VIRGINIACARTER U. S. Geological Survey Reston, VA22092 DONALDL. MALONE JAMES H. BURBANK Valley Authority Chattanooga, TN 37401 Wetland Classification and Mapping in Western Tennessee

Seasonal color IR photographs provide sufficiently detailed information to map wetland areas as small as 0.5 ha in size and 20 'm in width.

INTRODUCTION maps will provide baseline information for resource management including the infor- HE U.S. GEOLOGICAL SURVEY (USGS) and ,,tion needed for T the Tennessee Valley Authority (TVA) legislative or regulatory requirements; have recently completed a cooperative wet- location of seasonally inundated and land mapping project in western Tennessee. permanently flooded areas;

ABSTRACT:The U.S. Geological Survey and the Tennessee Valley Authority have recently completed a cooperative wetland mapping project in western Tennessee. High-altitude color infrared photo- graphs were acquired by the National Aeronautics and Space Ad- ministration during several seasons in 1974 and 1975. These photo- graphs supplied the information on hydrologic boundaries and vegetation that was needed for classification and mapping. Seasonal information was required to map the maximum number of cate- gories. The stage (water level) was determined for the time of over- flights for sites where gaging stations are in operation. A wetland classification system was developed for the Tennessee Valley Region based primarily on vegetation, and on frequency and duration of inundation. Using this new classification system, wet- lands at four sites were mapped at 1:24 000 scale as overlays on U.S. Geological Survey 7.5-minute topographic maps. Adjacent land use was also mapped, but in less detail than wetlands. The meth- odology for separating and delineating wetland classes was care- fully documented. Overlays for separate dates were combined to make the final camera-ready composite overlay. A lithographed map of wetlands and land use was made for one of the five quadrangles covering the site. At the Reelfoot Lake and Hatchie sites, the stage at time of photography was referenced to a stage-duration curve, placed on the map collar, to show that boundaries are repre- sentative of average water levels rather than extreme highs or lows.

This experimental project was initiated in decisions on sites for agricultural, resi- response to local, State, and Federal man- dential, or industrial development; agement needs and concern over the loss of wildlife management and habitat acquisi- wetland habitat in the area. The wetland tion;

PHOT~GRAMMETRICENGINEERISG AND REMOTESENSING, Vol. 45, No. 3, March 1979, pp. 273-284. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1979

development of recreational opportunities; mapping the proposed wetland classes and and subclasses as well as surrounding monitoring change in vegetative cover or land use in the Tennessee ValIey Region; wetland acreage and (3) use seasonal photographs to study wet- Recent studies and publications have land boundary dynamics (changes in shown that high-altitude color infrared (IR) hydrologic conditions) and to relate photographs can be used for Level I1 land- boundaries to stage (water level). use and land-cover mapping (Anderson et This report addresses objective 2, and its 1976; Stevens, 1973; Stevens 1974) al., et al., primary emphasis is on the preparation of and for more detailed mapping of inland wetland maps for four sites in western Ten- wetland vegetation cover types (Neilsen and Wightman, 1971; DeSteiguer, 1977; Carter nessee using the wetland classification sys- tem developed for the Tennessee Valley and Stewart, 1977; Carter et al., 1976). The Region. The wetland classification system use of high-altitude color IR photographs to will be discussed only in sufficient detail to document surface-water boundaries has been discussed by Carter and Stewart (1977) explain map preparation. and Moore and North (1974). The authors acknowledge the assistance of personnel from the Reelfoot Lake National Three major objectives of this project were Wildlife Refuge, the Hatchie River .-, and the Tennessee Wild- develop a wetland classification system life Resources Agency. The USGS office in for the Tennessee Valley Region; Nashville, Tennessee collected water quality test the utility of seasonal high-altitude data at Reelfoot Lake and compiled and (1:130 000 scale) color IR photographs for interpreted stage information.

