I take this opportunity to introduce myself. I worked as a research associate at Computer Vision Group, Department of Mathematics and Computer Science at University of Jena, Germany, after completing my MSc in Geo- informatics from Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, Netherlands and Bachelor of Technology degree in Information Technology from School of Information Technology, West Bengal University of Technology, India. I used to work as a Web Application Developer and Windows Administrator at Sarangtec Systems after completing my bachelors, but I found myself more interested in geospatial research rather than continuing as a Systems Engineer. I pursued my masters purely out of interest in Geo-information science and remote sensing and also for fulfilling my desire to study at ITC. The international standard of education at ITC helped me to make myself more proficient in various remote sensing techniques and technologies.
Additionally, programming skills in R, Python, SQL and C, which I have developed in my bachelors, masters and also during my two years of professional/trainee experience at Sarangtec Systems, and also from various projects that I did during my bachelors and masters level study. I am also a certified Linux Administrator from School of Mobile Computing and Communication, Jadavpur University, India and also have worked as a Windows Administrator during my professional stint at Sarangtec Systems.
Besides the above skills, I have great interest to work on geo-statistics, applications of GIS and data mining techniques that help to interpolate as well as to solve complex problems related to GIS and remote sensing. The advanced modules at ITC really helped to understand the fundamentals and also the advanced sections of image analysis using various machine learning techniques and also how geo-statistics can be implemented to interpolate the results using various models on a particular data. I have also implemented various machine learning and data mining techniques such as neural networks, SVM (Support Vector Machine), MRF (Markov Random Field), clustering and time series analysis during my course work. I have participated in a workshop on “Embedded and controls and software” conducted by Advanced Technology Development Centre, IIT-Kharagpur, India. I have worked on most open-source and proprietary GIS software’s since I started pursuing my masters. I have hands on experience in Digital Image Processing and Interpretation techniques using image processing and analysis software’s like ERDAS, ENVI, IDRISI and Arc GIS.
Some of my works in the peer-reviewed journals such as:
1. “Rainfall Mapping using Ordinary Kriging Technique: Case Study: Tunisia” – which shows how spatial interpolation techniques can be used to predict rainfall: using simple kriging with different variogram fitting models (Mukhopadhaya, 2016e).
2. “Land use and Land Cover Change Modelling using CA-Markov Case Study: Deforestation Analysis of Doon Valley” – the modelling of land use change for Doon valley was done using Markov Chain Analysis and Cellular Automata (CA) to predict the LULC of 2020 with data of 1999, 2006 and 2013 (Mukhopadhaya, 2016c).
3. “GIS-based Site Suitability Analysis: Case Study for Professional College in Dehradun” – this work explains the usage of GIS (Geographical Information System) and MCE (Multi-Criteria Evaluation) techniques for selection of most appropriate sites for developing a new professional college in the city of Dehradun, Uttarakhand (Mukhopadhaya, 2016b).
Other completed projects using remotely sensed data such as:
1. “Image Classification using comparison method of SVM-MRF and MRF-MLC” – It is mainly based on the comparison of SVM-MRF** and MRF-MLC** classification techniques, where linear/non-linear (using kernels) SVM (Support Vector Machine) were compared. This was done by elaborating the effect of feature selection on the classification, by identifying the best classification approach using R software.
**MRF (Markov Random Field), MLC (Maximum Likelihood Classifier).
2. “Land Information Management System for Agriculture” – It is mainly based on the ways of how agricultural land information can be handled using GIS technology along with web i.e., how a farmer/government official can use the information of the land. This project was done on a web platform using JSP, map server and web server.
3. “Hyperspectral Imaging System” – Here, the handling of hyper-spectral datasets was done, with more emphasis on the Hyperion data and also on its processing. This project was done on ENVI using Hyperion tool.
4. “Database Management System of Banks in an area” – Here, both spatial and non-spatial databases have been handled and managed with reference of the total banking databases including employees, customers, bank locations, ATM locations, etc. This project was done on an open source platform using QGIS and PostgreSQL for maintaining the database.
Summary of previous work:
During my tenure as a research associate at Computer Vision Group, I have worked on the causality of the carbon variables like GPP (Gross Primary Productivity), NEE (Net Ecosystem Exchange), TER (Total Ecosystem Respiration), SH (Sensible Heat), LE (Latent Energy), etc. The data was of the Hainich Forest, Germany. These variables were used to form a causal structure among the variables. This work was mainly done using statistical with data driven process to find the causal structure. I have worked on a project using multivariate time-series processing for non-stationary processes at my position as a research assistant at Shiv Nadar University, India. The project is based on network modelling of invasive species (Lantana Camara) spread in Rajaji Tiger Reserve. This project is mainly based on combining geographic information with network theory to explore network elements of invasion. In this research, we are studying Lantana Camara invasion in Rajaji Tiger Reserve (RTR). This critical tiger habitat may be lost in near future if the invasion of Lantana continues to spread. The spread of Lantana affects the presence of the herbivores present in the forest as this is not a proper food for them, which eventually affects the tiger population of the area. So, track the spread of lantana we have identified various elements which directly affect the spread such as: movement of pollen grains/seeds carried by birds or by the wind or even the human movement (if possible). We are working with the spread of lantana with respect to the climatic changes as well as with the interactions with the living beings. LISS IV, Panchromatic sensor satellite data and topographic maps are being analyzed to explore network elements, i.e., nodes and corridors for Lantana and also to detect the temporal change of the growth lantana for the past 5 years. Here, the multi-variables are taken into consideration to detect the change of spread of lantana and the variables are very much non-stationary.
