Remote Sensing Project Report

Joel Heilman

Urban Development in Ames, IA 1991-2011

Ames is a fast growing city in central Iowa. According to the United States Census

Bureau, the population grew from 47, 198 in 1990 to 58,965 in 2010, which is close to the time

period the project studies. This is a change of 23.7%. The state of Iowa grew by only 9.5% in the

same amount of time.

The city has not only grown in population but also in land area. This project sets out to

study this change in land use. I grew up in Ames during most of the study time and I have

observed a lot of changes. I wanted to see what these changes looked like as a whole, and see if

what I observed was corresponding to what actually happened.

There were three main objectives for the project. The first objective was to create a

change detection map for the Ames area. From that, the goal was to see how much Ames has

grown in the twenty years from 1991 to 2011. Also, I used the map to determine what areas of

Ames had grown the most.

From this map and the conclusions, the city of Ames can determine how to use its

resources. They can create policies to help smooth growth and maintain urban sustainability. The

school system could use it to determine bus routes, and find a good spot for a potential new

school. Iowa State University can also use the results for expansion and to determine bus routes

for CyRide, the university bus system.

The location of the project was the city of Ames, located at 42.0° N, 93.6° W with an

elevation of about 940ft. Ames is located on the Des Moines Lobe, the area of Iowa that

experienced the last glacier during the last ice age. This area of central Iowa is typically flatter

than the rest of the state and has very good farmland.

Ames is on the western side of Story County, located just 30 miles north of Des Moines.

As of 2010, the city was home to over 58,000 people. From which about half are enrolled in

Iowa State University. In fall 2013, the city reached record enrollment at 33,241 students. The

University is also the largest employer in town. Ames and Iowa State University have become

almost synonymous.

The first step of the project was to collect data to use. I found images on the Glovis

website. I found the right location and set the cloud cover to zero percent. Clouds can make a

project impossible. The ground cannot even be seen if clouds are in the way. Even a few are not

good, especially while performing a classification. Clouds, and their shadows, are usually

classified as something else, so having no clouds in the area is necessary. Despite the setting,

both my images had a few clouds, but fortunately none were in my study area, so I could still use

the images. I found images from August 1991, and July 2011.

Figure 1: Ames, IA

Once I downloaded the images and loaded them onto Erdas Imagine 2013, I had to

georeference one to the other. The images did not have the same reference. I could not tell which

one was correct, or if both had the wrong reference. But as long as the images matched, I could

make a change detection map from them. So, I georeferenced the 1991 image to the 2011 image.

The next step was to create subset images from the two images. The subset feature in

Erdas Imagine allows the user to crop the image to a certain area. This way I could look at just

Ames. This made the rest of the process much easier. I drew a rectangle around Ames so I could

look at all the change around and within the city. I saved the Aoi file so I could use the same area

for both images.

The next step was to classify the two subset images. Since I was very familiar with the

area, I decided to use supervised classification. However, the colors of the images for different

classes were very similar to each other, and I had to redo the classification several times to get it

Figure 2: Methods for Project

remotely accurate. I used about 20 training sites for each image. The grass and farmland

appeared to be the same color, so I joined the two into one class. This was fine since my main

concern was the urban class and its change. I also noticed the water was more apparent on the

2011 image. The rivers were wider, and the lakes were bigger. This actually made sense because

2011 was a wet year and Ames experienced some minor flooding. 1991 was a lot drier in

comparison.

Urban 10 Water 11 Cropland & Grass 12 Forest 13

Urban 1 10 11 12 13

Water 2 20 22 24 26

Cropland & Grass 3 30 33 36 39

Forest 4 40 44 48 52

Figure 3: The recoded classified images. 1991 is on the left and 2011 is on the right. Red is urban, blue is water, green is forest, and brown is cropland and grass.

I then recoded both images so that there were only four classes, instead of 20. I then

recoded the 2011 image again to get different values so I could multiply the image values to get

the proper classes. Figure 4 shows how this multiplication works.

I used the two image function to multiply the image values together. Figure 4 shows the

resulting values. These values allowed me to determine which areas had changed and precisely

how they had changed.

Figure 5: The final urban change detection map.

I then recoded the image to show only five classes: urban (no change), water, cropland

and grass, forest, and urban growth. Many of the changed classes were small and mostly due to

inaccurate classification. Most of the true change was urban, so I was able to fit each class into

the five categories to get the resulting map as shown in figure 5.

I also performed an accuracy assessment of the two classified images. I could tell already

that the images were not classified perfectly. Much of the farmland in both images had spots of

forest and urban intermixed which I knew was incorrect. The 1991 image didn’t detect the rivers

as water nor did it classify all the freeways as urban.

I chose to do the accuracy assessment for each image with 50 stratified random points. I

compared the points to the original image and used my knowledge of the area to determine the

accuracy of the individual points. The results were a lot better than I expected. The 1991 image

had an overall accuracy of 84% and the 2011 image had an overall accuracy of 86%.

