Overview of Change Analysis Statistics

An accurate, highly visual analysis of land cover changes in a region is the responsibility of the NOAA Coastal Services Center's Coastal Change Analysis Program (C-CAP). Satellite data, primarily Landsat Thematic Mapper (TM), plus extensive ground checks, provides the data and the maps that document and illustrate how land cover in a region changes from one time period to the next.

The focus of C-CAP's data acquisition is on coastal wetland habitats and adjacent uplands. When the satellite image exceeds this boundary, C-CAP often classifies the entire Landsat TM scene.

The image to the left displays the full extent of the C-CAP land cover and change analysis data for South Carolina and a portion of North Carolina. This data has been compressed with PKZIP® software and is located in the following directory: data/land_cov/lc_spi.zip.

Change analysis statistics for the region depicted are provided in the next section below.

To document specific changes in landcover from one date to another, a change table is used. Various types of land cover are categorized. The first vertical column represents categories in the 1990 land cover data. The the first horizontal row represents categories in the 1995 land cover data. Reading the chart from left to right you can obtain what a 1990 land cover class changed to in 1995. For example, notice the "mixed forest" row. This represents areas that were categorized as mixed forest in 1990. Each column that intersects with that row represents what the areas were categorized as in 1995. Notice the "cultivated" column. Follow it down to its intersection with mixed forest. During the change analysis period, 437 acres changed from mixed forest to cultivated land.

Make note of the diagonal cells in the table, which are bold. Cells on the diagonal did not change between the two dates. All cells that are off-diagonal represent change.

	Developed High Intensity	Developed Low Intensity	Cultivated	Grass	Deciduous Forest	Evergreen Forest	Mixed Forest	Scrub/Shrub	Palustrine Scrub/Shrub	Palustrine Forest	Palustrine Emergent Wetland	Estuarine Emergent Wetland	Unconsolidated Shore	Bare Land	Water	Totals
Developed High Intensity	118,939	0	0	0	0	0	0	0	0	0	0	0	0	0	0	118,939
Developed Low Intensity	746	257,228	0	0	0	0	0	0	0	0	0	3	0	0	0	257,979
Cultivated	383	428	2,168,092	43,299	0	0	0	14,508	0	1,574	14	2	1	530	696	2,229,527
Grass	1,009	2,157	12,178	1,056,508	0	0	0	128,118	0	5,807	153	11	11	4,527	522	1,211,002
Deciduous Forest	33	109	176	2,157	306,332	0	1,524	10,641	92	415	16	0	0	380	80	321,957
Evergreen Forest	886	2,290	914	71,686	0	1,956,648	3,127	216,343	5,867	10,241	941	29	11	16,311	692	2,285,986
Mixed Forest	265	912	437	18,198	213	15,604	783,392	62,410	1,642	1,664	111	6	1	2,653	244	887,753
Scrub/Shrub	1,273	3,661	4,941	73,812	1,039	176,938	6,177	3,102,976	4,300	7,606	587	54	18	14,351	1,559	3,399,293
Palustrine Scrub/Shrub	293	971	957	28,315	207	4,526	2,884	44,010	2,761,010	44,212	572	55	7	8,788	1,468	2,898,275
Palustrine Forest	105	190	943	5,577	5	6,451	370	11,398	3,648	311,361	527	77	13	1,652	1,237	343,552
Palustrine Emergent Wetland	20	26	79	2,207	0	0	0	1,538	294	2,616	63,672	151	12	1,437	279	72,332
Estuarine Emergent Wetland	73	28	1	1,383	0	0	0	711	0	2,056	0	369,826	282	1,000	750	376,109
Unconsolidated Shore	39	33	0	59	0	0	0	67	0	82	31	31	23,266	289	631	24,529
Bare Land	200	157	352	2,026	0	0	0	9,221	0	1,372	667	31	284	58,505	2,095	74,909
Water	58	44	55	682	0	0	0	655	0	1,897	315	128	131	1,570	5,568,719	5,574,253
Totals	124,322	268,234	2,189,127	1,305,910	307,796	2,160,166	797,473	3,602,597	2,776,854	390,902	67,607	370,404	24,037	111,993	5,578,974	20,076,395
													Total Acreage	20,076,395

