Date of submission: 12
Download 0.72 Mb.
|
- Navigate this page:
- 6.1 Map validation and accuracy assessment
- 6.2 General trends and dynamics of urban growth
6. Results and DiscussionThis chapter provides a description of the analysis and results that were carried out for this study. These are broken down into:
6.1 Map validation and accuracy assessmentFigure 31 displays a composite map of the changes across the three epochs. This is an overlay of figures 18, 19 and 20 and gives an indication of the change from 1988 to 2010. 
Figure 31: Composite change in NDVI, from 1988 - 2010 The study area size is 1365 square kilometres (527 square miles). The changes in NDVI were calculated for each of the epochs separately and are presented as below: Table 8: Amount of NDVI change across the study area for the three change epochs
Omitting the no change figures for clarity the results for change in the table above can be seen graphically below: Change in NDVI values across the three epochs Figure 32: Change in NDVI graphs for the three change epochs The graph above shows a significant change between the first and second epochs, with a marked decrease in NDVI experienced. The results for epoch 1 are somewhat surprising, since it may be expected that an overall increase in NDVI would not have been experienced given that the city has been growing. The map validation exercises undertaken to assess the accuracy of the NDVI differencing method composed of the two stages discussed in section 5.4. The amount of speckle found in the Austin parks and green areas was found to be fairly low. As a percentage of the total image area these were as below: Table 9: Accuracy assessment for decrease in NDVI – amount of speckle experienced
The second accuracy assessment exercise that took place was with relation to areas experiencing an increase in NDVI values. Although the areas of undeveloped land were found to be a factor (see figure 23), a number of other areas experienced an increase in NDVI values, in particular in epoch 1. These were investigated by visual inspection and a clear pattern could not be identified. The land-use had not appeared to have changed for many areas so the change in NDVI appeared associated with a choice of threshold for the lower change value. However, the main focus for the study is urban growth and the decrease in NDVI values have represented urban growth well. The errors of commission experienced were low and the sample taken and percentages found were then multiplied by the NDVI values and used as a proxy for urban growth. Based on these figures, urban growth estimates for the three epochs are as follows: Table 10: Urban growth experienced in Austin
A composite map for urban growth in Austin across all three epochs is displayed as below. 
Figure 33: Austin urban growth 1988 – 2010 6.2 General trends and dynamics of urban growthThe analysis indicates that there has been significant urban sprawl in the Austin area, around 309 km2 in the past 22 years, approximately 14km2 per year. The results suggest that the majority of the sprawl was within the second epoch, from 1995 to 2003. Urban sprawl is linked to a growing population and an increase in residential developments away from the traditional city centre and the pattern of urban growth in the Austin area over the three epochs has reflected this. The distance of growth from the downtown Austin area was calculated and is shown below in 5km bands. 
Figure 34: Distance of urban growth from the Austin CBD. The distance is displayed in 5km bands Using the raster calculator the mean (and maximum) distance of urban growth in each epoch away from the traditional CBD was calculated. The results show an increasing distance in urban growth over time. Table 11: Mean and maximum distance of urban growth from the CBD
This pattern of growth is indicative of urban sprawl, as development generally takes place gradually further away from the traditional urban centre. The amount of growth for each epoch has been displayed graphically below. The maximum distance for sprawl on the graph is set at 35km. A change in shape can be witnessed, from a single ‘spike’ in amount of growth in epoch 1 to a bimodal pattern for epoch 2 and 3, with more urban growth being witnessed further away from the CBD. 
Figure 35: Distance of urban growth from CBD – epoch 1 
Figure 36: Distance of urban growth from CBD – epoch 2 
Figure 37: Distance of urban growth from CBD – epoch 3 Roads are also stated as an important factor in urban sprawl so the mean distance of urban growth to roads was calculated, again using the Euclidean distance and raster calculator as above. The results as below were obtained – with average growth for all epochs being within 2km of a primary road or secondary road. Table 12: Mean distance of urban growth away from a primary or secondary road
The distances from roads were also plotted graphically in order to better understand the relationship between urban growth and primary roads. 
Figure 38: Distance of urban growth from roads – epoch 1 
Figure 39: Distance of urban growth from roads – epoch 2 
Figure 40: Distance of urban growth from roads – epoch 3 A clear trend can be identified in the graphs above – a large amount of urban growth close to roads which then gradually slopes off in a logarithmic fashion. In order to test for a correlation the data needed to be transformed, so the natural logs were taken so a linear correlation test could take place. A Student’s t-test was performed to observe the statistical significance of the relationship. The results for the three epochs are displayed below. Table 13: Pearson’s correlation and student’s t-test results - distance of urban growth from roads
The transformed data all displayed very strong and highly significant correlations (all were actually significant at the 0.001 level), which suggests that urban growth and proximity to roads are strongly correlated. To facilitate interpretation of the overall general pattern of sprawl observed, a 3rd order polynomial trend surface was created using the Land Change Modeler, as spatial trend analysis is an effective way of visualizing the general observed change over time. Figure 41 below makes evident that the greatest extent of urban growth over the three epochs is concentrated in the north east of the study area. Spatial Trend of Change 
Figure 41: Spatial trend of change over the three epochs As a final stage in assessing the location and trends of sprawl, the census tracts of Austin were analysed in order to identify areas in which the greatest sprawl has taken place. The census tracts which experienced the most urban growth (as a percentage of their total size) were calculated and mapped below. For the three epochs these were located as in figure 42. 
Figure 42: Urban growth in census tracts Although epoch 1 and 3 show slightly fragmented patterns there are still noticeable clusters in tracts in the south and north respectively. Epoch 2 shows a definite cluster in census tracts in the north east of the study area, which is likely to be linked to regional development trends at that particular time. Download 0.72 Mb. Share with your friends:
|