Watershed hydrology and catchment response

Watershed hydrology and catchment response

Motivation Studying stream response to water input events (catchment response) is important for: Water supply: Precipitation falling in the ground and then moving through stream networks represents a resource for agricultural, municipal and recreational water use. Flood prediction and forecasting: Flood forecasting, the design of dams, bridges and levees, and developing floodplain management plans requires knowing how streams will respond to precipitation events. Water quality: Water quality is strongly influenced by chemical and biological reactions as water moves over the land surface and then into streams.

Watershed hierarchy Small watersheds (e.g., a few km 2 ) can be part of bigger watersheds, and bigger watersheds can be part of even bigger watersheds. The example at right is the Patuxent river watershed, which drains into Chesapeake Bay. The biggest watershed in the U.S.A. is the Mississippi watershed; our own Boulder Creek watershed (or drainage) is part of the South Platte drainage, which is part of the Missouri drainage, which is part of the Mississippi drainage. http://www.uvm.edu/giee/av/pubs/hydro2/pap_hydro2.html

Viewed with respect to the Mississippi drainage as a whole, the Boulder Creek drainage seems very small. Boulder Creek is itself comprised of a number of smaller watersheds. Watershed hierarchy (cont.) http://czo.colorado.edu/html/sites.shtml http://www.salemstate.edu/~lhanson/gls100/gls100_hydro.htm

Stream order Streams can be defined in terms of their order. The joining of two first-order streams creates a second-order stream. The joining of two second order streams creates a third-order stream, and so forth. This system was developed by Strahler (1952), modified from Horton (1945). The Ohio River is an 8 th order stream, the Mississippi is a tenth order stream. The Amazon, the worlds largest river, is a 12 th order stream. Anything 7 th order and up is typically viewed as a river, but this is not a hard and fast rule in practice. http://www.epa.gov/owow/watershed/wacademy/acad2 000/stream/stream11.html

Drainage patterns Dendritic: The most common form. It develops in regions underlain by homogeneous material (the subsurface geology has a similar resistance to weathering so there is no apparent control over the direction the tributaries take). Tributaries join larger streams at acute angle (less than 90 degrees) Parallel: Forms where there is a pronounced slope to the surface. A parallel pattern also develops in regions of parallel, elongate landforms like outcropping resistant rock bands. Tributary streams tend to stretch out in a parallel-like fashion following the slope of the surface. A parallel pattern sometimes indicates the presence of a major fault that cuts across an area of steeply folded bedrock. Trellis: Develops in folded topography like that found in the Appalachian Mountains of North America. Synclines form valleys where the main channel of the stream is found. Short tributary streams enter the main channel at sharp angles as they run down sides of parallel ridges (anticlines) Tributaries join the main stream at nearly right angles. http://www.uwsp.edu/geo/faculty/ritter/geog101/textbook/fluvial_systems/drainage_patterns.html

Drainage patterns (cont.) Rectangular: Found in regions that have undergone faulting. Streams follow the path of least resistance and thus are concentrated in places were exposed rock is the weakest. Movement of the surface due to faulting off-sets the direction of the stream. As a result, the tributary streams make shape bends and enter the main stream at large angles. Radial: Develop around a central elevated point. This pattern is common to conically shaped features such as volcanoes. The tributary streams extend the headward reaches upslope toward the top of the volcano. Centripetal: The opposite of the radial as streams flow toward a central depression. Pattern is common in the Basin and Range province of the United States where many basins exhibit interior drainage. Deranged: Develop from the disruption of a pre-existing drainage pattern, such as when a dendritic pattern is overrun by a glacier. http://www.uwsp.edu/geo/faculty/ritter/geog101/textbook/fluvial_systems/drainage_p atterns.html

Measuring streamflow Measuring streamflow (discharge) generally involves four steps: 1) Measuring stream stage: This involves continuous measurement of the height of the water surface at a location along a stream with reference to some established altitude (close to the stream bed) where the stage is defined as zero. 2) Measuring discharge: Instruments are used to measure stream discharge 3) A relationship is formed between measured stage and measured discharge 4) Once the relationships is formed, then one can use stage information to get discharge. One must monitor the stage/discharge relationship as it may change. http://ga.water.usgs.gov/edu/measureflow.html

Measuring streamflow (cont) A typical USGS stream gauge that measures stage (http://ga.water.usgs.gov/edu /streamflow1.html) http://ga.water.usgs.gov/edu/streamflow2.html

Measuring streamflow (cont.) http://ga.water.usgs.gov/edu/streamflow3.html http://ga.water.usgs.gov/edu/streamflow3.html

Daily Snowmelt Input Rain / Snowmelt

Streamflow response is rapid in small catchments. Not all event precipitation becomes runoff Response to event is variable.