TABLE1. COMPARISONOF WETLAND CLASSES AND SUBCLASSESFOR THE TENNESSEEVALLEY REGION WITH USGS LEVELI1 CATEGORIES(ANDERSON et al., 1976) Tennessee Valley Region Wetland Classes Wetland Subclasses Level I1 Class FW-1 Bottomland (FW-la) Upper Forested Wetland Hardwood Bottomland Hardwood (FW-lb) Lower Bottomland Hardwood FW-2 (FW-2a) Forested Swamp (FW-2b) Shrub Swamp (FW-2c) Dead, Woody Swamp M-1 Marsh (M-la) Wet Meadow Non-Forested Wetland (M-lb) Emergent Marsh (M-lc) Seasonally Emergent Marsh M-2 Seasonally (M-2a) Vegetated Dewatered Flats (M-2b) Non-Vegetated M-3 Agriculture ngriculture Subject to Flooding OW-1 Open Water (OW-la) Vegetated dpen Water (OW-lb)Non-Vegetated WETLAND CLASSIFICATION AND MAPPING IN WESTERN TENNESSEE 275

the uscs land-use and land-cover classifica- tion system (Anderson et al., 1976) which was used for mapping adjacent land use. An For maximum utility of wetland maps, abbreviated definition of each class is in- the classification system used should meet cluded below, but subclass descriptions and state and local requirements and also be the lists of indicator species have been compatible with a regional or national sys- omitted for the sake of brevity. Figure 1 tem. Existing wetland classification sys- shows the entire Tennessee Valley Region tems were considered: some were too where the classification system is con- general to supply needed management in- sidered applicable. formation (Martin et al., 1953; Penfound, 1952), and others were not really applicable to the Tennessee Valley Region (Stewart Bottomland Hardwood (FW-1): and Kantrud, 1971; Golet and Larson, 1974). Wetland dominated by mixed hardwood Therefore, a new system was developed species and flooded annually during winter based primarily on vegetation and on fie- and(or) early spring. Flooding may be brief quency and duration of inundation (Carter or for long periods. The ground is usually and Burbank, 1978). Several modifications exposed in summer and fall although soil were made in the initial system as the map- may be saturated or covered locally with a pability of the classes was tested. The final few centimetres of water. system is, thus, designed to derive the opti- Swamp (FW-2): mum blend of interpretable and mappable Semipermanently or permanently flooded wetland information from the data source, in wetland dominated by woody vegetation this case high-altitude color IR photography. such as water tolerant deciduous trees, At the same time every attempt was made to aquatic shrubs or saplings, or dead trees provide a classification system useful for and shrubs. those engaged in ground inventory or habitat Marsh (M-1): evaluation. The U.S. Fish and Wildlife Ser- Wetland dominated by herbaceous emer- vice (FWS) has recently published an admin- gents. Flooding may be seasonal (generally istrative draft report on the new national for long periods in late winter and spring), wetland classification system (Cowardin et semipermanent or permanent, or it may be al., 1977) and our system appears to be temporary with soils remaining saturated for compatible with theirs. The Cowardin Sys- most of the year. tem was not used primarily because it was Seasonally Dewatered Flats (M-2): in an early stage of development when the Flats located along the margins of streams project began in 1974. and reservoirs and covered with water dur- Table 1 shows the wetland classes and ing most ofthe year. They may be uncovered subclasses and indicates how they fit into for varying amounts of time in periods of

FIG.1. Tennessee Valley Region, showing entire area included in wetland classification system. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1979

