Water Quality

The Effects of Vegetation Removal on Water Quality:
Implications for Management of Municipal Watersheds
in the Northeastern United States

A term paper prepared for Dr Donald Mader
Forestry 528: Forest Hydrology
University of Massachusetts, Amherst

Karl Davies
May, 1984

I.  Introduction

Forest watershed research accumulated over the past 30-40 years has shown that removal of forest vegetation by timber harvesting can increase watershed streamflow by significant amounts.  Douglass and Swank (1975) developed equations to predict first year and subsequent years  increases in water yield for hardwood forests in the Northeast:

    (1) AQ  = (x1/x2)b  first year change in yield
    (2) AQ1  = AQ + c(log Ti) subsequent changes in yield
    where AQ = first year increase in streamflow in inches
     x1 = percent of basal area cut
     x2 = potential annual insolation in langleys
    a,b,c  = coefficients
     AQ1  = increase in streamflow in year i
    T  = duration of yield increase

Increase in yield is a function of percent of basal area cut, insolation index (a function of the aspect and slope of the watershed), and length of time after the cut.  Heavier cutting yields more water, especially on steep, north-facing watersheds, and yield increases last longer with heavier cuts.  Use of the above equations allow some general statements about watersheds in the Northeast:  On fairly level watersheds clearcutting will increase water yield 40-50% in the first year and increases will last for about 20 years.  Cutting 50% of the forest vegetation will increase yield 15-20% in the first year and increases will last about 7 years.

Cutting of vegetation reduces the amount of precipitation lost to the atmosphere through evapo-transpiration (ET)   Some of the precipitation intercepted by trees is lost to evaporation: trees and other plants transpire heavily during warm weather in order to maintain temperatures appropriate for their metabolic activities.  Most of the increase in water yield from cutting occurs during summer months when demand for water is greatest.  August, September and October streamflows should be roughly 100% greater the first year after cutting all vegetation (Douglas and Swank 1975).  Lesser increases will be observed in all other months as well.

ET is a function of the amount of leaf surface area, the darkness (albedo) of the leaves, the length of time the leaves are on the trees, and the species' rates of transpiration.  Conifers have higher leaf areas than hardwoods--by a factor of about 3; they also have darker colors and retain their leaves year round (Douglas and Swank 1975).  While conifers' transpiration rates are lower then hardwoods', their total yearly transpirations are considerably greater then hardwoods.  Hardwoods in turn have greater yearly transpirations than shrubs or grasses which have much lower leaf surface areas.

The possibility of lowering ET and thereby increasing streamfiow by cutting forest vegetation is attractive to municipal watershed managers, especially when faced with potential summer water shortages.  Costs of procuring additional water from deep wells or other sources can be high and the quality of deep ground water may be suspect, particularly in the light of recent discoveries of many different chemical contaminants in ground water. 

Revenues from the sale of timber are additional incentives for forest management on municipal watersheds. Nevertheless, most watersheds are not managed for increased water or timber yield.  The primary reason for this is concern over possible loss of water quality due to cutting of trees.  This paper will address the principal areas of concern over possible loss of water quality from cutting of forest vegetation.

II.  Suspended Sediment

Patric (1977) emphasized the distinction to be made between "tree cutting" and "logging":

    "Tree cutting is severing the aerial portion of a tree from its roots.  Logging is cutting out the useful parts of severed trees, gathering them and delivering them to the marketplace.    In practice, tree cutting and logging are inseparable but their potentials for causing erosion differ enormously."

Tree cutting in and of itself causes no erosion of forest soils (Hewlett 1974).  Forest soil infiltration rates are high enough to absorb all rainfall and thus prevent overland flow, which is the means by which soil particles are detached and carried to streams, resulting in sedimentation and loss of water quality.  Forest soils have high infiltration rates because the thick layers of leaf litter and other accumulated organic matter have high porosity and the ability to absorb water rapidly.  Unusual erosion of forest soils occurs only when the protective organic layers are removed, and especially when the exposed mineral soils are compacted by logging equipment.  Compaction prevents infiltration and causes overland flow.

