WILLIAMMACCONNELL Department of Forestry 6 Wildlife Management University of Massachusetts Amherst, MA 01003 WILLIAMNIEDZWIEDZ Division of Environmental and Urban Systems Virginia Polytechnic Institute Blacksburg, VA24061 Remote Sensing the White River Vermont

Aerial photographs provide a data base for wildlife and fisheries biology studies of the effects of stream channelization on an excellent trout stream which was once a prime spawning ground for the Atlantic salmon.

INTRODUCTION River and its tributaries is known as an ex- ORRENTIAL FLOODING in the White River cellent trout stream, in the past as a prime T Watershed ofVermont during the period spawning stream of the Atlantic salmon June 26 to July 6,1973 led to a Declaration of (Salmo salar), and as the site of a National Disaster (FDAA #397-DR, dtd. July 6, 1973) Fish Hatchery devoted to Atlantic salmon issued under P.L. 91-606. All Vermont gen- culture and rearing. Additional stream alter-

ABSTRACT:Torrential rains and severe flooding in the White River Watershed in Vermont in June 1973 resulted in the issuance of a Declaration of Disaster, which led to channelization in the river and its tributaries. Conventional springtime panchromatic aerial pho- tography at a scale of 1:12,000 was taken the following year in order to establish a data base for wildlife and fisheries biology studies of the effect of the channelization on the river. The aerial photographs were used to assess the impact of stream channelization on the White River and its tributaries, measure vegetation and land use of value to wildlife along the river, determine stream bank characteristics, and delineate and measure water characteristics of value to fish in the river. Springtime photography without leaves on vegetation or snow on the ground prjved very useful in delineating channelized sections of the river; and in identifying vegetation, land use, and river bank types of importance to wildlife as well as water characteristics of importance to fish.

era1 laws relating to wetlands, stream pro- ation was implemented by the Agricultural tection, and water quality were nullified for Stabilization and Conservation Service a 60-day period by a gubernatorial declara- under P.L. 85-58 and by the U.S. Soil Con- tion. Under the authority of P.L. 91-606 and servation Service under P.L. 81-516. with Federal Disaster Assistance Adminis- This paper describes a study initiated to tration funding, the White River among document by aerial photographs the ecologi- other rivers in Vermont was channelized to cal changes resulting from channelization of reduce future flooding threats to public the White River and its major tributaries. It health, safety, and property. The White also measured the characteristics of the

PHOTOGRAMMETRICENGINEERING AND REMOTE SENSING, Vol. 45, No. 10, October 1979, pp. 1393-1399. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1979