SCALE OF MILES WHC I I I 20 10 0 20 40 60 FIG.2. Location of wetland sites: (1)Reelfoot Lake, (2) Duck River dewater- ing area, (3) Hatchie River Bottoms, and (4) White Oak Swamp. low water or by artificial lowering of water banks. In late winter and spring, and for levels (generally during late summer or short periods following very heavy rain, a fall). These flats become vegetated when wide area of forested flood plain and some exposed for a sufficient period of time before surrounding agricultural land is flooded. the first killing frost. The two remaining sites are in the Ten- Agriculture Subject to Flooding (M-3): nessee River basin and are more or less Lowland areas usually under cultivation affected by controlled water levels in the during the growing season but generally river. The Duck River dewatering area, flooded each year during winter and early located where the Duck River enters the spring. Tennessee River, is surrounded by a levee Open Water (OW-1): and the water level is controlled by TVA. Water up to 3.3 m deep, usually associated Many acres of agricultural land in the area with any of the other wetland classes. Sur- are flooded annually and, until recently, a face vegetation (rooted floating-leaved or program for control of vegetation for wild- free-floating aquatics) and (or) submergent life enhancement was carried on by TVA. aquatics may be present. White Oak Swamp is west of the Tennessee River. White Oak Creek has been channel- ized and dredge spoil piled on the channel Four test sites, Reelfoot Lake, Hatchie banks. The few connections left between River Bottoms, White Oak Swamp, and the main channel and low areas behind the Duck River dewatering area, were selected spoil banks are mostly dammed by beavers. for classification and mapping. These sites Impounded water has created many acres of (Figure 2) are considered representative of Dead Woody Swamp, partially regenerated the diversity of wetlands found in the Ten- to Shrub Swamp and surrounded by For- nessee Valley Region. Reelfoot Lake is a ested Swamp or Bottomland Hardwood. tectonic feature created on the River flood plain by the New Madrid Earth- quake of 1811-12 (Glenn, 1933). This shal- The primary data source for mapping the low and heavily vegetated lake, surrounded land use and wetland classes and sub- primarily by agricultural land and upland classes was high-altitude color IR photog- forest, has the most diversity in wetland raphy obtained by NASA with a Wild RC-10 types of any site mapped. Hatchie River 150 lnln focal length camera* (225 lnln image Bottoms is an extensive and classic example of Bottomland Hardwood, extending along * The use of brand names in this report is for both sides of the Hatchie River. During identification purposes only and does not imply much of the year, including the growing endorsement by the U.S. Geological Survey or season, the river is contained within its the Tennessee Valley Authority. WETLAND CLASSIFICATION AND MAPPING IN WESTERN TENNESSEE format) at a scale of approximately 1: 130 000. (USGS and FWS) and Hatchie River (USGS and Natural water-level fluctuations, seasonal U.S. Army Corps of Engineers) were used growth of emergent aquatic vegetation, to determine the stage on the dates of over- and continuous tree cover in many areas flights. Stage-duration curves were devel- necessitated the use of seasonal photographs. oped for both test sites based on the five- Photographic coverage was, therefore, ob- year period 1970-1975. tained in February (high water, leaves-off), In addition to the preliminary field check- May (early growing season), and November ing, final map products were extensively (low water, leaves-off) of 1975, but cloud checked in the field at the Reelfoot Lake cover rendered most of the May coverage site by the authors and by the Tennessee unusable. Fortunately, growing season Wildlife Resources Agency (R. Fox, written photographs from October 1974 were avail- commun., 1976), with somewhat less inten- able for Reelfoot Lake, the site where the sive checking at the other three sites. At need was most critical. All photographs Reelfoot Lake the checking included run- were of good to excellent quality and no ning several transects to determine accuracy enhancement techniques other than en- of classification. largement were needed to aid the interpre- tation. Plate 1 shows Reelfoot Lake in October 1974, February 1975, and Novem- Wetlands and adjacent land-use were ber 1975. mapped on a total of 15 uscs 7.5-minute A mapping scale of 1:24 000 was selected quadrangles: Reelfoot Lake (5), Hatchie for several reasons. The recognizable or in- River Bottoms (5), Duck River dewatering terpretable limit of the 1:130 000-scale area (2), and White Oak Swamp (3). The photographs is an area less than one-half final map product is a clear positive overlay hectare in size. The small size of some wet- keyed to the appropriate 1:24 000 sheet. The land classes necessitated a large mapping overlay is on scale-stable base material and scale. The 1:24 000-scale uscs topographic can be reproduced in several forms includ- map series covers all of the sample test sites ing clear positive, matte finish positive, or in a reasonable number of map sheets, and paper print. provides other data for evaluating wetlands. Wetlands in the Tiptonville quadrangle These data include topography, cultural (Reelfoot Lake site) were color separated features, and surrounding drainage patterns. and printed using the 7.5-minute topographic This established map series which meets the map for base information to produce a 5- National Map Accuracy Standard (NMAS) color lithographic product (Plate 1). On the provides reliable geometric control for lo- lithographed map, Bottomland Hardwood cating wetland and land-use classes. and Swamp are displayed in green (FW-1, The delineation of the six wetland classes FW-2), Seasonally Dewatered Flats (M-2) and 12 wetland subclasses and surrounding and Agriculture Subject to Flooding (M-3) land use was done manually by an exper- are shown in brown, and Marsh (M-1) and ienced photointerpreter. The photointer- Open Water (OW-1) are shown in blue. preter also took part in the preliminary field Colors for the land-use categories are yellow checking and assisted in modifying the sub- for urban areas, green for forested, and white classes for maximizing the information de- for cropland and pastures. Copies of this rived from the photographs. Wetland areas map are available through TVA Mapping as small as 0.5 ha in size and 20 m in length Services Branch, Map Information and were mapped. The interpreted data were Records Unit, 100 Haney Building, Chat- transferred to stable base drafting film keyed tanooga, Tenn. 37401. to the appropriate 1:24 000 map. The regis- The boundaries that are most variable in tration of wetland data was accomplished by terms of yearly or seasonal fluctuation are use of a Bausch and Lomb Zoom Transfer those between upper and lower Bottom- Scope. For the less detailed land use sur- land Hardwood, between Bottomland Hard- rounding the wetlands a Kelsh stereoplotter wood and Forested Swamp, between Sea- was used. sonally Emergent Marsh and Open Water, A variety of photographic mapping criteria and around Agriculture Subject to Flooding. was established for distinguishing among Dashed boundaries rather than color were the six wetland classes, and selected sub- used for the first three of these cases in classes (Table 2). A photointerpretation order to indicate this variability. In an effort key was developed for each test site. The to make the class and subclass boundaries key for Reelfoot Lake is an example (Table more meaningful to the map user, the stage 3). at the time of photography is shown on a Records from gages on Reelfoot Lake stage-duration curve on the map collar (Fig- PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1979