From the change detection map I was able to determine how much Ames had changed in

20 years. If the change corresponded to population, the result should be about a 23% increase in

population. Instead, the urban class had grown by 116%. According to the map, the size of Ames

had more than doubled.

The map also shows which areas grew the most. The most notable change was the

expansion of the city northward. The residential areas have grown here more than anywhere in

town, this matched my hypothesis. This expansion is directly related to the population growth,

but may or may not relate to the university’s growth.

Another notable area of expansion was the southwest part of town, along with other areas

of growth south of town. This development involves many types of new buildings. It involves

new housing and new apartments—some, not all, is for students. It also involves new stores and

restaurants, including Perfect Games—a hangout for both students and the community. Some of

it is also the expansion of the university, including a new athletic facility for the basketball teams

and research facilities. Much of the southern expansion is related to the university; in fact many

of the “holes” that are classified as cropland and grass is actually Iowa State athletic fields and

research fields for the college of agriculture and life sciences.

The third area of expansion was east of town. The area is home to many of the industrial

businesses in town. The area is mostly commercial buildings and factories. There’s also a

university research facility in the northern part of the region. A small Des Moines Area

Community College (DMACC) campus is also present, though there doesn’t seem to be any

significant growth. The area is also right next to the interstate which runs north to south just east

of town, so there are a good number of hotels and restaurants present.

The growth in the eastern part of town may or may not be related to the growth of the

university. The hotels accommodate for people visiting town, perhaps to visit the university.

Much of the industrial growth seems to be associated with the town, and not specifically the

university. However, the growth of Ames is usually related to the growth of Iowa State

University.

The expansion of Ames seems to make sense; it expands outward, especially near the

university and near the interstate. But there are other factors as well. For example, a flood map

can tell us a lot about why certain areas do not see urban growth.

Figure 6 shows the flood zones in Ames. These areas do not experience growth because

Iowa sometimes experiences flooding in the summer. Nobody wants to build on these lands

because the risks are too high. The map shows areas near Squaw Creek and the South Skunk

River only. A smaller creek, College Creek, also runs through the forested area northwest of

campus. These rivers and their flood zones account for the gaps in urban development. This is

the main reason why a lot of housing development has been pushed north—to avoid the

floodplains.

Figure 6: Flood Zones in Ames

The floodplains are not totally unused. Most of them are turned into parks and golf

courses. There are plenty of parks along the South Skunk River for example. There appear

classified as either cropland and grass or forest.

The biggest surprise of the project was the amount of growth; 116% growth seemed

absurd. The accuracy assessment was surprising; I did not think the classifications were as

accurate as they were. I decided to get a more accurate accuracy assessment would take many

more points.

However, through visual interpretation I could find some of the leading causes for the

problem. Both interstate 35 (runs north to south, east of town) and U.S. highway 30 (runs east to

west, south of town) are not classified correctly. The entirety of both highways is not classified

as urban, and most that is urban, is classified as urban growth. Both highways existed in 1991, so

they should both appear as no change. I also observed spots of urban growth outside of town.

Most of these areas were not urban at all.

Another problem was inaccurate georeferencing. Although the images had almost the

same reference, they did not line up exactly. I used about 15 points in my georeferencing, but

apparently this was not enough to get accurate results.

Although there was some error, Ames did grow more than I expected, and most likely

more than the 23% estimated. Part of this may be due to the housing boom in the early 2000s.

Another factor may be the increase in the size of houses. Many of the new homes in north Ames

are large houses with bigger plots. Therefore, the density of people is becoming smaller, which

means the land area of the city is growing faster than the population.

There are several things I would like to do for the future. One is to get better

classifications. This may mean doing a classification again with more training sites, but also I

think different raw images could give me different results. I also would need to be sure the

georeferencing is completely accurate.

I would also like to overlay the flood zone map with the classified change map. This

would give me a better understanding of how flood zones affect urban development. I could also

look at how different water heights affect urban development differently. The flood zone map I

used was also limited because it didn’t show College Creek or the areas south of U.S. highway

30.

Another interesting idea would be to try different years. Perhaps I could observe the

change half way during my study time by using an image from 2001. I could also use older

images from the 1980s or 1970s. It would be interesting to see the change all in one map

showing how much growth occurred each decade.

Despite difficulties, I was able to get some decent results. I produced a change map of

Ames between 1991 and 2011, and was able to determine which areas grew the most. Although

it is hard to say how much Ames did grow, it certainly grew more than I expected.

Works Cited

"Ames (city) QuickFacts from the US Census Bureau." United States Census Bureau, 27 June

2013. Web. 7 Dec. 2013.

"City of Ames." City of Ames, 2013. Web. 8 Dec. 2013.

"IFIS – Iowa Flood Information System." Iowa Flood Center, Web. 16 Dec. 2013.

"Iowa State University." Iowa State University of Science and Technology, 2013. Web. 7 Dec.

2013.

"USGS Global Visualization Viewer." USGS, Web. 2 Dec. 2013.