This week I’m sharing with you some land use-land cover (LULC) maps. LULC maps show what is on a given area of land: water, developed cities, forests, pastures, crops, etc. I’ve been looking for recent LULC maps of Missouri for some time, and only just now found them at the website of the Multi-Resolution Land Characteristics Consortium (MRLC). This is a cooperative project by several branches of the federal government. Modern LULC maps are created by sophisticated imaging systems on satellites. This process allows for new maps to be created every few years, and the source I found has maps for 2001, 2003, 2006, 2008, 2011, 2013, and 2016.
LULC data is used for scientific study and for planning. For the purposes of this blog, what would be most interesting would be data regarding how much land in Missouri has changed use, and whether the changes represent a degradation of environmental carrying capacity. It appears to me that such data is available, either at MRLC or at the U.S. Geological Survey, but that it requires special programs and knowledge to utilize, which I don’t have. The closest I can come is a map provided by MRLC that shows the Land Use Change Index (land that changed use at least once between 2001, 2003, 2006, 2008, 2011, 2013, and 2016).
Figure 1 shows the 2016 LULC map for Missouri and a little bit of neighboring states. Each pixel on the map is 30 x 30 meters, which is close to 100 by 100 feet. Each pixel is coded with a color representing 1 of 21 categories of land use, shown in the legend. The pinks and reds jump out visually; these represent developed land. It is easy to make out the St. Louis, Kansas City, Springfield, Joplin, St. Joseph, Columbia, and Jefferson City areas, as well as other, smaller developed areas. The brown areas represent cultivated crops. The Bootheel is very heavily cultivated, but so is much of northern Missouri, especially the northwestern part. Green stands for forests – dark green for conifer forest and light green for mixed forest. There is a lot of forest in the Ozarks, mostly mixed forest. Yellow stands for pastureland/hay, which is widely distributed throughout the state. Blue stands for water bodies.
Looking at the map and forming overall impressions, I was first struck by how fragmented land use is. Land use varies by small plots in many areas of the state. Second, I was struck by how the forest land in the state is concentrated in the Ozarks. There are small forested regions in the north, but no very large tracts like in the south. Similarly, I was struck by how cultivated crops are almost completely absent from the Ozarks. There may be some farming there, but it looks to be devoted to hay and pasture.
Figure 2 shows a map of the Land Use Change Index in the same area. The resolution of this map is the same, each pixel is 30 x 30 meters. The map codes each pixel according to whether the use changed at least once between the mappings in 2001, 2003, 2006, 2008, 2011, 2013, and 2016. The change could have occurred in either direction. For instance, blue codes water change: what was a body of water became something else, or something else became a body of water. There are orange dots across the top of Missouri: either land became hay/pasture, or hay/pasture became something else. In addition, the change doesn’t have to have been permanent; what was hay/pasture in 2003 could have become cultivated cropland in 2006, but then reverted back in 2008. The map doesn’t tell us, it only tells us that at some point the land changed.
It’s not clear to me why they would construct this index this way. Perhaps these changes are reversible only in theory. For instance, the pink dots represent land that has either either urbanized or “un-urbanized.” I think that in real life, it is nearly impossible for land to “un-urbanize” – streets and buildings don’t go away in a matter of a couple of years. But it would be much more useful to know which changes were permanent, and which represented a degradation of the land.
Dark green represents land that did not change use across the 12-year time period. Most of the state is dark green. However, it is surprising how much land has changed. There are a lot of orange dots across northern and western Missouri, indicating that a lot of land there either became or stopped being hay fields/pastureland. Across southern Missouri, there are a lot of light green dots, indicating lots of change in the forests there.
Pink dots are scattered around the St. Louis, Kansas City, and Columbia regions, indicating land that (probably) urbanized. Given all of the development that has occurred in Springfield and Branson, I’m surprised that there isn’t more pink in that region.
Figure 3 shows the Land Use Change Index for the eastern Continental United States (CONUS). What immediately stands out is the forestland change across the Southeast, the Far North, and Maine. The map shows that change is happening, but not what kind of change. Figure 3 also suggests that, while land use is changing more in Missouri than in some states, it is changing less than in others.
Figure 4 shows the Land Use Change Index for the western CONUS. Across the West, forest change seems to be the most common change, followed by persistent grass and shrub change. One wonders how much of this is due to logging and/or development. However, one also wonders how much of this is due to either the bark beetle infestations that have been killing trees all over the West, or to the huge wildfires that have been ravaging these areas? For instance, using the interactive map on the MLRC website, I can zoom in and identify forest changes inside Glacier National Park. Those are not logging changes, they have to be the result of wildfire. However, there have not been hundreds of wildfires all up and down the Cascade Mountains in Oregon, those are likely to be logging changes.