- ct. 1974 Growing Season 75 High water, lei off ~v.1975 Low watc aves off

PLATE1. Seasonal color IR photographs of Reelfoot Lake, Tennessee were used to map wetlands for the Tiptonville, Tennessee quadrangle (see Table 2 for classification). WETLAND CLASSIFICATION AND MAPPING IN WESTERN TENNESSEE 279

TABLE2. PHOTOGRAPHICCRITERIA FOR MAPPINGWETLAND CLASSES AND SELECTEDSUBCLASSES Class Subclass Mavvinn Criteria FW-1 Bottomland 330 percent tree cover; flooded on high-water leaves-off photo- graphs (Feb.): not flooded on low-water leaves-off photographs (Nov.). High-water leaves-off photographs needed to separate sub- classes. FW-la: Upper Bottomland Hardwood: not flooded on normal high-water photographs (Feb.). FW-lb: Lower Bottomland Hardwood: flooded on normal high-water photographs (Feb.). FW-2 Swamp 330 percent woody vegetation cover: flooded on high- and low water, leaves-off photographs (Feb., Nov.). FW-2a: Forested Swamp: 230 percent live trees. FW-2b: Shrub Swamp: 330 percent live shrubs-Sometimes narrow hinge bordering Open Water and Forested Wetland. FW-2c: Dead Woody Swamp: S30 percent live trees, 370 percent dead trees, (impounded water). M-1 Marsh S30 percent woody vegetation cover, 370 percent herbaceous vegetation cover: growing season and either Nov. or Feb, photo- graphs needed to separate subclasses. Wet Meadow: flooded in high-water spring or winter photographs (Feb.): not obviously flooded in growing season or fall photographs (Oct., Nov.). Emergent Marsh: surface water present, but hidden by persistent vegetation except on high-water photographs (Feb.) when some water may be visible. Seasonally Emergent Marsh: 310 percent emergent cover: flooded in high-water winter or spring photographs (Feb.), mostly dead by fall low-water photographs (Nov.), visible in growing season photo- graphs (Oct.). M-2 Seasonally Visible in late growing season photographs (Oct.) or fall (Nov.) Dewatered Flats photographs at low-water, flooded on high-water photographs (Feb). Vegetated when cover exceeds 10 percent. M-3 Agriculture Flooded on high-water photographs (Feb.), agriculture in growing subject to flooding season (Oct.) or fall (Nov.) photographs. OW-1 Open s30 percent live tree cover: separable at low-water (Nov.) if not Water vegetated or associated with marsh class. Growing season photo- graphs needed to delineate if marsh class is present. OW-la: Open Water Vegetated: surface vegetation present in growing season photographs (Oct.): ~10percent emergents; minimum size 1 acre. OW-lb: Open Water Non-oegetated: no surface vegetation. ure 3). This information gives a measure of abrupt topographic changes may serve to how often a specific water level is equalled indicate the boundary of a wetland or wet- or exceeded. land class. Often, however, the change in vegetation is gradual and it is difficult to place a meaningful boundary within the Wetlands are dynamic ecosystems define- continuum from permanently wet to per- able in terms of hydrology, vegetation, and manently dry. Mapped boundaries of wet- soils, but difficult to map because of water- land classes and subclasses, if based on level (boundary) fluctuations. A photograph persistence and extent of surface water or a field visit results in a record of the lo- alone, will vary from year-to-year depending cation of the water's edge at one isolated upon weather and date of acquisition of instant in time. Vegetative composition or photographs. TABLE3. REELFOOTLAKE PHOTOINTERPRETATION KEY*