Normal sediment erosion in undisturbed forests occurs only along stream channels, especially at times of peak flow (Patric 1977). Some erosion also occurs as water carries off dissolved minerals but this occurs as water percolates through the soil and not over it (this aspect of water quality will be discussed in section III below).  Patric (1976) estimated that the average annual total erosion loss from eastern forest land is only about .1 ton/acrel year.  This rate is much less than the rates for pasture or corn/hay rotation which are about .3 and 2.8 tons/acre/year, respectively (Brady 1974).

Removal of organic layers may result from winching (dragging) tree stems from where they fall to the skidder (tractor).  Pulling bunches of stems behind the skidder has the same effect. On skid trails which are frequently used, the organic layers are typically entirely removed and mineral soils are compacted.  At the log landing where stems are bucked (cut) into logs and loaded onto trucks for transport to the mill, additional surface soil removal and compaction occurs.  And where truck roads are constructed to give access to log landings, further severe exposure and compaction occurs.

Recovery from logging disturbance occurs naturally and rapidly in the humid Northeast.  Grass, weeds, shrubs and trees will become established soon after cessation of logging, but it may be 1-2 years before erosion is curbed (Patric 1977).  For this reason, exposed soils are normally seeded to grass after logging in sensitive areas.  Other measures are recommended for surface erosion control:  minimizing number of skid and truck roads; locating roads as far as possible from stream channels; use of water bars, water turn-outs, culverts, and grade restrictions to prevent water from running down or across roads; skidding on contours away from streams; avoiding operations on soft soils during wet weather; and avoiding any disturbance of stream channels (Patric 1977 and 1980, Ursic and Douglas 1978).

Recent studies have shown that the above measures will keep suspended sediment levels under 2 NTU (nephlamine turbidity units) in nonstormflow periods on clearcut watersheds (Patric 1980).  Since the current drinking water standard for turbidity is 1 NTU at the point of entry into a water distribution system (Shelton 1983), small clearcuts should have negligible effects on overall water quality.  Patric's data (1980) also indicated virtually no increase in average turbidity from light selection cuts which removed 25-30% of the basal area.  However, turbidities did increase during storms regardless of cutting intensity and were traced to muddy logging roads.

Other studies (Packer 1966, Douglas and Swank 1975) have indicated that poorly designed and poorly located roads are the principal cause of deterioration in water quality.  Careful planning of roads, supervision of logging crews, plus quick revegetation of exposed soil surfaces should therefore prevent suspended sediment from lowering water quality.  Partial cutting of trees would tend to lessen the area in roads and it would also lead to smaller increases in streamflow and therefore less risk of erosion from stream channels.

As suggested in section I, above, ET may be reduced by converting conifers to hardwoods and by converting conifers or hardwoods to grass.  However, such conversions do not occur without sacrifices in water quality.  Ursic and Douglass (1977) found that hardwood forests and grasslands yield about four times as much sediment as pine forests in the Coastal Plain states.  Kling and Olson (1975) found that similar relations exist for hardwood forests and pine plantations in east-central New York State, and that pastures and brushlands yield respectively 1 and 2 times more sediment than hardwood forests.

Both of the above cited studies were conducted in regions having large areas of recently abandoned, eroded agricultural land.  Hence, the levels of erosion were higher than the norms found for forested watersheds by Patric (1976).  With time and opportunity for stream channels to stabilize, erosion differences between cover types might decrease.

III.  Dissolved Minerals

After the cutting of trees, erosion is likely to increase by the export of mineral nutrients which are dissolved in water percolating through the soil. This is especially the case with nitrogen. Vitousek et al. (1979) briefly summarized how nitrogen losses occur:

    "After destructive disturbance, increased soil temperature and moisture availability accelerate mineralization of nitrogen ..., but nitrogen uptake by vegetation is reduced or eliminated.  The nitrogen mineralized immediately after destructive disturbance could be lost to streamwater or groundwater."