river, its banks and vegetation, and land use turbulent water areas are captured as mot- adjacent to the river which would have an tled, patchy areas on aerial photography. effect on wildlife and fish. The White River study attempts to classify Research objectives were first to provide a on aerial photographs river and stream water habitat data base of the White River and as habitat for fish, and land adjacent to the vegetation and land use adjacent to it for water as habitat for wildlife. The product fisheries biologists studying benthos and produced was used in research by wildlife fish habitat and wildlife biologists studying and fisheries biologists making on-ground wildlife and wildlife habitat. The second and in-water studies ofwildlife and fish. It is objective was to document the extent and expected that annotated negatives showing nature of stream channelization. The third conditions on the White River in the spring objective was to develop a fish and wildlife of 1974 will be printed for use by biologists habitat classification system to include veg- in the future, if the White River once again etation, land use, stream bank parameters, becomes important for Atlantic salmon stream-water characteristics, and chan- spawning. nelizing activity. Flood damage assessment and flood pre- diction are necessary components to any The White River Watershed is located in successful land-use planning process that central Vermont and consists of a drainage covers flood-prone rivers and shorelines. area of 1844 km2 (712 mi2). Headwaters of Many flood-prone areas are highly popu- the White River and its tributaries are lo- lated, heavily industrialized, or prized for cated on the slopes of the Green Mountains, agricultural uses. Flood-plain development and it joins the Connecticut River at White in flood-prone areas has increased the need River Junction. Included in this study are for careful planning and management by the White River; the First, Second and Third local, state, and federal agencies. Accurate, Branches of the White River; Stony Brook; rapid, and economical methods are required Gilead Brook; and the West Branch of the to provide data necessary for water resources Tweed River. Most of these tributaries have assessment for planning-management func- clear water with a rock and gravel stream tions (Meyers and Welch 1975). Remote bed. Agriculture and urban land uses domi- sensing techniques have been utilized for nate the lowlands adjacent to the river while many years to aid the process of studying a mixed hardwood-conifer forest (Tsuga floods and flood-caused damage. However, canadensis, Picea spp., Acer spp., and Betula the thrust of Remote Sensing applications spp.) borders the tributaries on the uplands. has been toward flood inundation mapping The river and strips of riparian land to a dis- and property damage assessment (Hoyes et tance of 305 m (1000 feet) on both sides of al.. 1973: Piech and Walker, 1972). the river make up the study area. kerial photographs at scales ranging from 1: 10,000 to 1:15,000 can be used directly to obtain information for quantitative descrip- Standard panchromatic aerial photographs tion of drainage systems (Meyer and Welch, at a scale of 1:12,000 were taken of the study 1975). As outlined by Bunik and Turner area on May 2, 1974 when snow was gone (1971) black-and-white photography at these and trees were without leaves. The Zeiss scales can be utilized to produce a data base RMK-A camera used had a 15 cm (6 in.) focal defining stream-drainage basins to their length and a 23 by 23 cm (9 by 9 in.) contact lowest orders. Meyer and Welch (1975) indi- film format. Specifications for photography cate that cross-sectional information is pos- were those defined by the U.S. Department sible for streams where photography pene- of Agriculture (1969). To capture maximum trates to the streambed or with ground truth photographic detail of streambanks and data. water characteristics, photographs were ex- Panchromatic, infrared, color infrared, and posed during the brief period between loss color photography as well as thermal imag- of snowcover and budding of deciduous ery techniques have been used successfully vegetation. Four power pocket stereoscopes to assess water patterns related to changes in were used for interpretation and .000 diam- channel conditions. Coleman (1969) and eter technical ink fountain pens were used to Coleman and McIntire (1971) have demon- annotate both the photographic prints and strated that water surface turbulence can be corresponding negatives (Figure 1). Dot correlated with channel topography. Their grids were used to measure area types, map studies show that relatively calm water nor- measures were used to measure line types, mally overlies shallow bars, while highly and point types were counted for tabulation. REMOTE SENSING THE WHITE RIVER IN VERMONT 1395

FIG.1. Reproduction of a photo copy of a 1:12,000 scale annotated negative showing water characteristics, stream hank types, and vegetative and land-use types near the mouth of the White River in Vermont. Boundaries of the stream bank types are marked with Vs and the type symbols are shown on the water adjacent to the stream banks. Water and stream bank type symbols are described in this paper while vegetative and land-use types on the bank of the river are described in a bulletin by the authors cited in the references. (Photo by J. W. Sewall Co., Old Town, Maine)

-Large boulders with diameters A water and river bank classification sys- greater than 38 cm in diameter tems for use with 1:12,000 scale leaves-off -Dead woody debris piled along the panchromatic photographs was devised to streambank from past floods provide biologists with needed information E - Eroded streambank about fish and wildlife habitat in and near P - Rip-rap the water. Water in the river, the character of W - Retaining wall its banks, and land adjacent to the river were Elevation of Streambank all classified and measured employing the 1 -A beach-like situation with no ab- following system. rupt drop to the water 2 -An abrupt drop to the water which is less than 3 m (10 feet) 3 -An abrupt drop to the water of 3 to 6 Channelization m (10 to 20 ft.) C -Evidence of stream channelization 4 -An abrupt drop to the water which is on the banks more than 6 m The ClGl type would be a channelized Shade Provided by Streambank stream with little shade, and gravel banks Vegetation which have a beach-like character. 1 -Little shade on the water, 0-30 per- cent 2 -Light shade, 31-50 percent 3 - Medium shade, 51-80 percent 4 -Heavy shade, 81-100 percent 0 -Pools over 30 m (100 ft) in diameter with fish holding capacity are out- Composition of the Bank lined. Pools less than 30 m in diam- M -Mud, clay, or soil other than sand or eter with fish holding capacity are gravel marked as a point type. L - Ledge or bedrock -Exposed streambeds over 30 m in G -Sand, gravel, and stones less than 13 diameter during low water in the cm (5 in.) in diameter summer are outlined. Exposed R - Rock materials from 13 to 38 cm (5 to streambed less than 30 m in diame- 15 in.) in diameter ter are marked as a point type. PHOTOCRAMMETRIC ENGINEERING & REMOTE SENSING, 1979

