|
Classification of River Channel and Riparian Features
P. McDowell and M. Hughes
Project: "Floods and Human Response:
Implications for Geomorphic Adjustment and Fish Habitat in the Interior
Pacific Northwest" |
|
Go to: General
issues Classification
system figures Classification
system definitions |
|
We have adopted the following file naming protocol: Scanned aerial photos that have not been georectified (base photos): project code (2 characters), flight line (3 characters), photo number (3 characters). Example: The scanned image of photo NZ-9ee-85 (from the 1964 ASCS project) is named nz9ee085.tif. Rectified aerial photos: Add “re” to front of file name. Example: Rectified file of nz9ee085.tif is named renz933085.tif. The minimum size feature to be digitized is greater than or equal to 25 square meters (25 pixels) total area, and greater than or equal to 2 meters (2 pixels) wide on average. Small bars or other features that are smaller than this will be ignored. We will digitize at approximately 1:2000. The digitizer will shift between scales to recognize and make interpretations, typically shifting between larger scales (1:5000) and smaller scales (1:1000). It also helps in interpretation to have the unrectified image available in another window while you are digitizing. In digitizing, we are creating shapefiles, not coverages, at present. We haven’t decided which approach is best: 1) initially create shapefiles and keep all work and analysis in shapefiles; 2) initially create shapefiles and later covert them to coverages and fix topology before doing analysis; or 3) initially create coverages. The feature definitions below are a preliminary classification. They are intended to be the conceptual features we will use for analysis. In digitizing, the features initially digitized may not fit these definitions. Some of features are created in a second step by overlaying and clipping. We need to evaluate the feature definitions and determine whether they are the appropriate units for analysis. We suggest the following sequence for digitizing. (These are the ones we’ve done so far, in our image library examples.) 1) Digitize the outermost active channel boundary, creating the acb shapefile. 2) Digitize islands within the active channel, creating the isl shapefile. 3) Digitize bar complexes and individual bars, creating the bar shapefile. 4) Digitize chutes within bar complexes, creating the chu shapefile. 5) In a later phase, different land covers on floodplains will be digitized. 6) We have not yet separated main channel and side channel. Also, although we have digitized bars, what is left within the active channel includes wetted active channel and other dry active channel (i.e. active channel bed). You will see this in image 1971_RM63_map.pdf . During the analysis phase, these files will be used to create shapefiles (or coverages?) consistent with our definitions -- a total active channel shapefile, a shapefile of all bar units, etc. -- for calculating areas and other types of analysis. For an example of how this will work, see Figure 5. Digitizing sequence.
Figure 1. Top-level features. (Go to Fig. 1 as a pdf)
Figure 2. Floodplain features. (Go to Fig. 2 as a pdf)
Figure 3. Active channel features (Go to Fig. 3 as a pdf)
Figure 4. Schematic map showing relationships between the major features.
Figure 5. Digitizing sequence (Go to Fig. 5 as a pdf) Definitions for Land and Channel Features Go
to: Top-level units
Types of active channel
Subdivisions of
active channel Bar
types These are definitions, not a key in the strict sense. See Figs. 1-3 (tree diagrams) for levels and relationships. These definitions emphasize characteristics visible on aerial photos. If field checking is possible, many other useful criteria may be employed. Top-level units: active channel, floodplain, terrace, upland Active channel (AC): Area frequently or recently reworked by fluvial processes during high flow events. Roughly equivalent to scour zone of Poole et al (2002). The active channel includes only channel-like features (i.e. elongated, appear to have been carved by flowing water); other water bodies are classified within the floodplain. On aerial photos (usually taken during low flow season), the area in the active channel is occupied by water, bars, exposed channel bed, or colonizing (pioneer) vegetation, while the floodplain has established vegetation, fields, structures/roads, and also some water. The active channel boundary (ACB) is the boundary between active channel and floodplain or higher surface. The ACB is defined by the edge of established (or permanent) vegetation, which may be trees, shrubs or herbaceous. The key point in identifying the ACB is distinguishing between more or less continuous vegetation cover that has been established for a number of years, and bare surface or colonizing vegetation. For more details, see Figure 6 (ACB decision tree). Go to: Figure 6 in html Figure 6 in pdf Floodplain (FP): relatively flat land surface outside the ACB, by fluvial deposition. The floodplain is the lowest extensive land surface; higher fluvial surfaces (terraces) are excluded. The floodplain may include islands that are separated from the continuous floodplain by active channel (see discussion below under Bars). Land cover of the floodplain is permanent, established vegetation (not showing signs of recent scour), agriculture, human-built structures, or isolated