Also, logging residues left on the forest floor decompose, making additional nitrogen available (Roskoski 1980).  Other minerals have insignificant and inconsistent losses after cutting (Martin et al 1981).  Nitrogen (as nitrate N03) losses are much more pronounced in the northern hardwood forests of northern  New England (Vitousek et al 1979, Martin et al 1981).  Studies conducted at the Hubbard Brook Experimental Forest in New Hampshire (Likens et al. 1970) caused some alarm among environmentalists in the early 1970's when large nitrate losses were shown to result from clearcutting and chemical prevention of revegetation.  Streamwater concentrations were well above the 10 ppm standard for drinking water (Shelton 1983).

Since the early 70's several researchers have indicated that such high nitrate losses do not occur in other parts of the East (Aubertin and Patric 1974, Douglas and Swank 1977, Martin et al 1981)  and that even in New Hampshire much lower losses result when revegetation is not prevented (Martin and Pierce 1980).  Swank and Waide (1980) showed that much more nitrogen is stored in the thick forest floor organic layers at Hubbard Brook than at other experimental watersheds.  Thus, much more could be lost by leaching after clear-cutting.  Losses after clearcutting are in fact directly proportional to amounts of nitrogen in the forest floor.

In their recommendations for harvesting of watersheds in New Hampshire, Martin and Pierce (1980) suggested use of buffer strips along streams, less cutting in upper portions of watersheds, and progressive strip cutting of every third strip along the contour perpendicular to the stream channel.  These measures were found to very significantly reduce stream concentrations of nitrate and other nutrients and keep them within water quality standards.  Patric (1980) found that light selective cutting resulted in no significant loss of dissolved minerals in West Virginia, while clearcutting resulted in only slightly increased nutrient concentrations. 

These studies indicate that partial cutting will lower nutrient concentrations in stream water flowing out of watersheds with soils having thick organic layers (New Hampshire), but that nutrient concentrations are little influenced by intensity of cutting on watersheds with less organic matter (West Virginia).

Watersheds having thin or moderate organic soil layers are unlikely to have nitrate concentrations in excess of drinking water quality standards, regardless of intensity of cutting.  But clearcutting in excess of 20% of coniferous watersheds may significantly increase stream water acidity (Martin et al. 1981). And converting forest land to grasses will greatly increase nitrate and other nutrient losses (Swank and Douglass 1977) if the usual practices of liming and fertilizing are employed. 

Other intensive forestry practices, such as site preparation by clearing and discing, followed by tree planting, will probably also increase nutrient losses (Douglass and Swift 1977) over the short term. But pine plantations, once established, will reduce nutrient losses compared to grass and coppice hardwoods (Swank and Douglass 1977).

IV.  Water Temperature

Cutting all trees in the riparian zones along streams will increase water temperatures 5-10F (Patric 1980, Lynch et al. 1980).  Temperatures will be lowered by downstream cover.  And regrowth of riparian vegetation will restore water to pre-cutting temperatures in 3-4 years for small streams (Patric 1980):

    "The 3 ha protective strip completely shaded the stream channel preventing any rise in   stream temperature through 1972. . . . After this strip was cleared during the winter of 1972-73, the water temperature increased. . . . By 1975, dense revegetation had begun to reshade the channel and the stream . . . was not more than 2.20C warmer . . . Channel shading was sufficient by 1977 to return stream temperatures to pre-clearcutting levels."

Increased water temperature, in conjunction with increased nutrient levels in stream water, will increase growth of certain algae (Lynch et al. 1980, Martin et al. 1981); also, growth of flowering plants on stream bottoms will increase.  More primary production will mean higher populations of macroinvertebrates, especially mayflies (Lynch et al. 1980).  Higher temperatures may also lead to lower levels of dissolved oxygen in stream water; this change may negatively impact populations of stream invertebrates and vertebrates.  If suspended sediments are also present, feeding habits may also be negatively affected.

