mid-term seminar
DESCRIPTION
Basics of Cellular Automata (CA)-Markov Model for Land Use land Cover modellingTRANSCRIPT
Modelling and Analyzing the Watershed Dynamics using Cellular Automata (CA) -Markov Model –A Geoinformation Based Approach
SCHOOL OF WATER RESORCES
INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR
Mid-Semester Seminar
15-10- 2009
Prepared by
SANTOSH BORATE
08WM6002
Under the guidance of
DR. M. D. BEHERA
CONTENTS• Introduction
• Review of Literature
• Aim and Objectives
• Study Area
• Methodology
• Model Description
- Markov Chain Analysis
- Cellular Automata(CA)
- Cellular automata-MCA in IDRISI- Andes
• Work Done
• Work to be done
• Conclusion
Introduction
• Watershed, Land Use/ Land CoverDefinition
• In order to maintain equilibriumbetween surrounding environment andclimate
Need of Watershed Modelling
• Prerequisite for Land Use Land CoverChange (LULCC) detection
Image classification
• Understand relationships & interactionswith human & natural phenomena tobetter management
Change detection
• Remote sensing & GIS tools providessynoptic coverage & repeatability thus iscost effective
Use of advanced spatial technology
tools
Introduction
Review of Literature
Methodology
Aim and Objectives
Study Area
Model description
Work to be done
Work Done
Acknowledge-ment
Review of Literature
Review of Literature
IntroductionAnuj Kumar Singh (2003) conducted study of LULCC with Cellular Automata(CA)which has advantage, that it incorporate the spatial component. Suggest thatCellular Automata(CA) Model is highly depend on Spatial variables taken in toconsideration . More variables can increase the accuracy of the model.
Daniel G. Brown(2004) Introduced the different type of models for LULCCModeling in relation to the purpose of the model, avaibility of data , driversresponsible for LULCC.
Antonius B. Wijanarto(2006) Described Markov Change Detection is oneapplication of change detection that can be used to predict future changes basedon the rates of past change. The method is based on probability that a given pieceof land will change from one mutually exclusive state to another. Theseprobabilities are generated from past changes and then applied to predict futurechange.
Thomas HOUET, Laurence HUBERT-MOY(2006) Cellular automata (CA), thatprovide a powerful tool for the dynamic modeling of land use changes, is acommon method to take spatial interactions into account. They have beenimplemented in land use models that are able to simulate multiple land usetypes.
Research Papers
Methodology
Aim and Objectives
Study Area
Model description
Work to be done
Work Done
Acknowledge-ment
Review of Literature continue……
Soe W. Myint and Le Wang(2006) This study demonstrates the integration ofMarkov chain analysis and Cellular Automata (CA) model to predict the Land UseLand Cover Change of Norman in 2000 using multicriteria decision makingapproach. This study used the post-classification change detection approach toidentify the land use land cover change in Norman, Oklahoma, betweenSeptember 1979 and July 1989 using Landsat Multispectral Scanner (MSS) andThematic Map (TM) images.
Huiping Liu (2008) Research shows that Land use/land cover change detectionusing multi-temporal images by means of remote sensing and ration research ofmodel of urban expansion by GIS are good means of research of urban expansion.
BOOKS1. Introduction to probability.
- Charles M. Grinstead, J. Laurie Snell2. Probability and statistics for Engineers and Scientists.
- Ronald E. Walpole3. Markov Chains Gibbs Fields, Monte Carlo Simulation and Queues.
- J.E. Marrsden4. Introduction to Geographic Information System(GIS).
-Kang-tsung Chang
Methodology
Aim and Objectives
Study Area
Model description
Work to be done
Work Done
Acknowledge-ment
Review of Literature
Introduction
Aim and Objectives
AIMTo Model and Analyze the Watershed Dynamics using Cellular Automata(CA) -Markov Model and predict the change for next 10 years.