Change Analysis: Interpreting the Data

To be applicable for change analysis, it is necessary that land cover categories reflect transitional features of the landscape. For example, a time series of classified scenes for a land area occupied by mature forest that is harvested might exhibit a transition from forest land to bare land to grassland to scrub/shrub and back to forest. Another example are transitions from natural categories to low and high density developed, encroachment or changes in wetland categories, phenomena such as shoreline movement, and loss or addition of aquatic emergent and submersed habitats.

Much of the change exhibited in the South Carolina study area falls into predictable change categories. These could be linked to major economic activities that take place along the coastal plain. There are two major change categories that are most observable in South Carolina. The greatest amount of change is related to silviculture activities. Significant change could also be noted around population centers with the increase of land development for human needs.

Forestry as an Agent of Change

Silviculture, or forestry, is a land intensive practice that imposes an often-dramatic impact on the landscape. The impact of forestry is evident when viewing remotely sensed imagery. Vast areas are carved out for agricultural and rangeland purposes, as are the transportation infrastructure and urbanized features of the land. Forestry transitions detected by the C-CAP include the cutting of trees, building of roads, development of dikes and trenches to drain the land, planting of trees, prescribed burning, stages of growth of same age trees, forest thinning, and transition of mixed forest habitats to monoculture.

Changes in evergreen, deciduous, mixed, and palustrine forest types are monitored with this land cover and change analysis product. In 1990, for example, 2,285,986 acres of evergreen forest were recorded for the land classified in the South Carolina mosaic. During 1995, a total of 2,160,166 acres of evergreen forest were recorded. This results in a loss of 125,820 acres of evergreen forest. Part of this loss is due to urban development.

Using the table, note that between 1990 and 1995 886 acres of evergreen forest changed to developed high intensity. This means that evergreen forest was cleared for urban development. Additionally, 2,290 acres of evergreen forest were converted to developed low-density. This type of change, evergreen forest to developed low-density, has less of an impact on the affected evergreen forests because not all of the trees are removed during this type of change. The developed low-density class represents areas that are developed but retain 20-50% vegetative cover.

Change from evergreen forest to grassland is a typical transition noted in forestry practices. Some time during the change period, 71,686 acres of evergreen forest was cleared and grasses colonized the land. Changes to grassland are also a component of human development. Often areas of grassland can be distinguished within urban and suburban developments. One major component of change to grassland in developing areas is golf courses, which can be identified as numerous, closely organized strips of grassland.

Change to scrub/shrub is another typical feature of forestry. Early in the change period 216,343 acres of evergreen forestland was cleared and made the transition from bare land to grassland to scrub/shrub.

Evergreen forest to bare land is another feature of forestry practices. Between 1990 and 1995, a total of 16,311 acres of evergreen forest were converted to bare land.

Change Due to Urban/Suburban Growth

The coast of South Carolina has been experiencing accelerated development due to expansion in recreation, retirement, and industrial markets. The Grand Strand in Horry county, the Charleston metropolitan area, and Beaufort county are dramatic examples. The transformation of the landscape associated with this growth is represented by changes to a number of classes. These include developed high-density, developed low-density, grassland, and bare land.

Developed high-density is land cover that consists of greater than 80% constructed surfaces. These changes are often found along major road and bridge construction, industrial and business development, and shopping plazas. Developed low-density land cover consists of greater than 50% constructed surfaces. The remaining area is usually vegetated cover. These changes occur with road and bridge construction, subdivision development, and divergent building construction. Change to grassland is witnessed when areas under development are converted to features such as lawns, parks, cemeteries, and golf courses. Grassland is also a transitional feature related to development. As the predominant land cover is removed grass is often planted to stabilize the soil while construction is taking place. Similarly, bare land is another transitional feature of development.