Type Tone Texture Shape Location Upper Bottomland 0ct.-bright red Very rough Variable-boundary Large areas throughout the project Hardwood FW-la eb.-lt. brown indistinct between area including the area Nov.-dk. blue to FW-la and FW-lb Feb.-lt. brown Lower Bottomland 0ct.-bright red Very rough Variable--boundary Hardwood FW-lb Feb.-dk. brown to indistinct between It. gray FW-lb and FW-2a Nov.-lt. brown to It. blue Forested Swamp 0ct.-blue-green Very rough Elongated-variable Large areas throughout the project FW-2a with some red & from large to extra including the Lake Isom area yellow small Feb.-dk. brown Nov.-dk. blue Wet Meadow M-la 0ct.-pink Medium to fine Small variable shape Near North end of Running Slough Feb.-not visible textured Nov.-lt. pink to It. blue Emergent Marsh 0ct.-red, blue, Fine to medium Variable-square to From North end of the project to M-lb green textured elongated the South end; mainly along Reel- Feb.-lt. pink- foot Lake, also in gray Lake Isom area Nov.-It. pink to gray Seasonally Emergent 0ct.-pink Fine to smooth Variable-small to From Upper Blue Basin to south of Marsh M-lc Feb.-not visible texture elongated Reelfoot Lake Blue Basin. Also in Nov.-not visible Lake Isom Seasonally Dewatered 0ct.-reddish blue Fine to smooth Variable-small to Along the MS River near Winches- Flats Vegetated M-2a to blue texture medium ter Towhead. North end of the Feb.-not visible project Nov.-not visible Seasonally Dewatered 0ct.-white to It. Smooth Large elongated Along the MS River near Winches- Flats Non-vegetated blue ter Towhead. North end of the M-2b Feb.-not visible project Nov.-not visible Agriculture Subject 0ct.-pink to red Coarse to smooth Variable Throughout project in the flat agri- to Flooding M-3 Feb.-not visible culture part Nov.--pink to bright blue Open Water 0ct.-pink to red Very fine to smooth In southern part of Reelfoot Lake Vegetated OW-la Feb.-not visible area Nov.-not visible Open Water Non- 0ct.-dark blue Uniform smooth Variable-elongated Throughout the whole project in vegetated OW-lb Feb.-light blue the Reelfoot Lake area and MS Nov.-light blue River. Also in Lake Isom area