Basic aspects of stream response Consider a small upland watershed. Precipitation may move into a stream from overland flow or as a groundwater flow. The streamflow (the flow that would be gauged at a particular point) is a spatially and temporally integrated response to: (1) spatially and temporally varying water input rates; (2) the time it takes for a given drop of water to travel from where it strikes the surface of the watershed to where it enters the stream network; (3) the time it takes for the water to travel from where it enters the stream channel to the point of measurement. Dingman 2002, Figure 9-1

Flowpaths

A typical storm response The figure at the right shows a typical hydrograph of a stream in response to a isolated precipitation event of significant magnitude and areal extent. At the gauging site, one is measuring the passage of a flood wave. The nature of the this wave (shape, amplitude, duration) depends on precipitation intensity, extent, duration (as seen in the hyetograph) and form, topography, soils (which affect infiltration), and the density of the stream channel network. In a small (less than 50 km 2 ) watershed, the travel time of water to a watershed outlet is primarily a function of the hillslope travel time; for larger watersheds, travel time in the stream network is more important. If the precipitation event is snowfall, there can be a long lag (even seasonal) in streamflow response. Dingman 2002, Figure 9-2

Some terminology Watershed, catchment, drainage basin: three names for the same thing Hydrograph: A graph of stream discharge at a point (x axis) by time (y axis) Hyetograph: A graph of water input (y axis) versus time. Peak discharge: The value of maximum streamflow in response to a water input event that follows the hydrographic rise and is then followed by the hydrographic recession. hydrograph watershed hyetograph http://water.me.vccs.edu/math/tabular.htm commons.wikimedia.org http://echo2.epfl.ch/vicaire/mod_1b/chapt_2/main.htm

Contributing area The water in a stream identified as the response to a given precipitation event can originate from only a small fraction of the watershed,; this fraction is termed the contributing area. For an event such as an isolated thunderstorms, the contributing area could be small (several km 2 ); for a more general precipitation event, the contributing area can be much larger (hundreds of even thousands of km 2 ), which, depending on the size of the watershed being considered, may cover most or all of the watershed. The extent of and location of the contributing area may change during the course of a precipitation event. https://www.meted.ucar.edu

Hydrograph response The figure at right shows hydrographs for a series of gauging stations progressively downstream along the Sleepers River in Danville VT in response to a intense rainfall event (hyetograph at the top). Note how the hydrograph shape for the smallest (top) watershed is closely allied with the shape of the hytograph. Lower downstream, the hydrographs are increasingly affected by tributary inputs and storage effects of the stream channels, leading to an increase in the lag time between precipitation input and the hydrograph peaks, as well as smoother hydrographs. Dingman 2002, Figure 9-3

Hydrograph separation Streamflow can be separated into the contribution from event flow Q ef taken to be the direct response to a given water input, and a base flow Q b not associated with a specific event, commonly (albeit not always correctly) assumed to be due to ground water inputs. The figure at right shows a hyetograph for an isolated rainfall event of about 24 hours duration (Sleepers River, Danville VT) and the corresponding hydrograph. In this case, Q b could be estimated as the low flow before the rainfall event. However, in practice, separating Q b from Q ef can be very difficult. For example, a stream might be still responding to a past precipitation event or series of events at the time when a new precipitation even occurs. Graphical separation methods and chemical tracers provide ways of separating base flow from event flow. Dingman 2002, Figure 9-4.