Other Notable Change Classes

A dominant feature of the South Carolina coast is the extensive salt marsh habitat. Salt marsh or estuarine emergent wetland is composed of expansive communities of smooth cordgrass growing on a tidally influenced substrate. This habitat can be seen behind the barrier islands and in the estuaries all along the coastline. Numerous creeks and channels leading to open sounds or the ocean dissect the marsh flats. Another species associated with this land cover category is bulrush, which is usually found in communities on slightly higher ground. Research has shown that salt marsh habitat is a rather stable environment, demonstrating little change during a twenty-year research period.

Changes that do occur in estuarine emergent habitats are generally related to human impacts or storms. Notable human changes are often state or federally permitted activities such as road and bridge construction. In the Charleston metropolitan area change to the estuarine emergent environment include the Isle of Palms Connector crossing the marsh from Mount Pleasant to Isle of Palms and the James Island Connector linking Charleston and James Island.

Other impacts due to human activity are related to the economic history of coastal South Carolina. The rice culture of the 1600-1900s left an indelible imprint on the landscape. Coastal marsh and wetlands were trenched, diked, and built with elaborate control structures. These structures took advantage of South Carolina’s high tidal difference (6 foot average) to allow filling and draining of impoundments during the rice growing cycle. Many of these structures still function today to control water for wildlife, while others are deteriorating allowing saltwater intrusion into the canal system. Natural and human forces often determine whether fresh or salt water dominates in these abandon rice impoundments. Therefore, it is likely that changes may be noted in these areas from an estuarine to palustrine system. That is the case in the southern Santee Delta.

Another consideration is the fact that Hurricane Hugo made landfall September 21, 1989. Hugo was a category four storm with a 15 - 20 foot storm surge and winds reaching speeds of 138 miles per hour in some regions. The storm left its mark on the South Carolina coast. The first date of the satellite imagery used in the land cover classification study was acquired in 1990, only a year after the hurricane. The storm opened rice impoundments, pushed saltwater landward, deposited sand and shell along the coast and in the salt marsh, and washed out areas of higher ground. Some of the changes found between Bulls Bay and Winyah Bay are due to these effects.

Environmental Considerations

Failure to understand the impact of various environmental characteristics on the remote sensing change detection process can lead to inaccurate results. For example, when examining coastal areas, tides can have a significant impact on classification of low lying areas. With the Landsat TM satellite passing overhead every 16 days, it is extremely difficult to acquire two dates of imagery in which the tides are perfectly coordinated. It should also be noted that due to the large swath width of the TM sensor, tides can vary dramatically within a given image. If tides are too high, marsh areas can be erroneously classified as water, causing an underestimation of marsh area and inaccurate results when the image change comparison is performed. C-CAP protocols are designed to assure that high tide conditions do not cause underestimation of marsh. However, due to the difficulties in obtaining two or more image dates with similar tidal conditions, change in unconsolidated shore should be viewed with caution.

It is also important to consider seasonal and annual phenological cycles in vegetation. Obtaining near anniversary images greatly minimizes the effects of seasonal differences that may appear to be categorical change in the imagery. However, phenological cycles can also be used to the image analyst's advantage. C-CAP obtained leaf-on and leaf-off imagery for the entire South Carolina study area. Leaf-off imagery can be used to more accurately classify forested wetlands, which are more difficult to distinguish from upland forests during leaf-on conditions. Leaf-off imagery also can aid in detection of low intensity developed areas. For example, in older neighborhoods with heavy deciduous tree cover, branches that overhang roads and homes can obscure the underlying structures, leading to an underestimation of the developed - low intensity class. Leaf-off imagery can help to rectify this problem.