OBJECTIVES To generate land use / land cover database with uniform classification
scheme for 1972, 1990, 1999 and 2004 using satellite data
To create database on demographic, socioeconomic, Infrastructureparameters
To derive the Transition Area matrix and suitability images based onclassification
Analysis of indicators and drivers and their impact on watersheddynamics
Projecting future watershed dynamics scenarios using CA-Markov Model
Methodology
Review of Literature
Study Area
Model description
Work to be done
Work Done
Acknowledge-ment
Aim and Objectives
Introduction
River basin map of India
STUDY AREA
• Drainage Area = 195 sq.km• latitude- 20 29 33.39 to 20 40 21.09 N•Longitude- 85 44 59.33 to 85 54 16.62 E•Growing Industrial Area
Mahanadi River Basin
Methodology
Review of Literature
Aim and Objectives
Model description
Work to be done
Work Done
Acknowledge-ment
Study Area
Introduction
Parameters to be considered
A) Biophysical Parameters: B) Socio-economic Parameters
1. Altitude 1. Urban Sprawl2. Slope 2. Population Density3. Soil Type 3. Road Network4. LU/LC classes 4. Socioeconomic Environment
a) Wetlands Policies b) Forest 5. Residential developmentc) Shrubs 6. Industrial Structure d) Agriculture 7. GDPAe) Urban Area 8. Public Sector Policies
5. Extreme Events 9. Literacya) Flood b) Forest Fire
6. Drainage Network 7. Meteorological
a) Rainfall b) Runoff
Methodology
Review of Literature
Aim and Objectives
Model description
Work to be done
Work Done
Acknowledge-ment
Study Area
Introduction
Acquired Satellite Data
LandsatMSS
PATH 150
ROW 46
Resolution 79m
LandsatTM
PATH 140
ROW 46
Resolution 30m
Satellite data for time period 1972 – procured from GLCF site
Satellite data for time period 1990 – procured from GLCF site
Satellite data for time period 1999 – procured from GLCF site
GLCF – Global Land Cover Facility
LandsatETM+
PATH 140
ROW 46
Resolution 30m
Satellite data for time period 2004 – procured from GLCF site
LandsatTM
PATH 140
ROW 46
Resolution 30m
Methodology
Review of Literature
Aim and Objectives
Model description
Work to be done
Work Done
Acknowledge-ment
Study Area
Introduction
Data Collection
1. Population Density
2. Land Use Land Cover
3. Soil Map
4. Rainfall
5. Road Network
6. Urban Sprawl
7. GDPA
8. Literacy
9. Residential development Methodology
Review of Literature
Aim and Objectives
Model description
Work to be done
Work Done
Acknowledge-ment
Study Area
Introduction
METHODOLOGY
Data download and Layer stack
Georeferencing and Reprojection
Area extraction
Multitemporalimage
Classification
Preparing Ancillary
Data
Statistics
TAM and Suitability Images
Simulation
Analysis
Prediction
Classification of the satellite data
Drainage Network Soil Type Altitude
Population Density
Road network
Calculation of LU/LC area statistics for different classes (for different periods)
Obtain Transition Area Matrix (TAM) by Markov Chain Analysis and Suitability Images by MCE
METHODOLOGY
Industrial Structure
Urban Sprawl Slope
Run CA- Markov model in IDRISI- Andes by giving -1) Basis land Cover Image , 2) TAM and 3) Suitability Image as inputs
Analysis of drivers responsible for watershed change
Predict future watershed dynamics for coming 10 years from the obtained trend
Toposheet 1945 MSS 1972 TM 1990 ETM+ 1999 TM 2004
CA-Markov Model Description
IDRISI Software
Markov Chain Analysis
Cellular Automata (CA)
CA-Markov Model in IDRISI AndesInput files- 1) Basis land Cover Image ,
2) Transition Area Matrix3) Suitability Image from MCE
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Work Done
Review of Literature
a) Research papers
b) Books
Formulation of Methodology
Analysis of parameters which to be consider
Acquisition, Georeferencing, Reprojection of Remote Sensing Data
Collection of data like DEM data, road network, drainage network, LULCC, Population, Rainfall etc.
Extraction of Study Area.
Unsupervised Classification of reprojected images
Introduction with Geoinfomatics software's ERDAS IMAGINE 9.1,
ArcGIS 9.1 , IDRISI Andes.
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Model description
Acknowledge-ment
Work Done
Introduction
Work to be done
Prepare the spatial layers of socio-economic parameters considered.
Obtain Transition Area Matrix by Markov Chain Analysis and Suitability Images by MCE
Run CA- Markov model in IDRISI- Andes
Analysis of drivers responsible for land use land cover change in watershed
Predict the watershed dynamics for next future 10 years
Study Area
Review of Literature
Aim and Objectives
Methodology
Work done
Model description
Acknowledge-ment
Work to be Done
Introduction
Acknowledgement
Prof. S.N Panda gave the guidance on Modelling of watershed.
Prof. C Chatterjee guided in selection of watershed
Prof. M.D. Behera guided in developing overall methodology and gave ancillary data.
Christina Connolly who gave the trial version of IDRISI Software from Clark Lab.
SAL (Spatial Analytical Lab) of CORAL Department and JRF and SRF in Lab.