Maps like these are created using large databases; each pixel is identified and has data coded for it. Theoretically, one should be able to use the database to construct summary statistics. I found a couple of databases, but unfortunately, they required specialized software to use, and/or they were multiple terabytes in size. Further, the agencies that create these maps don’t seem to think that states make useful boundaries. I will continue to keep my eyes open, however, and if I should come across summary data, I will do a future post on it.
Land cover change 2001-2004-2006-2008-2011-2016. Multi-Resolution Land Characteristics Consortium. Downloaded 2019-07-02 from https://www.mrlc.gov/viewer.
Screen capture from Multi-Resolution Land Characteristics Consortium. 2016 CONUS Land Cover. Downloaded 2019-07-02 from https://www.mrlc.gov/viewer.
Developed land is on the increase, while cropland, pastureland, and rangeland are on the decrease, according to the 2012 Natural Resources Inventory. The U.S. Department of Agriculture has conducted the inventory every 5 years since 1982, but it takes several years to put the report together, so the inventory for 2017 is not yet available.
Figure 1 graphs the surface area of the contiguous 48 states by land cover/land use in 2012. The top 3 uses were forest land, rangeland, and federal land, each of which accounted for 21% of the total. When the USA was first settled, forest land and rangeland were much more extensive, but they have been converted into cropland and developed land. In addition, we think of our country as having huge freshwater lakes, but only about 3% of the surface area is water. Freshwater is very precious and special.
Of course, federal land could also be categorized into forest land, rangeland, cropland, and the other categories, but the Natural Resources Inventory does not do so.
Figure 2 shows the change in land cover/land use since 1982. Over that time, cropland decreased and developed land increased by more acres than did any other category. “CRP Land” is land placed in the Conservation Resource Program.
The Natural Resources Inventory grew out of the National Erosion Reconnaissance Survey, conducted in 1934 because of severe dust storms and erosion during the Dust Bowl. Thus, since its inception, the report has been concerned with erosion. Figure 3 shows the estimated erosion rate on cropland in 1982, and Figure 4 shows the same data for 2012. You can see that in 1982, erosion was most severe in a region centered on Iowa’s borders with Illinois, Missouri, and Nebraska, but also extending along the Mississippi River into western Tennessee. In 2012, that region remained the one with the most severe erosion, but the rate had been significantly reduced. Across northern Missouri in 1982, more than 10 tons of soil eroded from each acre of cropland each year! In 2012 that had been reduced by 50% or so.
Figure 5 shows land use in Missouri from 1982 – 2012 in a few broad categories. The green areas of the columns represent federal land, which is not broken-out according to use. The red areas represent water. The two blue areas represent non-federal land, and they are broken into two categories: developed (light blue) and rural (dark blue). You can see that rural land represents by far the largest use of land in Missouri. In 2012, it represented 86.8% of Missouri’s surface area, while federal land, water areas, and developed land represented 4.5%, 2.0%, and 6.7%, respectively. Over the 30-year period, federal land increased slightly, water areas increased slightly, and developed areas increased by a whopping 38%, all being converted from rural land.
Figure 6 looks at Missouri’s non-federal rural land more closely. In 2012, more land was used for crops than for any other purpose (36% of rural land), followed by forest land (32%) and pastureland (27%). Over the 30-year period, the amount used for cropland decreased slightly, pastureland has decreased 17%, and rangeland, which was already such a small portion of the land that you can barely see it on the chart, declined 62%. Forest land and other rural land have increased. The Conservation Reserve Program (CRP Land) began after 1982, peaking in 1997, and declining since then.
This report is compiled and published by the U.S. Department of Agriculture, and from an environmental perspective it may be a bit misleading. Figure 5 shows that developed land represents only 6.7% of all Missouri land. However, Figure 6 shows that almost 1/3 of rural land is cropland, and another 27% of it is pastureland. It is not as if these lands are undeveloped. While they may not be covered in asphalt or highly populated, they are intensively used. They may be subject to high levels of erosion, as shown in Figure 3, or they may be disturbed by tilling and the application of agricultural chemicals. Pig farms and feed lots, for instance, are located in rural areas, but they are highly developed operations, in many cases resembling factories.
Thus, the Natural Resources Inventory probably provides the most comprehensive look at land cover/land use in the USA. It does not, however, provide an in depth review of the ecological status of the land.
