Using Idrisi to analyze Urbanization at Bohai Bay, China



by: Po Hu

Dec. 7th 2011

Earth Science Department

Advanced Image Processing, Emporia State University



Introdution

Bohai Bay is one of the three bays forming the Bohai Gulf. There are several oil fields located offshore in the Bohai Bay. By the drilling oil, urbanization developed fast on land around Bohai Bay in last few decade. The urbanization had been reforming the land. From 1996 to 2009 land cover had changed dramatically. Landsat 5 dataset were used to conduct land cover change analysis. By using visible, near-infrared (NIR) and mid-infrared (MIR) bands (see table 1) the land changes were much more visualize.



Table 1. Landsat 5 TM Spectral/Spatial Resolutions
Band Spectrum Spectral Resolution (micrometers) Spatial Resolution
1 Visible (Blue) (0.45 - 0.52 ) 30 m
2 Visible (Green) (0.52 - 0.60 ) 30 m
3 Visible (Red) (0.63 - 0.69 ) 30 m
4 Near-Infrared (0.76 - 0.90 ) 30 m
5 Near-Infrared (1.55 - 1.75 ) 30 m
6 Thermal (10.40 - 12.50 ) 120 m
7 Mid-Infrared (2.08 - 2.35 ) 30 m



Composite Images

The images above show the east part of the Bohai Bay. Images are shown in natural color.





We can easily see the urban area which is grey white shown on the both of the image. In 2009 the white area is much larger than 1998. Many port building appeared along the north of the bay, stretching out from land to sea, in the 2009 image. In the natural color image it is easy to see the urban areas, roads, building. But it is hard to distinguish other land cover.


In order to know more about the changes of the Bohai Bay area. Several false-color composite images were produced for the land cover distinguishing.



The false color composite 234 is the standard false color composite image. In the image blue represent water. Dark red represent active vegetation. The light red represent inactive vegetation. The city area is cyan. We still can see many dark brown or black areas that probably the agriculture field of rice, which grew in water.



In the false color composite 147 images. The urban areas which are purple are much easy to observe. The urban area is more than twice larger in 2009 than 1998. The urban area developed quickly along the river.


Vegetation Index


Vegetation indices for 1998 and 2009. The indices indicate the active vegetation at the area of Bohai Bay. The images are processed by using NDVI vegetation index of Idrisi. The index is calculated by using the band equation NDVI=(B4-B3)/(B4+B3).


The NDVI values rank from -1 to +1. Some vegetation place in low values could be related to high moisture or even they grow in water. Negative values represent inactive vegetation. The positive values represent active vegetation. In order to do more analysis with the NDVI vegetation index, NDVI values need to be changed to byte- binary data. First, use Scalar to add 1 in both of the images to make all value positive. Second use scalar to multiply the images by 125. Third use Convert to change the images from real-binary to byte-binary with rounding. In this way the NDVI images are converted to be byte- binary data which is available to be composited with other band images.

The below image is composite by using band 1, converted NDVI image and band4.



In the composite images the water are dark blue and the city area is Sapphire blue. Active vegetation is bright green and inactive vegetation and bare soils are dark green. It is easier to apart urban area from vegetation and soil as well as to observe condition of active and inactive vegetation .


Overlay images

In order to show the vegetation which is in low value on NDVI images which is effected by moisture. I used OVERLAY to show the water content in different area. Use OVERLAY to make 7/3 Ratio image which can show the different water content. The water is much less reflective in Mid-Infrared light than red light so that the lower values represent high water content. Then Scalar the largest value to 255 by multiply. Use band 1, overlay image and band 4 to make composite image. (See below)


(Left :1998 Right :2009)


In the image urban area is cyan vegetation is greenish yellow. The vegetation which live in water or in high moisture are brown. In image of 1998 there are many brown areas around city and north to the bay, but in 2009 most of the brown area disappeared.


Cluster Analysis


To create cluster images, I used CLUSTER model which is under Image Processing (Hard Classifiers). The cluster images are used to identify and classify different land cover at the area. When we start the CLUSTER, I choose 6 as the number of files and then input the band 1-5 and band 7 images. Once the images are produced, the different land covers will be shown in different color by random. And then we can use our knowledge or information on the composite images to identify what different color represent for.


There are 14 clusters in image of 1998 and 10 clusters in image of 2009. In the image of 1998, we can hardly see the shape of the urban area. Whereas in the image of 2009 apparently we can see that the urban area is cyan in the image. It matches well with the same area on composite images. Clearly the cyan represents buildings and structures both in urban and small towns. In this way it is much more obvious to know that the urbanization developed significantly fast from 1998 to 2009.


Conclusion


The purpose of the project is to realize the high urbanization around Bohai Bay. Based on all images produced by Idrisi including natural color images, false color images, NDVI vegetation index and composite images by using overlay image and NDVI image, the urban area expanded fast from 1998 to 2009. Urbanization also changed the vegetation at that area, especially vegetation growing in water. Satellite remote sensing datasets can provide real-time detection of urban area change. For further urban change assessment, we need to update the images of the area and use more tools for analysis.



Rederences

Clark Labs IDRISI Software (2003). http://www.clarklabs.org .


USGS Global Visualization Viewer. http://glovis.usgs.gov. (28/11/2011)


United States Geological Survey, EROS Data Center, World Wide Web homepage URL: http://eros.usgs.gov/.




This paper was written by Po Hu for the Advanced Image Processing Class at Emporia State University (ES775), fall session 2011, Instructed by Dr. James Aber.