Water temperatures may be controlled by leaving buffer strips of uncut trees along the edges of streams.  Buffer strips must be of sufficient width to prevent heat and light from entering from adjacent cutover areas; 50-100' should be wide enough.  Buffer strips may also absorb dissolved nutrients in ground water flowing into streams (Patric 1977).  However, Swank and Caskey (1982) suggested that higher biological activity of algae and bacteria in streams may reduce nitrate levels after clearcutting; additional nitrate is denitrified by stream sediments.  Thus, buffer strips may have a net negative effect on drinking water quality if they reduce primary production too much by keeping temperatures low.

V.  Discussion

Most watershed research has focused on the effects of clearcutting forest stands: The reduced efficiency of partial cutting, particularly under unevenaged management, has generally been attributed to the ability of the remaining vegetation to consume nearly as much water as the fully intact forest.  This is possible because the remaining vegetation has roots which may extend for considerable distances into opened areas.  Increased crown exposure may also result in greater transpiration rates in the residual vegetation. (Mrazik et al. 1980)

Furthermore, research has been done predominantly with hardwood stands on steep slopes.  Since many municipal watersheds in the Northeast have very significant areas in softwood cover, since most are located on rolling rather than steep topography, and since clearcutting is generally not practiced for aesthetic and/or timber production reasons, the applicability of most of the research may be indirect.  In general, effects on water quality of partial cutting of softwoods or mixed stands on rolling topography should be significantly less than the effects of most experimental cuttings.

Patric's research in West Virginia (Patric 1980) indicated what might be expected to result from light, selective cuttings: insignificant or no increases in suspended sediments, dissolved minerals, and stream temperature.  This is because the residual stands protect the forest floor from increased solar heating and decay of organic matter; they also absorb what nutrients may be released and they filter out much of the sediments which may result from logging scarification and road construction.  Furthermore, since streamflow from partial cuts is not as great as from clearcuts, the risk of sedimentation from channel scour is much less.

Partial cutting will not increase water yields as significantly as clear-cutting over the short term, but periodic partial cutting should increase water yields 5-15% over the long term (Douglass 1974).  Summer yields should increase 10-30% from partial cutting.  Partial cuttings will normally yield important financial returns when they are part of a timber production plan, while periodically repeated clearcuts may yield little or no return after the initial harvest. This is because very little commercial value accrues in the first 20-30 years of regeneration following a clearcut and, as stated in Section I, 20 years is the length of time for water yield increases from a clearcut.

Overall comparisons of water plus timber yields resulting from different cutting systems have not been made, although computer simulations have indicated that optimal combined value production occurs with a 12 year partial cutting cycle on a 120 year rotation for a Ponderosa pine covered watershed in Colorado (Black 1963).

Regardless of the cutting practice, careful attention should be given to planning and supervising the cut, particularly where road construction is involved. Removal of protective cover should be minimized and quick revegetation of exposed surfaces should be assured.  These criteria would apply especially to more erodable, fine textured soils.  If forest soils have high levels of organic matter, partial (strip or selective) cuts should be used, especially when subsurface soils have little capacity for resisting loss of dissolved minerals (Martin et al. 1981).  In general, careful attention should be given to soil characteristics in planning road construction and silvicultural prescriptions.