Another project in parallel to the previous is to understand urban heat island and building energy management through mobile sensing. This project is based on the change of land use and land cover from pervious to impervious material interferes with the natural heat loss phenomenon for maintaining diurnal temperature range. A network of mobile sensors based on open source microcontrollers (Arduino UNO) is being deployed indoors and outdoors to detect the temperature and humidity using the DHT22 sensor. The energy composition with urban heat island can be detected with the help of humidity and temperature change in the urban areas having various LULC. GIS aids in visualizing spatial and temporal patterns in energy and comfort levels. We have taken the university campus (350 acres) as a prototype of an urban area having various LULC such as vegetation, open land, urban buildings, parks and small forest area. The elements considered for the change in urban heat temperature are: demography of the place, total population in that area (like a classroom of 300 students will have a different energy dissipated with respect to a conference room having 10 faculties or a playground which is open to natural light) and the amount of precipitation (to be considered after the above two factors are justified). Here, the multi-variables taken into consideration for modelling the urban heat are also non-stationary, as the population is changing and it’s random over a temporal phase.
I have previously worked on a project for rainfall mapping using the ordinary kriging technique. This work presents how spatial interpolation techniques can be used to predict the rainfall: using simple kriging with different variogram fitting models. The technique has been illustrated using annual rainfall observation measured at 75 climatic stations in a 165759.24sq.km region of Tunisia. Cross-validation has been used to compare the prediction performances of the geostatistical interpolation methods with the kriging method. The geostatistical method of ordinary kriging is quite useful for understanding the trend and can also be used for prediction. Here, the modelling using geostatistical method was completed with a stationary variable i.e. rainfall (Mukhopadhaya, 2016e).
I have also worked on GIS-based site suitability analysis using multi-criteria evaluation. — Geographical Information System (GIS) and Multi-Criteria Evaluation (MCE) are the standard techniques used to examine the possible sites for development in the urbanization of an area. The most important issue for a developing a city is the identification of proper locations for urban development. Site suitability is a way for understanding the existing site locations and also the elements that will help to decide the sites for a certain activity. This work explains the usage of GIS and MCE techniques for selection of most appropriate sites for developing a new professional college in the city of Dehradun, Uttarakhand. For this cause, Topo-sheet and satellite data has been used to produce different thematic layers by using software like ArcGIS. Criteria using connectivity of roads but away from highways, away from major residential areas but not far from city, land use/land cover, land proximity and other geographical information has been used for the analysis of the suitable site by properly evaluating the land. By measuring each criterion according to the importance, certain weights of each criterion is created. These weights and maps have been combined using ArcGIS tools and the final map was prepared showing the suitable sites (Mukhopadhaya, 2016b).
I have also worked on Land use and land cover change modelling using CA-Markov. In this work, the modelling of land use change for Doon valley was done. The modelling was begun with known land use at two different periods (2006 and 2013) and used them to project and model change into the future (2020). The two techniques used to model land use change are Markov Chain Analysis and Cellular Automata Analysis. This transition probability matrix was developed to show the land use land cover change from 2006 to 2013, which was used to predict for 2020. The transition probabilities were accurate on a per category basis, but there was a lack of knowledge of the spatial distribution about the instances within each land use category, i.e., there was no spatial component in the modelling outcome. Cellular automata (CA) helped to add character which had been spatially distributed to the model. CA_Markov was used to ‘grow out’ land use from 2013 to 2020 along with contiguity filter. In essence, the CA filter would develop a spatially-explicit weighting factor which would be applied to each of the suitability, measuring more heavily areas that were close to existing land uses. This ensured that land use change occurs proximate to existing land use classes, and not wholly random. Here, the transition area matrix was non-stationary, as the change was very random due to the constantly growing population in the area. The results also showed the same. The predicted 2013 LULC MAP was compared with actual 2013 LULC MAP which gave a comparable result with an overall error of 0.019%. The final predicted 2020 LULC MAP showed an overall increase in agricultural land, fallow land and vacant land by 124.42%, 0.77%, and 3.77% respectively, and an overall decrease in forest land and waterbody by 15.20% and 17.32% respectively (Mukhopadhaya, 2016c).