* All arras referenced in plrotointerl~rrtatio~~key are shown on the Tiptonvillr, Term.-Mo.-Ky.; Ridgrly, Tmn.,Samburg, Tenn., Hornbeak. Tenn., and Bondurant. Ky.-MI).-Tenn.,quadrangles. PHOTOGRAMMETRIC ENGINEE:RING & REMOTE SENSING, 1979

REELFOOT LAKE STAGE-DURATIONCURVE September being less risky in terms of early Bared on Period of Record 1970.75 frost. Datum ol Gage 07027000 = 270.22 It above Mean Sea ~evsl 286.00 Interpretation of the February photo-

285.50 graphs was complicated by the presence of

285.00 muddy water due to a recent storm. Silted water may displace clear water under trees _ 284.50 284.00 during flooding, especially in lower areas or

283.50 along drainage channels where there is : 283.00 greater flow of water. The presence of two : 282.50 very different water "signatures," one dark H 282.00 blue to black or dark brown and the other light blue to white, was confusing, and for- 281.50 ested areas flooded with clear water were 0.01 0.1 1 2 5 10 50 90 99 99.99 misinterpreted as not flooded by the com- Percent of Tlme Gags Henght was Equalled or Exceeded Aernal Photography used tn Wetlands Msppnng piler until field checking in a beaver flooded Date: Lake Stage: area near the Hatchie River clarified the ostober 10. 1974 a 282.12 situation. February 25. 1975 0 282.86 The classification system, not being November 12. 1975 0 282.41 solely confined to criteria available through FIG.3. Reelfoot Lake stage-duration curve. remote sensing, contains some categories that are difficult to map from aerial photo- graphs. Depth of water was impossible to The boundaries of some subclasses (upper ascertain; rather than place an arbitrary and lower Bottomland Hardwood, Swamp) boundary on Open Water, we simply iden- on the wetland maps prepared in this project tified all Open Water regardless of depth. are based primarily on the presence or ab- The strong reflectance of both duckweed sence of water on the seasonal aerial photo- and the white water lily makes it possible graphs (see Table 2) rather than species to over- or under-estimate the extent of Sea- composition. This is partly because dis- sonally Emergent Marsh. Because there is crimination between deciduous species is relatively little white water lily present at difficult with high-altitude photography our sites, the difference is probably not and partly because we felt that hydrology significant. It is likely that inclusion of should take precedence over species compo- free-floating aquatic vegetation in the Open sition where the subclass was in question. Water class is unwise (even though they are By showing the relationship of the bound- usually associated with submergent aquatic aries taken from the photographs to informa- vegetation) because they can be relocated by tion on water-level frequencies with the wind action. Detection of submerged aquatic stage-duration curve, we hope to give the vegetation was not possible because we map user a better understanding of the basis could not establish a signature for it, even for our wetland boundaries. in areas where it could be seen with ground To map the maximum number of classes, observations and low-altitude color obliques. the need for seasonal photographs was sub- The Open Water Vegetated class is un- stantiated. The February and November doubtedly underrepresented on the map. photography was well timed, although the Ditching and beaver activity cause some tendency of oaks to retain their dead leaves interpretation problems. For example, the gave some problems in determining the Bottomland Hardwoods may be flooded November low-water boundary. Growing- permanently, or Forested may be season photographs are essential for several seasonally drained. An extended period of wetland classes. By comparing the May inundation may result in a rather rapid 1975 with the October 1974 photographs, we change from Bottomland Hardwoods to were able to see that the extent of Seasonally Forested Swamp to Dead Woody Swamp. Emergent Marsh is strongly dependent on Where there-is a rapid change from Forested seasonal water level, a fact verified by the Swamp to Bottomland Hardwood, or vice manager of the Reelfoot Lake National Wild- versa, the species composition is not as life Refuge. While May appeared to be a listed for the water regime, so we have arbi- little too early for full emergent growth, trarily decided that the water regime as October was quite satisfactory. Any time established from photographs and ground from August to mid-October should be opti- observations takes precedence over the mum for identifying the open-water and species composition. marsh categories, with August to mid- The techniques developed in this