Cutting softwoods on watersheds should have a much greater effect on water yield than cutting hardwoods (Douglass and Swank 1975), while impacting water quality to a lesser extent (Martin et al. 1981).  Since conversion to hardwoods is often a natural result of cutting softwoods in the humid Northeast  such natural conversions should increase water yields 30-40% over the long term (extrapolated from Douglass and Swank 1975), but at a slight sacrifice in water quality (Swank and Douglass 1977)

Conflicts between water quantity and quality occur to a much greater extent when forests are converted to grass (Kling and Olson 1975, Swank and Douglass 1977, Ursic and Douglass 1978).  Seedbed preparation exposes soils to erosive forces and soil amendments added to assure establishment are likely to leach out into stream water.  Once established, the grass cover is less effective in preventing erosion of sediments and dissolved minerals than are forest covers. 

The increased water yields reported by Swank and Douglass (1975) were in the range of 15-25% when the grass was not growing vigorously, and less than that when fertilization increased growth.  These yield increases are relative to yields from previous hardwood forest cover.  Much greater yield increases would have presumably resulted from converting softwoods to grass.  But the value of the increased yields is suspect when loss of water quality is considered.

The actual quantities of suspended sediments and dissolved nutrients eroding from forests converted to grass have not been documented for the periods immediately following conversion, although research is in progress (Douglass and Swift 1977). Good data are also lacking for softwood to hardwood conversions. Other aspects of water quality relations needing more research are the relative effects of cutting on north- and south-facing slopes, and the effects of various residual stand densities.

It should be kept in mind that a watershed managed for sustained yield of timber plus increased water run-off would only have small areas cut each year. With a 12 year partial cutting cycle, as suggested by Black (1963), only 50% of the volume on 1/12th of the watershed would be cut each year.  If the effects of cutting on water quality disappear after 3-5 years (Lynch et al. 1980), then only 1/4-2/5 of the watershed would have higher erosion rates at any particular time. The effects of dilution from water without increased sediments or minerals would reduce any possible negative impact.

VI.  Conclusions

Careful clearcutting will substantially increase water yields without exceeding water quality standards in most areas in the Northeast, even immediately after cutting:

    "Tree cutting reduced evaporation from forest land and increased heating of the forest   floor thereby increasing streamflow and the outflow of plant nutrients dissolved in it.    Channel clearing raised water temperature and with logging roads, increased stream sediment.  The regrowth returned most of these conditions nearly to precutting levels within 5 years. . . . Throughout the experiment, loading of solids dissolved and suspended  in streamfiow was within the range of variation accepted as the geologic erosion rate." (Patric 1980)

While levels of sediment and minerals in streamwater may be statistically significant, they are not likely to exceed drinking water quality standards. Exceptions to this general rule will only occur when trees on soils with high levels of organic matter are heavily cut, and when careful logging practices are not followed.

Partial cuttings on a 10-20 year cycle will increase water yields 5-15% over the long term with a much lower risk of loss in water quality.  Water yields will increase 10-30% during summer months.  Partial, periodic cuttings are more likely to optimize combined values of water and timber yield from municipal watersheds, and the effects of dilution from uncut areas of the watershed should eliminate any negative impacts on water quality.

Converting softwoods to hardwoods, and forests to grass will increase water yields, but with some loss in water quality.  Converting forests to grass is probably undesirable because of the water quality problem, but converting softwoods to hardwoods should not significantly impair water quality.

Watershed managers in the Northeast would be well-advised to consider forest management practices as practical means of increasing water yield, especially in summer months, without sacrificing water quality.  Interpretation and application of existing research on watershed management should prove to be profitable in terms of increased yields of water and timber from municipal watersheds.


Aubertin, G.M. and J.H. Patric.  1974.  Water quality after clearcutting a small watershed in West Virginia.  J. Environmental Quality 3(3).

Black, Peter E.  1963.  Timber and water resource management.  Forest Science 9(2).

Brady, Nyle L.  1974.  The Nature and Properties of Soils (8th Edition, Chapter 9). MacMillan.  New York.

Douglass, James E.  1974.  Watershed values.  J. of Forestry 72(10).

Douglass, James E. and Wayne T. Swank.  1975.  Effects of management practice on water quality and quantity:  Coweeta Hydraulic Laboratory, North Carolina. in USDA For. Serv. Gen. Tech. Rep. NE-13.  Broomall, PA.