Missouri Department of Natural Resources. 2018. Soil and Water Conservation Program. Viewed online 4/18/2018 at https://dnr.mo.gov/env/swcp.
U.S. Department of Agriculture. 2015. Summary Report: 2012 National Resources Inventory, Natural Resources Conservation Service, Washington, DC, and Center for Survey Statistics and Methodology, Iowa State University, Ames, Iowa. http://www.nrcs.usda.gov/technical/nri/12summary.
Into the discussion of whether sprawl leads to economic growth or economic growth leads to sprawl comes an interesting article by Alan Flippen in The Upshot, a blog published in the New York Times. He and Annie Lowrey ranked counties on 6 measures of education, income, unemployment, and health. They posted a wonderful interactive map on the Times webpage. It shows all the counties in the country. You mouse over them, and their ranking pops up, along with data on the 6 measures. (Find the map here.)
A county that is doing well would be one that is healthy, educated, and has a good economy. One that is struggling has poor health, low education, and a poor economy.
As you look at Missouri on their map, it is clear that taken as a whole, the state would not fare well–most of the state’s counties are struggling, only a few are doing well. There are states that are doing much, much better, but there are also states doing considerably worse – almost the entire South, plus Appalachia.
In Missouri, most of the southern part of the state is seen as struggling, while metropolitan regions and the far Northwest of the state are doing well. While this would appear on the surface to be related to the discussion of sprawl, I really think it is more of an urban-rural dichotomy. Urban centers tend to do better on measures of health, education, and the economy. There is also a north-south split in Missouri, with the struggling counties centered on the Ozarks. This may have something to do with the difference in topography: relatively flat, fertile farmland in the North, and rocky, mountainous land in the South.
This article appeared after I wrote the two preceding articles, but before they went live on the blog. It is strange how often something like that happens. It isn’t a government report or a study in a reviewed journal, so take everything it says with a grain of salt. Still, it is interesting and you might want to check it out.
Flippin, Alan. June 26, 2014. “Where Are the Hardest Places to Live in the United States?” The Upshot Blog in The New York Times. http://www.nytimes.com/2014/06/26/upshot/where-are-the-hardest-places-to-live-in-the-us.html?ref=us.
It would be easy to read my previous two posts about the Compactness Index and get the impression that sprawl is an evil in-and-of itself, while compactness is similarly a virtue in-and-of itself. I would caution against such easy assumptions, however.
Sprawl is the tendency for people to move away from densely populated central areas into less dense areas on the edge of cities or just outside them. People have been doing it for a long time, as the word “suburban” was apparently invented by Cicero, the ancient Roman orator. The fact that people have been doing it for such a long time, and that the trend continues, might be taken by some as prima facie evidence that suburbs are “better” than city cores.
Sprawl became a topic of significant discussion for American urban planners when the automobile became ubiquitous after World War II, freeing millions of Americans to move to the suburbs. Jobs and commercial development went with them. The causes are complex, but many urban centers depopulated and suburban areas grew like wildfire. Urban centers were sometimes left with shrunken tax bases, reduced populations, empty and decaying buildings, and large infrastructure networks that were no longer fully used and difficult to maintain with decreased revenues. Often it was the impoverished dispossessed who remain behind. This obviously represented problems for those urban areas, and sprawl has been a contentious political issue ever since.
There has been a proliferation of research purporting to show certain advantages or problems associated with both high density and low density. The research is difficult because of inherent problems separating which factors are causes and which are effects, and because there are many, many complicating factors that are virtually impossible to control.
I am comfortable with the idea of developing the Compactness Index. As I noted in the previous post, there may be other factors that tell you more about the differences between regions, nevertheless having a measure of sprawl is bound to help study the phenomenon. Smart Growth America is a national organization that researches and advocates for what have come to be called “smart growth” policies. They are often anti-sprawl. They were one of the sponsors that paid for development of the Compactness Index discussed in my previous two posts. The development of the index and the index values for MSAs and counties was published in Measuring Urban Sprawl and Validating Sprawl Measures, a report by the index developers that is available on the website of the National Cancer Institute.
In addition, however, Smart Growth America published its own report describing the index: Measuring Sprawl 2014. In addition to describing the index and giving index values for MSAs and counties, this report provides illustrations of the way the index can be used in research, and makes some claims for what has been discovered about sprawl. One of these claims disturbed me: “Compactness has a strong direct relationship to upward economic mobility,” the report claims in bold, colored lettering. (p. 9)
I came away from reading this with the impression that the report was claiming that compact areas were economically more vital than sprawled areas. Living in the St. Louis region, where such a claim is obviously not true, I was appalled. Then I read more closely. I could see that it was easy to get that impression, and if you, dear reader, read the report, you might come away with it, too. But it is not exactly what the report claims.