Ri -Rills with white water and some ex- shade types, nine river bank composition posed rocks with diameters greater types, and four bank elevation types. than 38 cm on streams over 30 m Streambanks were classified as a linear type wide are outlined. Rills on streams on both sides for streams that maintain a less than 30 m wide are dashed lines minimum width of 3 m. Smaller streams with the Ri symbol. were classified as mountain streams, and no Rf - Riffles have no white water but the streambank or water characteristics infor- water is fast moving and rough with mation was added to that stream type. rocks of unknown size beneath the Land-use and vegetation types on 305 m water. Riffles are outlined on strips were, however, added to both sides of streams over 30 m wide. Riffles on mountain streams. streams less than 30 m wide are solid The water sub-system consisted of three lines with the Rf symbol. water characteristics of known value to fish. S -Mountain stream less than 3 m wide Pools over 30 m in diameter were outlined with fast moving, highly aerated and measured for area on the photographs. water under complete shade. This is Rocky rills with white water and rocky riffles a line type marked periodically with without white water were also outlined and the letter S and it is not assigned measured for area on streams over 30 m in streambank or water characteristics. width. On streams less than 30 m riffles and Q -Trees down in the water providing rills were shown as line types-riffles shown cover for fish are marked as a point as solid lines and rills as dashed lines. Both type with symbol shown. area and line types were recorded by towns, river tributary, and for the entire river sys- The classification system has three sub- tem. Pools less than 30 m in diameter were classes: vegetation and land use as area shown as point types without a boundary types for 305 m (1000 ft) strips adjacent to the and were tabulated by count. Other point river, streambank characteristics as linear types included partially submerged fallen types for both banks, and water characteris- trees and streambed that would be exposed tics shown as area, linear, or point types. The during low-water periods. land-use and vegetation classification sub- system used on the 305 m strips (MacCon- nell, 1975) was devised for New England and has been applied to towns and counties The boundaries of 305-m strips of land on (MacConnell and Niedzwiedz, 1975),and on each side of the river were marked on 112 a statewide basis for Massachusetts (Mac- alternate aerial photographs in the 32 flight Connell, 1975) and Rhode Island (MacCon- strips with a segment of the river or tributary nell, 1974). No interpretation problems oc- centered in each flight line. Photo interpret- curred with the well tested land-use and ers then spent several weeks in Vermont to vegetation types on the 1:12,000 scale aerial reconnoiter the river, verify the classifica- photographs used. This classification system tion system, and establish thousands of that describes the nature of the land, vegeta- ground truth points so that the classification tion on the landscape, and land use is so system could be consistently and accurately similar to other systems in use that there is applied under all conditions. Interpreters, no need to describe the 104 types here. all experienced with the vegetation and The streambank and water classification land-use part of the classification system, developed for this study contains character- concentrated on the more difficult water and istics of value to biologists making fish and river bank types. During the interpretation wildlife habitat studies. The streambank work several more trips were made to the classification sub-system is similar to one White River watershed to ground check devised by MacConnell and Archey (1969) work completed and to reconnoiter and es- for classifying the banks of the Connecticut tablish more ground truth points in new River in a recreation study. Streambanks are areas to be typed. Subsequent summertime described by four characteristics indicated trips were valuable in checking exposed by a letter or number: evidence of channeli- river bottom. zation on the banks, amount of shade pro- One of the interpreters had experience vided to water by trees on the streambank, with a streambank classification system but composition of the bank, and elevation of the the water classification, developed for use in streambank above the water. Eighty-six this study, was new to everyone. Interpreta- streambank types were found on the river by tion of streambank composition posed few combining one channelization type, four problems after the extensive reconnaissance. REMOTE SENSING THE WHITE RIVER IN VERMONT