Douglass, J.E. and L.W. Swift, Jr.  1977.  Forest Service studies of soil and nutrient losses caused by roads, logging, mechanical site preparation and prescribed burning in the Southeast.  in Watershed Research in Eastern North America.  Smithsonian Institution, Edgewater, MD.

Hewlett, J.D.  1974.  Comments on letters relating to role of subsurface flow in generating surface runoff to upstream source areas by R. Allan Freeze. Water Resources Res. 10(3).

Kling, G.F. and G.W. Olson.  1975.  Role of computers in land use planning. Info. Bull. 88.  N.Y. St. Coll. Ag. and Life Sci., Cornell U., Ithaca, NY.

Likens, G.E., F.H. Bormann, R.S. Pierce, J.S. Eaton, and N.M. Johnson. 1970. Effects of forest cutting and herbicide treatment on nutrient budgets in the Hubbard Brook watershed ecosystem. Ecol. Monogr. 40.

Lynch, James A., Edward S. Corbett, and William E. Sopper.  1980.  Evaluation of management practices on the biological and chemical characteristics of streamflow from forested watersheds. Inst. for Res. on Land and Water Resources.  PA St. U., University Park, PA.

Martin, C. Wayne and Robert S. Pierce.  1980.  Clearcutting patterns affect nitrate and calcium in streams of New Hampshire. J. of Forestry 78(5).

Martin, C. Wayne, Diane S. Noel, and C. Tony Federer.  1981.  The effect of forest clearcutting in New England on stream-water chemistry and biology. Res. Rep. 34.  Water Resources Research Center, U. of NH, Durham, NH.

Mrazik, Brian R., Donald L. Mader, and William P. MacConnell. 1980. Integrated watershed management: an alternative for the Northeast. Res. Bull. 664. MA Ag. Exp. Sta., U. of MA, Amherst, MA.

Packer, P.E.  1967.  Forest treatment effects water quality.  in Forest Hydrology. W.E. Sopper and H.W. Lull, eds.  Pergamon Press.  New York.

Patric, J.H.  1976.  Soil erosion in the eastern forest.  J. of Forestry 74(10).

Patric, James H.  1977.  Soil erosion and its control in Eastern woodlands. Northern Logger 25(11).

Patric, James H.  1980.  Effects of wood products harvest on forest soil and water relations.  J. of Environmental Quality 9(1).

Roskoski, Joann P.  1980.  Nitrogen fixation in hardwood forests of the northeastern United States.  Plant and Soil 54.

Shelton, Theodore B.  1982.  Interpreting water quality analysis:  What do the numbers mean?  Coop. Ext. Serv. St. U. of NJ, Rutgers U., New Brunswick, NJ.

Swank, Wayne T. and William H. Caskey.  1982.  Nitrate depletion in a second-order mountain stream.  J. of Environmental Quality 11(4).

Swank, W.T. and J.E. Douglass.  1977.  Nutrient budgets for undisturbed and manipulated hardwood forest ecosystems in the mountains of North Carolina. in Watershed Research in Eastern North America.  Smithsonian Institution, Edgewater, MD.

Swank, Wayne T. and Jack B. Waide.  1980.  Interpretation of nutrient cycling reserach in a management context:  evaluating potential effects of alternative management strategies on site productivity.  in Proc. 40th Ann. Biol. Colloq. OR St. U. Press, Corvallis, OR.

Ursic, Stanley J. and James F. Douglass.  1978.  The effects of forestry practices on water resources.  in the W. Kelly Mosley Env. Forum Proc. Auburn U. Press, Auburn, AL.

Vitousek, Peter M., James R. Gosz, Charles C. Grier, Jerry M. Melillo, William A. Rivers and Robert L. Todd.  1979.  Nitrate losses from disturbed ecosystems.  Science 204.