At the end, the performance of the designed network has been evaluated by a set of field MT data and its results has been compared with those produced by a common smooth rapid relaxation inversion (RRI) method. The comparison indicates that the results of these two different procedures are in close agreement.