There are several ways to separate Event Flow from Base Flow Event Flow

New Water (Event water) vs Old Water (Pre-event water) Environmental isotopes: 16 O, 18 O, 1 H, 2 H (Deuterium), 3 H (Tritium) Geochemical tracers: 2+ Ca, 2+ Mg, + Na, - Cl, - HCO 3, 2- SO 4

Event flow For most regions, the ratio of event flow Q ef to total rainfall W is considerably less that 0.5 and often less that 0.1. The figure at right shows the ratio expressed as a percentage for the southeastern United States. Note the strong spatial variability. These results indicate that much of the streamflow travels to streams via delayed routes as base flow, in large part due to regional groundwater flows Dingman 2002, Figure 9-6

Event flow (cont) The ratio Q ef /W also varies strongly from event to event for a given watershed. The figure at right shows hytographs and corresponding hydrographs for two similar summer storms on a small1.6 km 2 watershed at Barrow, AK. For the first event, Q ef /W is 0.63; for the second event, Q ef /W is 0.012. The first event was during a very wet summer while the second was during a very dry summer. Much of the precipitation went into soil moisture recharge in the second case. Dingman 2002, Figure 9-9

Event flow (cont) The analysis and modeling of event hyetographs and response hydrographs involves a dedicated terminology Dingman 2002, Table 9-1 Dingman 2002, Figure 9-10

Hydrograph shape Hydrograph shape and the terms describing hydrograph response in the previous figure are determined by storm size, watershed size, soils and geology, slope and land use. Consider the two situations at right. The lag time (the centroid lag T LC in the previous figure, taken as the time difference between the centroid of effective water input and centroid of the response hydrograph) is smaller over an urban watershed than over the agricultural watershed (we are considering response to the same hytograph in each case). This reflects the low hydraulic conductivity of the urban watershed (water falls on pavement, allowing for rapid surface runoff) compared to the agricultural watershed. In turn, the peak discharge is greater over the urban watershed. http://web.cortland.edu/barclayj/hydrograph.jpg

Overland flow Overland flow is produced by two basic mechanisms 1) Infiltration excess, or Hortonian Overland Flow (panel at left) results from saturation from above where the water input rate w(t) exceeds the saturation hydraulic conductivity K* h of the surface for a duration exceeding the time of ponding t p. 2) Saturation (or Dunne) Overland Flow (panel at right) results from saturation from below ; water is added to the top, but the soil is saturated so that overland flow occurs. Saturation overland flow also includes return flow contributed by the breakout of ground water from upslope. http://www.meted.ucar.edu/ http://www.flickr.com/photos/15157983@n00/211869881

Overland flow (cont). Hortonian overland flow tends to be associated with conditions of high precipitation intensity (e.g., a decent thunderstorm), low soil permeability (clay, asphalt) and sparse vegetation. It appears a thin sheets of water (sheet flow), small threads of water, or rill erosion. The figure at right illustrates idealized relationships between the rate of Hortonian overland flow q ho (t) and the infiltration rate f(t) with a constant water input rate w(t). Note that: q ho (t) = w(t) f(t) Dingman 2002, Figure 9-19

Overland flow (cont.) Compared to Hortonian overland flow, saturation excess overflow tends to occur with lower precipitation intensity and longer-duration precipitation events. It is favored with more permeable soils (along with lower precipitation intensity, this means that ponding is less likely) as well as in moist soils where the water table is near the surface. At the onset of precipitation, overland flow is absent and only regional ground water flow is occurring (top panel of figure). With continued precipitation, the water table rises and the surface becomes saturated. Once the surface is saturated, direct precipitation can no longer infiltrate and becomes runoff (bottom panel of figure). There can be a contribution to the saturation overland flow by a return flow contributed by breakout of groundwater from upslope regions. Return flow is usually, but not always, a fairly minor contributor to stream flow event response. In humid regions, saturated overland flow is the major mechanism producing event response. Dingman 2002, Figure 9-21

Overland flow (cont). While saturated overland flow tends to be most apparent near stream areas (where the soil is most likely to be saturated) it can also occur where subsurface water flow lines converge in slope concavities (hillslope hollows) (panel a at right), along slope breaks where the hydraulic gradient is reduced (panel b), where soil layers are locally thin (panel c) and where hydraulic conductivity changes (panel d), resulting in perched zones of saturation that reach the surface. Dingman 2002, Figure 9-22

Overland flow (cont.) Within a given watershed, the extent of areas saturated from below varies widely with time, which is in large part responsible for the large variability of storm runoff observed in many regions. This is important for understanding and modeling event response. The example at right is for a drainage in central Vermont with gentle slopes and moderate to poorly drained soils. A s is the area that is saturated. Dingman 2002, Figure 9-23