For example, sand and gravel bank areas are caused turbulence or riffle areas that appear consistently displayed in light tone and fine mottled in tone on aerial photographs. In- texture on the photographs while other Ver- creased water depths of riffles increased mont soils encountered were dark-toned. light absorption, causing them to register in Channelized areas were easily identified generally darker tones than rill areas. Rocks because of their light tone, fine texture, in riffles and rills provide good cover for fish. smooth surface, and gentle gradient. A fea- The mountain stream designation was as- ture common in channelized areas was pres- signed to those streams that averaged less ence of bulldozer blade deposits that appear than 3 m in width. This designation was as linear mounds of light-toned gravel lo- used because the streams were too small to cated along the outer reaches of the stream- apply stream bank or the water character bank. Man-made structures, such as rip-rap identifications. Mountain stream courses on and retaining walls, were easily identified this rugged terrain could be determined, but from their consistent linear or curvilinear trees on the streambanks usually masked nature. Ledge, large and smaller jumbled water characteristic information. Field re- rock, eroded streambank, and piles of woody connaissance indicated mountain streams debris left from the flood all had a consistent were characterized by fast moving, highly signature on the photographs. Determina- aerated water with low temperatures. These tion of streambank elevation required sim- streams had small pools, and gravel bottoms ple height measurements and abundant field and banks. Any stream less than 3 m in checks. All these interpretations were most width, covered by vegetation and located in difficult to make where the streambank was moderate to steeply sloping terrain, was partially masked with coniferous forest classified as mountain stream. growth, but additional field work in conifer- Pools, riffles, and rills were shown as area ous areas prevented errors. types where sufficiently large and as line or Identifying stream water characteristics point types where small. Trees down in the proved to be the most challenging interpre- water, available as cover for fish, were tive problem. Light penetration into water is marked as point types. Bottom exposed often the limiting factor in water related during low water periods was also annotated studies (Geary, 1967; Polcyn, 1969). Light as a point type. penetration is a function of water depth, water clarity, suspended particles in the water, and surface disturbance on the water that is in turn caused by wind or water ve- An extensive search was made for good locities. Disturbance of moving water is con- large-scale base maps of the White River so trolled by water depth and c:mposition of that land-use and water characteristics maps the bottom. Since the water characteristics at a scale of 1:12,000 could be made in classification was designed to measure fish quantity. Most suitable were USGS maps at habitat, many factors limiting light penetra- 1:24,000 and 1:62,000scales. The latter scale tion that have been a hindrance in other was blown up 6.25 diameters and tested for studies aided accurate identification of water cartographic accuracy. Enlarged 1:62,000 characteristics in the classification used scale maps proved too coarse and inaccurate here. For example, pools with fish holding for use as a base map at the 1:12,000 scale capacity are large, relatively smooth- chosen for this study. Next, some of the an- surfaced, and deep. Because little light was notated aerial photos were produced as reflected from these deep pools, they were Mylar transparencies with a resolution of 60 displayed as dark toned areas on the photo- linestcm (150 linestper in.) that could be re- graphs. Rills with white water were easily produced by Diazo process. Mylar photo- identified by the presence of large rocks copies did not show the detail required by above the water line, and light-toned white researchers and this technique was aban- water on the downstream side of the rocks doned. Land-use and water types were caused by turbulent surface water. Rills have added to the original photo negatives by a generally light-tonal appearance caused by tracing from the annotated photographs. On stream bottom reflectance through the shal- printing, the photographs show type bound- low water. aries (Figure 1) in white while all original Riffles are similar to rills except that in- detail of the photograph is preserved. Four creased water depths cover the rocks. Since copies of the annotated negatives were stream bottom materials do not penetrate the printed for use by various White River chan- surface water, no