Instead, the report specifically claims that if a child is born into the lowest 1/5 of the income distribution, it has a better chance of making the top 1/5 of the income distribution by age 30 in a compact region than in a sprawled region. Now, this is a bit of an arcane claim. I’m not sure why the extreme of moving from the bottom 1/5 to the top 1/5 should be of general interest. It might seem that the more likely scenario of moving up one quintile would be more important.
I’m also struck by the use of quintiles to begin with. The simple laws of mathematics require that if some people are moving to higher quintiles, they have to be balanced by other people moving to lower quintiles. So, it must be equally true that more people drift down toward the lowest income level in compact areas also. As you might suspect, Smart Growth America doesn’t mention it.
The report gives an example: “The probability of an individual in the Baton Rough, LA area (index score: 55.6) moving from the bottom income quintile to top quintile is 7.2 percent. In the Madison, WI area (index score: 136.6) that probability is 10.2 percent.” (p.9) Thus, by moving from the 6th most sprawled MSA to the 13th most compact MSA , a person’s chances of moving up have increased less than 3%. Not a very big difference, frankly.
First, I looked up the Forbes Magazine list of the most desirable and least desirable places to live in America. I used the Forbes list because the primary factors they consider in their rankings are economic growth and the unemployment rate. I took the 10 most desirable locations and computed the average Compactness Index for them.Then I took the 10 least desirable locations and computed the average Compactness Index for them. There were a couple of places for which there was missing data because Forbes and the Compactness Index defined their metropolitan areas differently. But I found exactly the inverse of what Smart Growth America seemed to be claiming. (See table at right.) The 10 best places to live, presumably the ones with the highest economic growth and lowest unemployment, had an average Compactness Index of 103.01. The 10 worst places to live, presumably the ones with the lowest economic growth and highest unemployment, had a Compactness Index of 109.20. The high economic growth areas were more sprawled, the low economic growth areas were more compact.
I wasn’t comfortable stopping there, however. Forbes is a mass media publication, and the articles I saw didn’t describe their data methods in sufficient detail. So I did my own study. I compared the economic growth of the most compact and least compact MSAs. Specifically, using data from the Bureau of Economic Analysis, I computed the average economic growth rate for 2010-2012 for the 10 most sprawled MSAs and for 10 of the 11 most compact MSAs (economic data for one MSA was not available). The second table at right shows the results. The most sprawled MSAs grew an average of 2.21% per year, while the most compact MSAs grew an average of 1.19% per year. That is more than a 1% per year difference, a very large difference!
Now, these are informal “back-of-the-envelope” studies, and there are some significant methodological problems. The fact that the various lists may not have defined metropolitan areas similarly may not matter, but it might matter a lot. My little studies don’t prove anything. But they do seem to align with common perception, and perhaps they are sufficient to cast doubt on the importance of the Smart Growth America claim.
They also illustrate the difficulty separating cause from effect. Economic growth seems associated with sprawl, but is that because sprawl creates economic growth, or because economic growth creates more sprawl? Interesting question, hard to answer!
The reason I’ve gone into these issues in such detail is because they illustrate two issues that are very important to this blog. I had no problem with the report issued by the developer of the Compactness Index. It was posted on a federal government website. It appeared to me to be a reasonable and competent bit of research. It was only in the report issued by the interest group that I found the problem. That is why I tend not to use reports by interest groups in this blog. There are wonderful interest groups out there, and I don’t mean in any way to impugn the overall work of Smart Growth America. But I can’t use reports from interest groups for this blog.
Second, it illustrates a problem with all statistics. The famous quote says that there are lies, damn lies, and statistics. To really understand something, you have to go beyond the raw statistics and understand what is really being said. This blog is all about statistics. I try to read the statistics I report in this blog with a thoughtful and skeptical eye, but I am not an environmental scientist. If you are aware of problems with statistics I reporrt, I hope you will comment and let me know.
For Cicero inventing the word “suburb”: “Suburb.” Wikipedia. Viewed 6/17/14 at http://en.wikipedia.org/wiki/Suburb.
Ewing, Reid, and Shima Hamidi. 2014. Measuring Urban Sprawl and Validating Sprawl Measures. Salt Lake City: Metropolitan Research Center, University of Utah. Downloaded 6/13/14 from http://gis.cancer.gov/tools/urban-sprawl/.