GLCF (Global Land Cover Facility) – RS data download.
SRTM (Shuttle Radar Topography Mission )- DEM data download.
NRSC (National Remote Sensing Centre)- LULC data
Study Area
Review of Literature
Aim and Objectives
Methodology
Work done
Model description
Work to be done
Acknowledge-ment
Introduction
17
Markov Chain Analysis
Subdivide area into a number of cells
On the basis of observed data between time periods, MCA computes the probability that a cell will change from one land use type (state) to another within a specified period of time.
The probability of moving from one state to another state is called a transition probability.
Let set of states, S = { S1,S2, ……., Sr }.
where P = Markov transition probability matrix P i, j = the land type of the first and second time period Pij = the probability from land type i to land type j
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Markov Chain Analysis
Example: Forest in 2000 is change into two major classes in 2001,paddy field and residential; 33 % of forest is changing to residential,while 20 % changing to paddy field.
Forest
Residential
Paddy
2000 2001
F R P
F .47 .33 .20
P= R PRF PRR PRP
P PPF PPR PPP
transition probability matrix
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Markov Chain Analysis
Transition Area Matrix: is produced by multiplication of each column inTransition Probability Matrix (P) by no. of pixels of corresponding class inlater image
Disadvantages:
Markov analysis does not account the causes of land use change.
An even more serious problem of Markov analysis is that it is insensitiveto space: it provides no sense of geography.
- Although the transition probabilities may be accurate for a particular class as a whole, there is no spatial element to the modeling process.
- Using cellular automata adds a spatial dimension to the model.
F R P
F 94 66 40
A= R ARF ARR ARP
P APF APR APP
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Cellular Automata (CA) Model
Spatial component is incorporated
Powerful tool for Dynamic modelling
Each row represents a single time step of the automaton’s evolution.
St+1 = f (St,N,T)
where St+1 = State at time t+1
St = State at time t
N = Neighbourhood
T = Transition Rule
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Cellular Automata (CA) Model
Transition Rules Heart of Cellular Automata Each cell’s evolution is affected by its own state and the state of its
immediate neighbours to the left and right.
Fig. Von Neumann’s Neighbor and Moore’s Neighbor
Suitability Maps: Ex- To check the suitability of pixel for Settlement or Agriculture It depends on various Factors : biophysical and Proximity Factor like
altitude, rainfall, distance from road etcSc = Su + N……………………(1)
Su = (∑Wi * fi) ……………………………………(2)∑ Wi
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Cellular Automata (CA) Model
ClassesBiophysicalFactors
Settlement(Weights)
Agriculture(Weights)
Rainfall 4 8
Slope 8 2
Altitude 5 1
ClassesProximateFactors
Settlement(Weights)
Agriculture(Weights)
Distance From Road 10 6
Distance From City 5 7
Distance From Industry 3 3
Table.1. Allotment of Weights for Settlement and agricultural class
If SSet ≥ SAg ………then state = Settlement
If SSet ≤ SAg ………then state = Agriculture
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Cellular Automata(CA) –MCA in IDRISI -Andes
• Combines cellular automata and the Markov change landcover prediction.
• Adds knowledge of the likely spatial distribution oftransitions to Markov change analysis.
• The CA process creates a suitability map for each classbased on the factors (Biophysical and Proximate) andensuring that landuse change occurs in proximity to existinglike landuse classes, and not in a wholly random manner.
• In each iteration of the simulation each class will normallygain land from one or more of the other classes or it maylose some to one or more of the other classes.
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
Conclusion
Morkov Model does not incorporate the spatial component inmodelling Land Use and Land Cover prediction Integration of Markovchain analysis and Cellular Automata (CA) model adds knowledge ofthe likely spatial distribution of transitions to Markov change analysis.
Integration of Markov chain analysis and Cellular Automata (CA)model to predict the Land Use Land Cover Change is reasonablyaccurate , since it produces overall accuracy above the 85% whencomparing predicted map to the original satellite image
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
1. First (earlier) land cover image2. Second (Last) land cover image3. Prefix for output Conditional Probability Image4. No. of time period between first and last land cover image5. No. of time period to project forward from second image
Markov Spacelessness
Study Area
Review of Literature
Aim and Objectives
Methodology
Work to be done
Work Done
Acknowledge-ment
Model description
Introduction
1. Basic Land Cover image2. Markov Area Transition File3. Transition Suitability Image Collection4. Out Put Land Cover Projection5. No. of Cellular Automata iterations
1972 1990 1999 2004
Forest
Agriculture
Settlement
Wetland
Water Body