white water is present but nelization research teams. The U.S. Fish and the water is swift, and the rocky bottom Wildlife Service owns the negatives and PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1979 uses copies of them as a tool to train wildlife was in power line right-of-way. There were and fisheries biologists in aerial photoin- 6825 ha (2763 ac) of urban land that ac- terpretation. counted for 7 percent of the total land area within 305 m of the river or its tributaries. Forty-six percent of the urban land was Since no maps were made from the aerial single family dwellings, 12 percent indus- photographs, areas of land use and water trial or commercial, and 35 percent was in types were measured directly on annotated divided highway, an urban type in the sys- aerial photographs by dot grid. A dot grid tem used. Seven percent of the urban land with 100 dots per square inch was used, and was in public use. Approximately 1.5 per- each dot counted in a type represented 0.57 cent of the total study area was used for ha (0.23 ac). Linear distances of the line mining and waste disposal and about one- types were determined with a map measure, third of one percent was in outdoor recre- and point types were counted. Tabulations ation use. were separated by water course and towns and were summarized for counties and the CONCLUSIONS White River study area. One hundred four Standard panchromatic aerial photographs land use types were measured in 305 m at a scale of 1:12,000 can be used success- strips along the river, and 86 bank types, 5 fully to study stream alteration activities, water types, and 3 point types were also and to identify and annotate streambank and measured or counted and tabulated. stream water characteristics as well as to RESULTS identify vegetation types and land use. Suc- cessful analysis of a river system in northern The White River and its tributary streams climates requires photography taken in the in this study totaled 287 km (167 mi). Ofthis, spring after snowmelt and before the buds 19.3 km (12.1 mi) or 7 percent of total river on deciduous vegetation begin to swell. length in the study area were classified as Photography should be flown as soon as pos- channelized on aerial photographs. A severe sible after stream alteration activities have storm after channelization scoured away been completed in order to reduce risk of much of the channelization evidence before losing alteration evidence due to additional aerial photographs were taken. Con- flooding and erosion after original stream sequently, channelization data are conser- alterations have been completed. The vative. photographs used in this study were taken The total drainage system in the study area one year after stream alteration and after a included 4356 ha (1764 ac) of open water in second severe flood had erased much of the that part of the river system over 3 m wide. channelization evidence, making measure- Of this, 257 ha (104 ac) were in pools over 30 ment of channelization uncertain on either m in diameter, 788 ha (319 ac) were in riffles, the photographs or on the ground. Vegeta- and 526 ha (213 ac) were in rills. There were tion, land-use, and streambank types proved also 16.6 km (10.3 mi) of riffles and 11.4 km useful to wildlife biologists studying the im- (7.1 mi) of rills on streams less than 30 m pact of channelizing on songbird and small wide. There were 174 pools less than 30 m in mammal populations adjacent to the river diameter, 148 trees down in the water, and (Possardt and Dodge, 1978). Fisheries 220 areas where the bottom would be ex- biologists have also found streambank and posed in the low water period during the water characteristics of value in studying summer. fish habitat (Dodge et al., 1976). Vegetation land-use data on the 305 m strips bordering the river and its tributaries in the study area included 104 types from the Massachusetts Classification System. There Research reported herein was funded by were a total of 96,910 ha (39,235 ac) in these USDI, Fish and Wildlife Service Contract strips. 47,550 ha (19,251 ac) or 49 percent is 14-16-0008-1149. This is a contribution of forest; 46 percent of the forest is hardwood, the Massachusetts Agricultural Experiment 31 percent mixed wood, 22 percent Station and the Massachusetts Cooperative softwood, and 1 percent was forest planta- Wildlife Research Unit jointly supported by tion. Forty-one percent or 39,357 ha (15,934 the U.S. Fish and Wildlife Service, Mas- ac) in the strips was in agricultural use. Of sachusetts Division of Fisheries and this, 34 percent was tilled, 65 percent was in Wildlife, University of Massachusetts, and pasture or abandoned fields, and 1 percent the Wildlife Management Institute. REMOTE SENSING THE WHITE RIVER IN VERMONT 1399