Smart Growth America. 2014. Measuring Sprawl 2014. Downloaded 6/13/2014 from http://www.smartgrowthamerica.org/measuring-sprawl.
Greenburg, Zack O. “America’s Most Livable Cities,” Forbes Magazine, 4/1/2009. Downloaded 6/13/14 from http://www.forbes.com/2009/04/01/cities-city-ten-lifestyle-real-estate-livable-cities.html.
“Gross Domestic Product by Metropolitan Region,” Regional Economic Accounts, Bureau of Economic Analysis. Data downloaded on 6/18/2014 from http://www.bea.gov/regional.
In the previous post I reported on efforts to create a single index to represent how sprawled or compact a metropolitan region was. I included 2 tables, one showing the Compactness Index for 29 counties in Missouri, and another showing the Compactness Index for 3 Missouri metropolitan statistical areas (St. Louis, Kansas City, and Springfield).
The Compactness Index was originally constructed using data from the 2000 census, and recently updated using data from the 2010 census. The composition of the index was changed between the two, but in order to make comparisons between the years consistent, the authors went back and recalculated the 2000 data for counties using the new index.
The chart at right shows the change in compactness for the 26 Missouri counties for which both 2000 and 2010 data were available. Values below zero mean that the county became less compact, more sprawled. Twenty-two out of 26 counties in Missouri became less compact. Bates County led the way, with a whopping 24.16 decline in compactness.
(Click on chart for larger view.)
Similar comparisons for the St. Louis, Kansas City, and Springfield MSAs are unavailable.
Now, here’s a question: Bates County is on this list because it is part of the 14-county Kansas City MSA. However, it is a county of 17,049 souls about midway between Kansas City and Joplin. Butler is the largest town, with a population of 4,219 in the 2010 census. Outside of Butler, the population density is about 20 people per square mile. If a county is not part of an MSA, the Census Bureau requires at least 1,000 people per square mile for it to be urban. I noted in a previous post that small towns can sprawl just like large cities do, but I’m not sure why it makes sense to analyze an entire county like Bates County for urban sprawl.
The same could be said for other counties on the list. Caldwell County, for instance, is part of the Kansas City MSA, but it has only about 21 people per square mile. Lincoln County is part of the St. Louis MSA, and it has about 82 people per square mile. Polk County is part of the Springfield MSA, and it has about 48 people per square mile. Including these counties in analyses of larger MSAs seems like one thing, but why does it makes sense to analyze them separately for urban sprawl?
Let’s take a couple of additional examples to illustrate the point that other factors may tell you more of what you need to know about a region than the sprawl index. The Kansas City MSA has a Compactness Index of 77.60. The next lower MSA on the list is Palm-Bay-Melbourne-Titusville, Florida. This is the coastal region alongside the Cape Canaveral Space Center. Despite being similar on the Compactness Index, I suspect that these two regions are more different than the same.
The St. Louis MSA has a Compactness Index of 82.06. The next higher MSA on the list is Bakersfield-Delano, CA. St. Louis sits alongside the nation’s largest river, a former industrial powerhouse that was founded in 1764. The demographic majority is White, with African-American being the largest minority. Bakersfield was founded a century later. Three of its largest four employers are farming companies, and it is also the seat of the county that produces more oil than any other in the lower 48 states (Kern County). The largest demographic group is Hispanic, with Non-Hispanic Whites being the largest minority. Despite being similar on the Compactness Index, I suspect that these two regions are more different than the same.
So, Missouri is sprawling, as are most places in America. I’m just not sure what it means.
Ewing, Reid, and Shima Hamidi. 2014. Measuring Urban Sprawl and Validating Sprawl Measures. Salt Lake City: Metropolitan Research Center, University of Utah. Downloaded 6/13/14 from http://gis.cancer.gov/tools/urban-sprawl/.
For county-level oil production: County-Level Oil and Gas Production in the U.S., Economic Research Service, United States Department of Agriculture. Data downloaded 6/19/2014 from http://www.ers.usda.gov/data-products/county-level-oil-and-gas-production-in-the-us.aspx#.U6LaMagU_5I.
“Greater St. Louis,” Wikipedia. Viewed 6/16/2014 at http://en.wikipedia.org/wiki/Greater_St._Louis.
“Bakersfield, California,” Wikipedia. Viewed 6/16/2014 at http://en.wikipedia.org/wiki/Bakersfield,_California.