re- search project could be used to map the re- The methodology for separating and delin- mainder of the Tennessee Valley Region, or eating classes and subclasses is documented other areas containing inland wetlands, in this paper. using either the Carter and Burbank (1978) Overlays for separate dates were com- or the Cowardin (1977) classification sys- bined to make the final camera-ready com- tem. The accuracy and detail of mapping posite overlay. One of the five quadrangles would depend upon the availability of for Reelfoot Lake was made into a litho- seasonal photography. Table 2 shows that graphed product. Dates of photographs for it would be impossible to map all sub- Reelfoot Lake and Hatchie River are related classes without the seasonal photographs to available stage information in order to unless collateral data were available. High provide the user with information about the water, leaves-off photographs establish the range of water-level fluctuation and fre- wetlandtupland boundary and are used to quency of occurrence of the stage at which identify Agriculture Subject to Flooding and the boundaries were mapped. the two Bottomland Hardwood subclasses. Notable problems were experienced in Low water, leaves-off photographs are the interpretation of photographs, the sea- necessary to separate Bottomland Hard- sonal aspects of the data, and the wetland wood from Swamp unless it is known that classes themselves. Problems in interpreta- there is no Swamp present in an area. Grow- tion were generally resolved by field checks ing season photographs are needed to map during map preparation and following com- Flat, Open Water, and Marsh subclasses. In pletion. Submerged aquatic vegetation could areas where there are no Seasonally Emer- not be detected on the photographs and was gent Marshes and the wetlands are primarily not mapped. forested, a combination ofhigh water, leaves- In conclusion: off and low water, leaves-off will give all (1) A new classification system was de- forested classes except Dead Woody Swamp. veloped for wetlands in the Tennessee If it is impossible to obtain seasonal photo- Valley Region: graphs, the trade-offs must be evaluated and (2) Seasonal color IR photographs provide the date of photography will depend on sufficiently detailed information to map management priorities. wetland areas as small as 0.5 ha in size and 20 m in width; a minimum of ground truth is required, although field check- ing of final or interim products is always advisable; and The U.S. Geological Survey and TVA have (3) Dates of photographs, and thus wetland recently completed a cooperative wetland boundaries, can be related to stage records mapping project in western Tennessee. The to give an idea of range in water-level wetland classification system used is based fluctuation and placement of boundaries primarily on vegetation, and on frequency within this range. and duration of inundation. There are two forested wetland classes: Bottomland Hard- wood and Swamp, and four non-forested Anderson, j. R., E. E. Hardy, J. T. Roach, and R. wetland classes: Marsh, Seasonally De- E. Witmer, 1976, A land-use and land-cover watered Flats, Agriculture Subject to Flood- classification system for use with remote ing, and Open Water. There are a total of sensor data, U.S. Geol. Surv. Prof. Paper 964, 12 subclasses, five forested wetland and 28 p. seven non-forested wetland. Mapping Carter, Virginia, and J. H. Burbank, 1978, Wetland criteria have been developed for each class classification system for the Tennessee Valley and photointerpretation keys were de- Region, Tennessee Valley Authority Techni- veloped for each site. cal Note No. B24, 36 p. Using the classification system, wetlands Carter, Virginia, M. K. Garrett, Lurie Shima, and at Reelfoot Lake, Hatchie River Bottoms, Patricia Gammon, 1976, The Great Dismal Duck River dewatering area, and White Oak Swamp: Management of a hydrologic re- Swamp were mapped as 1:24 overlays source with the aid of remote sensing, Water 000 Resources Bull., v. 13, no. 1, pp. 1-12. on Geological Survey 7.5-minute maps. Carter, Virginia, and W. R. Stewart, 1977, Season- Adjacent land use was also mapped, but in al color-infrared photographs for mapping in- less detail than wetlands. NASA seasonal land wetlands on U.S. Geological Survey 7.5- high-altitude color IR photographs were minute quadrangles, 5th Biennial Workshop used as the primary data source. The sea- on Color Aerial Photography in the Plant sonal photographs were required in order to Sciences; 1975 Proc., Am. Soc. of Photogram., map the maximum number of categories. pp. 143-161. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1979

