Date of submission: 12

Download 0.72 Mb.

Page	1/17
Date	20.10.2016
Size	0.72 Mb.
	#6129

1 2 3 4 5 6 7 8 9 ... 17

Using GIS, remote sensing and a land cover change model to assess urban sprawl dynamics and forecast future change in Austin, Texas

David Hunt

MSc Thesis submitted in partial fulfilment of the requirements of the degree of Master of Science in Geographic Information Science at Birkbeck, University of London
Date of submission: 12^th September 2012

Total word count: 9982

Abstract

The ability to map and monitor the extent of urban sprawl over time has important environmental, economic and societal relevance. In this study I developed a methodology to map and describe urban growth using Landsat 5 data in the city of Austin for a three epoch time-series dating from 1988 to 2010. The NDVI differencing change detection method was employed to establish urban growth and filtered to avoid confusion with agricultural variability. The results highlight significant urban growth throughout the Austin area, with the city expanding at approximately 14km²per year over the 22 year period. Trends typical of urban sprawl were witnessed, such as greater urban growth away from the city centre over time and a strong correlation between urban growth and roads. The change maps were then input into the Idrisi Land Change Modeler in order to make a prediction for the amount and location of future urban sprawl. The distance from previous sprawl and distance from roads were input in order to create a transition potential for the model. A map for 2010 was created and compared to the actual 2010 land cover map. The model was able to predict urban growth and was very close to the actual land cover map in terms of amount of sprawl. Finally a simulation was run for the year 2015. The study highlighted the potential of using multi-temporal Landsat data and remote sensing methods to provide an accurate and economical means to map and analyse urban growth over time. The maps created, along with the prediction for the future could provide useful tools in land management planning and policy decisions to tackle urban sprawl.

Acknowledgements

Abstract 1

Acknowledgements 3

List of Figures 5

List of Tables 7

Declaration 8

1.Introduction 9

2.Aims and Objectives 12

2.1 Research Hypothesis 12

3. Literature Review 13

3.1 Urban Sprawl 13

3.2 Remote Sensing 19

3.3 Change Detection 22

3.4 Land Use/Cover Change Models 24

3.5 IDRISI Land Change Modeler 25

3.6 Multi-Layer Perceptron Neural Network 26

3.7 Predicting Change – Markov Chain 28

4. Study Area 29

5. Methodology 33

5.1 Data Sources 33

5.2 Image Pre-processing 36

5.3 Change Detection 42

5.4 Validation and Accuracy Assessment 53

5.5 Spatial Analysis 56

5.6 Land Cover Change Modelling 58

5.7 Transition Potentials 63

6. Results and Discussion 72

6.1 Map validation and accuracy assessment 72

6.2 General trends and dynamics of urban growth 76

6.3 Austin population change 85

6.4 Creation of a change map for 1988-2010 85

6.5 Validation of change map 88

6.6 Predicted urban growth map for 2015 91

6.7 Summary of Results 92

6.8 Limitations 97

7. Conclusion 100

7.1 Future Research 101

References 103

Appendix 107

Appendix A – NDVI threshold selection 107

Appendix B – Austin Census Tracts 109

List of Figures

Figure 1:	Urban sprawl in Melbourne, Australia……………………………………………………	13
Figure 2:	Neural Network Diagram……………………………………………………………………….	24
Figure 3:	Location of Austin in the USA and Texas………………………………………………..	26
Figure 4:	Austin, Texas, with primary/secondary roads present………………………..…	27
Figure 5:	Austin’s urban area expansion from 1970 – 2004…………………………………..	28
Figure 6:	Generalised work flow for creation of change detection maps……………..	30
Figure 7:	Austin urban area, as delineated by the US Census bureau, 2010………….	34
Figure 8:	Austin false colour composite for 1988…………………………………………………..	35
Figure 9:	Austin false colour composite for 1995…………………………………………………..	36
Figure 10:	Austin false colour composite for 2003…………………………………………………..	37
Figure 11:	Austin false colour composite for 2010…………………………………………………..	38
Figure 12:	VEGINDEX module in Idrisi – NDVI selected…………………………………………..	39
Figure 13:	Austin NDVI map for 1988………………………………………………………………………	40
Figure 14:	Austin NDVI map for 1995………………………………………………………………………	40
Figure 15:	Austin NDVI map for 2003………………………………………………………………………	40
Figure 16:	Austin NDVI map for 2010………………………………………………………………………	40
Figure 17:	Austin agricultural areas, identified from the Austin land use data set….	41
Figure 18:	Change detection map for epoch 1, between 1988 and 1995………………..	43
Figure 19:	Change detection map for epoch 2, between 1995 and 2003………………..	44
Figure 20:	Change detection map for epoch 3, between 2003 and 2010………………..	45
Figure 21:	Accuracy assessment for errors of commission…………………………………….	46
Figure 22:	Raster calculator to assess errors of commission…………………………………	47
Figure 23:	Accuracy assessment for increase in NDVI…………………………………………..	48
Figure 24:	Calculating the distance of urban growth away from the CBD……………..	49
Figure 25:	Generalised workflow for the future urban growth prediction map……	51
Figure 26:	Change in urban growth between epoch 1 and epoch 2………………………	52
Figure 27:	Distance from roads calculated in Idrisi………………………………………………..	54
Figure 28:	Distance from previous sprawl calculated in Idrisi……………………………….	55
Figure 29:	MLP Neural Network sub model run……………………………………………………	57
Figure 30:	Transition potential map – from Austin to a decrease in NDVI……………	58
Figure 31:	Composite change in NDVI, from 1988 – 2010…………………………………….	61
Figure 32:	Change in NDVI graphs for the three change epochs………………………….	62
Figure 33:	Austin urban growth 1988 – 2010………………………………………………………..	64
Figure 34:	Distance of urban growth from the Austin CBD…………………………………..	65
Figure 35:	Distance of urban growth from CBD – epoch 1……………………………………	66
Figure 36:	Distance of urban growth from CBD – epoch 2……………………………………	67
Figure 37:	Distance of urban growth from CBD – epoch 3……………………………………	67
Figure 38:	Distance of urban growth from roads – epoch 1…………………………………	68
Figure 39:	Distance of urban growth from roads – epoch 2…………………………………	69
Figure 40:	Distance of urban growth from roads – epoch 3…………………………………	69
Figure 41:	Spatial trend of change over the three epochs…………………………………..	71
Figure 42:	Urban growth in census tracts…………………………………………………………….	72
Figure 43:	Austin population figures from 1990 – 2010……………………………………….	73
Figure 44:	Change demand modelling dropdown in Change Prediction tab, LCM..	74
Figure 45:	Comparison of actual decrease in NDVI map for three epochs and prediction created in LCM……………………………………………………………………	75
Figure 46:	Validation results…………………………………………………………………………………	76
Figure 47:	Validation map showing false alarms, misses and hits…………………………	77
Figure 48:	Predicted Urban growth map for 1988 – 2015……………………………………..	78
Figure 49:	Time series of urban growth in Austin…………………………………………………	81