Subsurface storm flow Regional groundwater flow is usually the source of most streamflow (base flow) between event responses. While residence times of groundwater are generally too great to contribute to event flow, conditions may occur in which such subsurface flow enters a stream quickly enough to make a contribution. This in termed subsurface storm flow. The figure at right shows a situation with subsurface storm flow above impermeable bedrock. http://www.meted.ucar.edu/

Subsurface storm flow (cont.) As the water table adjacent to streams in humid regions is near the surface, percolation recharges groundwater near stream areas before upslope locations, producing a mound in the water table, steepening the hydraulic gradient both towards and away from the stream. The steepened streamward gradient can give a sustained contribution to streamflow. In some situations, pressurization of the capillary fringe from percolation can rapidly produce a groundwater mound. The figure below shows the simulated response of the near-stream water table to pressurization of the capillary fringe from a rain event on sandy soil. The lines show the position of the water table at successive times after the onset of rain. Dingman 2002, Figure 9-25

Subsurface storm flow (cont.) In many regions, hillslopes have a thin layer of permeable soil, overlying relatively impermeable layers. Infiltration and percolation commonly produce a thin saturated zone unconnected to the regional groundwater flow, and downslope flow in this zone can contribute to event response. This is referred to as a sloping slab; this is the basic situation depicted in the previous slide in which the permeable layer is underlain by bedrock. As seen in the figure at left, the depth of the saturated zone increases downslope, in some cases (see bottom panel) the sloping slab is attended by a breakout near the stream, producing saturation overflow along with subsurface storm flow. Dingman 2002, Figure 9-26

Role of drainage density fao.org (Food and Agricultural Organization of the United Nations) Drainage density is the total length of all streams in a watershed divided by the total area of the drainage basin. It depends upon climate and the physical characteristics of the drainage basin. Impermeable ground or exposed bedrock will lead to an increase in surface water runoff and higher density. Areas of steep topography also tend to have a higher drainage density than areas with gentle topography. Watersheds with a high drainage density have a shorter response time to a precipitation event and a sharper peak discharge.

Modeling runoff with TOPMODEL http://iflorinsky.narod.ru/ti.htm TI derived from a DEM of the Kursk region, Russia. TOPMODEL (Box 9-3) is a framework for modeling runoff in humid areas by identifying the time-varying portions of a watershed that can produce saturation overland flow. The watershed is taken to be covered by a uniform thin layer through which downslope saturated flow occurs below a water table parallel to the soil surface (the sloping slab). At each point in the modeled watershed, the production of saturation overflow is proportional to the tendency to collect subsurface flow from upslope areas and inversely proportional to the tendency to transmit that flow downstream. These opposing tendencies are expressed in a topographic index (TI): TI = ln(a/tanβ) Where tan β is the local slope and a is the areas draining to a given point per unit contour length; these terms can be computed from a DEM

TOPMODEL (cont.) The saturated hydraulic conductivity decreases exponentially with depth to near zero at the base on the soil. Subsurface flow q i at each location (i) in the watershed is treated using a modified version of Darcy s law: q i = T o.s i. exp(-d i /M) Where T o is the transmissivity of the soil when saturated to the surface and M characterizes the rate at which the conductivity decreases with depth. The local soil water storage deficit d i (the value of the difference between the current soil water content and the saturated content) is linked to the watershed mean storage deficit <d> as d i = <d> + M.(<TI> - TI i ) Where the angle brackets denote the watershed mean The value of <d> is calculated at each time step by keeping track of the watershed water balance (precipitation, evaporation and outflow). At each time step, points capable of generating overland flow are those for which d i equals zero.

TOPMODEL (cont.) d i = <d> + M.(<TI> - TI i ) Points with a large drainage and small slope have high TI values these tend to correspond to near-stream areas and swales where water flow will tend to be convergent hence likely to produce saturation overland flow TI for the Kursk region, Russia Points with a low TI (e.g., steep terrain) will tend to have divergent flow and not produce saturation overland flow TOPMODEL was developed in 1979 and has seem many developments and versions. http://iflorinsky.narod.ru/ti.htm