REFERENCES Agr. Expt. Sta., Univ. of Massachusetts, Bunik, J. A., and K. A. Turner, 1971. Remote Amherst, 79 pp. sensing applications to the quantitative MacConnell, W. P., and W. Archey, 1969. a he use analysis of drainage networks. Proc. Fall Mtg. of aerial photographs to evaluate the recre- Am. Soc. Photogrammetry, Falls Church, Va. ational resources of the Connecticut River in 71-313. Vermont. Bull. 575, Mass. Agr. Expt. Sta., Coleman, J. M., 1969. Brahmaputra River: chan- Univ' Amherst' 85 pp' riel processes and sedimentation. Sedimen- MacConnell, W. P., and W. R. Niedzwiedz, 1975. tary Geology. 3:131-239. Remote sensing 20 years of change in Wor- Coleman, J. M., and W. G. McIntire, 1971. Tran- cester County Massachusetts, 1951-1971. siting coastal river channels. International Bull. 625, Mass. Agr. Expt. Sta., Univ. Mas- Hydrographic Review 48:13-43. sachusetts, Amherst, 173 pp. Dodge, W. E., E. E. Possardt, R. J. Reed, and Meyer, W., and R. I. Welch, 1975. Water resources W. P. MacConnell, 1976. Channelization As- assessment. Manual Remote Sensing, Am. sessment, White River, Vermont: Remote Soc. Photogrammetry, Falls Church, Va., sensing, benthos, and wildlife. Bordeaux Co., 1479-1551. West Springfield, Mass. 74 pp. Piech, K. R., and J. E. Walker, 1972. Thematic Geary, E. L., 1967. Coastal hydrography. Photo- mapping of flooded acreage. Photogrammet- grammetric Engineering. 34:44-50. ric Engineering 38: 1081-1090. Hoyer, B. E., G. R. Hallberg, and J. V. Taranik, Polcyn, F. C., 1969. Depth determination by 1973. Seasonal multispectral flood inundation measuring wave surface effects. NASA Ann. mapping in Iowa. Proc. Symposium Manage- Earth Resources Program Review. ment and Utilization of Remote Sensing Possardt, E. E., and W. E. Dodge, 1978. Stream Data. Am. Soc. Photogrammetry, Falls channelization impacts on songbirds and Church, Va., 130-141. small mammals in Vermont. Wildlife SOC. MacConnell, W. P., 1974. Remote sensing land- Bull. 6:18-24. use and vegetative cover in Rhode Island. USDA, ASCS, 1969. Aerial photography specifi- Bull. 200, Coop. Ext. Service, University of cations, (ASCS-AP-201 Revision 4), USDA, Rhode Island, Kingston, 93 pp. ASCS Washington, D.C. 43 pp. , 1975. Remote sensing 20 years of (Received October 18, 1978; revised and ac- change in Massachusetts. Bull. 630, Mass. cepted June 25, 1979)

The Third Conference on the Economics of Remote Sensing Information Systems Hyatt Lake Tahoe Incline Village, Nevada November 6-8, 1979

The central theme of this conference series, sponsored by the University of Nevada, Reno, San Jose State University, and Resources Development Associates, is the cost, effectiveness, and quantified benefits of information systems involving remote sensing technology. Repre- sentative topics within the scope of this conference are

8 comparison of remote sensing platforms for natural lewurce explorat~onand mapping, 8 the problems and plomlw of geographic lnfo~mationsystem\, 8 the role of government and private industry in land-use Inventory and planning, 8 the value and limitations of remote ~en\ingin pollut~onmonitoring, 8 the expected cost of implementing agricultural surveys; and 8 the relative effectiveness of aerial photography and manual or computer processed and sat data for resource5 survey. For further information please contact Ms. Terri Wise Third Conference Coordinator Resources Development Associates P.O. Box 239 Los Altos, CA 94022 (415) 961-7477