In my master’s thesis, I have worked on image analysis using different similarity and dissimilarity measures along with fuzzy classifier. Here, I have mainly worked on handling mixed pixel problems. The handling of mixed pixels are quite difficult as the reflectance value of the mixed class pixels are random as there can a mixture of 2 classes or 3 classes or sometimes can be of 4 classes. These mixtures mainly occur at the boundary of the classes and the randomness of the reflectance values changes with the composition of the classes in the mixed areas. To detect the mixture, fuzzy c- means (FCM) classifier has been studied with nine other similarity and dissimilarity measures: Manhattan distance, chessboard distance, Bray-Curtis distance, Canberra, Cosine distance, correlation distance, mean absolute difference, median absolute difference, and normalized squared Euclidean distance. Both single and composite modes were used with a varying weighted constant (m) at different α-cuts. Formosat-2 image and Landsat-8 image of 8m and 30m spatial resolution were used to implement the weighted norms respectively. The results showed that the best single and composite norms were obtained by optimizing the weighted constant (m). This helps in controlling the degree of fuzziness at various α-cuts. Here, FCM classifier was trained to detect the randomness in the mixed pixels and also to classify the untrained classes properly. I have worked extensively on multispectral datasets such as Landsat8 and Formosat-2 datasets for my MSc research hence have a good knowhow about the advantages and limitations of the same. The processing of the data was done with open source coding software R and also on a Java based in-house tool named as SMIC (Sub-pixel Multi-spectral Image Classification) tool (Mukhopadhaya, 2016a).
For the final year project during my bachelors, my group initiated a project based on online portal and we came up with the project titled as “Chess Masters Club”, which is an online portal for chess game providing human to human capability as well as human to computer capability for both playing and learning respectively. The main objective of this work was to design an online portal for the game of chess, where various users can register themselves as players. By using artificial intelligence (AI) approach, the functionalities of the special moves for each components of the game can be managed. Here, the users can not only play the game but also learn the nitty-gritties of the game using various tutorials and free online games at the beginners level. Advanced or professional players can also be a part of tournaments organized online. The work was done using secured hypertext transfer protocol for web services, along JSP and Java Beans at the application server and the database used was of IBM DB2. This work was solely guided by Dr. Rituparna Chaki, Associate Professor, University of Calcutta, India (Mukhopadhaya, 2016d).
If selected, this domain knowledge will really help me to do the work of development of algorithm easily and my experience of working with remotely sensed data will also be helpful for the development of the same. The various geo-statistical processes that I have learnt in the advanced modules will also help me to validate the work and work on the errors that may be observed or obtained from the data collected. My background of Information Technology will help to develop the software required for the project.
Considering the kind of work that I have done in past and my academic qualifications, along with the current work that I have completed in my MSc research, and also my desire to work further in my domain of interest, I strongly believe that, I do have the skills and qualifications required for taking up the position. I believe that I hold the required skill-set as the position asks for. I am really passionate about my work. I feel there is a lot of scope to learn and develop as a professional on the given topic and, my past work experience suits me for the above mentioned position and, also an opportunity to my future career plans of pursuing. I am confident that you would find in me an enthusiastic, motivated and dependable professional who would be an all-round asset to your team. I do have the potential of a fast learner and to complete work within the project time-lines and deliver under pressure and also my experience of studying at ITC will help me to deliver the desired outputs on time.
New Zealand is a very beautiful country, and to work at the Boffa Miskell along with the specialists in the interdisciplinary field will be a lifetime opportunity for me. This opportunity will not only help me to grow as a professional but also help me to learn a lot of new dimensions of the work associated with it. My previous experience in the field of using machine learning techniques using remote sensing and Geo-Information Science (GIS) will also be advantageous for the work designed. I look forward to pursue this beautifully designed course of work.
Thanking you for looking into my application. I look forward to your favorable reply.
Mukhopadhaya, S. (2016a). Exploring measures of similarity / dissimilarity for fuzzy classifier : from data quality to distance quality. (A. Stein & A. Kumar, Eds.). Enschede: University of Twente. Retrieved from http://www.itc.nl/library/papers_2016/msc/gfm/mukhopadhaya.pdf
Mukhopadhaya, S. (2016b). GIS-based Site Suitability Analysis : Case Study for Professional College in Dehradun. Journal of Civil Engineering and Environmental Technology(JCEET), 3(1), 60–64.
Mukhopadhaya, S. (2016c). Land use and land cover change modelling using CA-Markov Case study : Deforestation Analysis of Doon Valley. Journal of Agroecology and Natural Resource Management(JANRM), 3(1), 1–5.
Mukhopadhaya, S. (2016d). Online Chess Portal- Learning and Playing. Advances in Computer Science and Information Technology (ACSIT), 3(2), 110–115.
Mukhopadhaya, S. (2016e). Rainfall Mapping using Ordinary Kriging Technique : Case Study : Tunisia.
Journal of Basic and Applied Engineering Research (JBAER), 3(1), 1–5.