Cowardin, L. M., Virginia Carter, F. C. Golet, and Neilsen, U., and J. M. Wightman, 1971, A new E. T. LaRoe, 1977, Classification of wetlands approach to the description of the forest re- and deep-water habitats of the , gions of Canada using 1:160,000color infared U.S. Fish and Wildlife Serv., Office of Biol. aerial photography, (Canadian Forestry Serv., Washington, D. C., 100 p. Service) Forest Management Inst. Inf. Rept. DeSteiguer, J. E., 1977, Forest type mapping of FMR-X-35, 25 p. the Atchafalaya River basin from satellite Penfound, W. T., 1952, Southern swamps and and aircraft imagery, 5th Biennial Workshop marshes, Bot. Rev., v. 18, no. 6, pp. 414-446. on Color Aerial Photography in the Plant Stevens, A. R., 1973, Applications ofhyper-altitude Sciences, 1975 Proc., Am. Soc. of Photogram., color infrared photography to land use map- pp. 129-142. ping in the Tennessee River watershed, In Glenn, L. C., 1933, the Geography and Geology proceedings, 39th Annual Meeting, 1973, of Reelfoot Lake, Jour. Tenn. Acad. Sci., v. 8, Am. Soc. of Photogram., Washington, D. C.: no. 1, p. 3-11. pp. 201-209. Golet, F. C., and J. S. Larson, 1974, Classification Stevens, A. R., C. W. Craven, W. H. Ogden, and of freshwater wetlands of the glaciated H. B. Wright, 1974, Alternatives for land use1 northeast, U.S. Fish and Wildlife Serv. cover mapping in the Tennessee River water- Resource Pub. 116, 56 p. shed, In proceedings, 34th Annual Meeting, 1974, Am. Cong. on Surv. and Mapping, St. Martin, A. C., Neil Hotchkiss, F. M. Uhler, and Louis, Mo.: pp. 533-542. W. S. Bourn, 1953, Classification of wetlands of the United States, U.S. Fish and Wildlife Stewart, R. E., and H. A. Kantrud, 1971, Classifi- Serv. Spec. Sci. Rept. 20, 14 p. cation of natural ponds and lakes in the glaciated prairie region, U.S. Fish and Wild- Moore, G. K., and G. W. North, 1974; Flood inun- life Serv. Resource Pub. 92, 57 p. dation in the southeastern United States from aircraft and satellite imagery, WaterResources (Received May 13, 1978; revised and accepted Bull., v. 10, no. 5, pp. 1082-1097. December 4, 1978)

The Fifth Purdue Symposium on Machine Processing of Remotely Sensed Data West Lafayette, Indiana June 27-29, 1979 Sponsored by the Laboratory for Applications of Remote Sensing, Purdue University, and cosponsored by the American Society of Photogrammetry, the Symposium will focus upon the theory, implementation, and novel applications of machine processing of remotely sensed data. The emphasis will be on research results in the following three broad areas: Digital representation and understanding of remotely sensed scenes. Utilization of digitally processed Earth resource data. Extraction of information primarily from digital remotely sensed Earth resource data. The program will be designed to provide an opportunity for scientists working in these areas to present current research and applications results and to describe new technological developments and novel applications. For further information please contact Dr. Luis A. Bartolucci or Dr. LeRoy F. Silva Laboratory for Applications of Remote Sensing (LARS) 1220 Potter Drive, Purdue University West Lafayette, IN 47906 Tele: (317) 749-2052

Forthcoming Articles Dr. Youssef I. Abdel-Aziz, An Application of Photogrammetric Techniques to Building Con- struction. N. Y. Chu and P. E. Anuta, Autonratic Color Map Digitization by Spectral Classification. Dr. Alden P. Coloocoresses, Proposed Parameters for Mapsat. Dr. Lawrence T.Fisher, Dr. Frank L. Scarpace, and Richard G.Thomsen, Multidate Landsat Lake Quality Monitoring Program. Alan M. Hay, Sampling Designs to Test Land-Use Map Accuracy. Mike L. Mathews and Garland N. Mason, High Intensity Dot Grids. Major L. G. Thompson, Determivation of the Point Transfer Error.