water features. Terrace (T): a relatively flat surface that appears to have been formed by fluvial deposition but is higher than the floodplain. Older, higher terraces may have a somewhat dissected surface rather than a flat surface. There should be a fairly distinct boundary (scarp) between the floodplain and terrace, but the boundary between terrace and upland may be distinct or gradual (if covered by colluvium). Usually land cover on the terrace is different from the floodplain. Many rivers have more than one terrace level; we’ll lump them all together for now. Upland (U): a surface higher than the floodplain that does not appear to be underlain by alluvium. May be smooth or rugged. Generally not flat but may be gently rolling. Bedrock or non-alluvial composition is indicated by ruggedness, slope, relief, and/or rock outcrops. Types
of active channels Ephemeral channel (EC): a channel that is entirely dry or does not have a continuous thread of water within it on the photo date; may have disconnected pools of water. Perennial channel (PC): a channel that has a continuous thread of water within it on the photo date. The area within a perennial channel includes dry (but scoured) area as well as water surface; the distinction between these is made when water is digitized (see below). Main channel (MC): the widest, continuous channel of the main river. Side channel (SC): a perennial channel connected to the main channel at both ends, but narrower than the main channel. Tributary channel (TC): a perennial channel connected to the main channel at only one (downstream) end, and originating outside of the floodplain. Tributary channel vs. small re-entrants: if length is > 2 x width, it is a tributary channel; if < 2 x width, it is simply a bulge in the active channel. Springbrook channel (SB): a perennial channel originating within the floodplain and connected to the main channel or a side channel at only its downstream end. Recharge channel (RC): a perennial channel originating within the floodplain and connected to the main channel or a side channel at only its upstream end. Isolated scour features: patches surrounded by floodplain that are not connected to the main channel or other channels but appear to be shaped and/or formed by recent scour. Usually they will have a linear shape and evidence of bright areas representing recent scour. They may or may not contain water. (We’re not sure we’ll see these, but they may occur, so they are included here.) Subdivisions within active channel area Water: water surface (at time of photo, generally low flow) within active channel. Water surface may be subdivided later; see water surface types below. Note: While doing the image library examples 1964_RM48 and 1971_RM63, we identified chutes as a subtype of type water surface. Chutes occur within and divide bar complexes. Bar: positive relief bedform within active channel and above water level; see bar types below. Other active channel: any other area within active channel not identifiable as water or bar; may include exposed channel bed that does not have positive relief. (Will we be able to recognize this?) Note: While doing the image library example RM48--1964 photos, we identified an area of scoured floodplain and a dry chute that we put into this feature type. (Reviewers, please look at these features on 1964_RM48_map.pdf. Do you agree with our interpretation? Should we add these to the definition of other active channel?) Bars can be digitized simply as “bar”. The bar type will be defined by entries in the following attribute fields: bar position, bar subtypes, simple/complex, typical/different, and comments. This will allow flexibility in analysis. We recommend that all attributes be filled out at the time digitizing is done, because it will probably be much less efficient, and lead to changes in digitized lines, to do it separately or by different people. Note: While doing the image library examples 1964_RM48 and 1971_RM63, we did not add attributes. Distinguishing between
bars
and islands: Bar position field: attached, mid-channel, or cross-channel Attached bar: not surrounded by water; contiguous (at least over a short area) with the active channel boundary Mid-channel bar: surrounded by water or other active channel; not touching the active channel boundary; typical of braided reaches; includes longitudinal, transverse, linguoidal, diagonal bars, some convergence/confluence bars, and others. Cross-channel bar: bar that is attached to both left and right edges of active channel; many are riffles, but there may be other types (push-up dams?). Additional attribute fields for bars: Subtypes based on
morphology or location: Simple or complex: A simple bar is a single bar occurring by itself, not part of a bar complex. A bar complex is a group of adjacent bars that appear to have a common origin. Bar complexes are typically formed as a single large bedform during maximum sediment transport in a large peak flow event, and they are dissected into separate units later in the same event or during subsequent events. Typical or different: Flag it as different if there is something you need to say in the comments. Genetic and other comments: Usually used to identify bars that were problematic in classification, or to explain a bar complex. Examples: looks like dissected mid-channel bar, others?? Rationale for our bar classification system We started by recognizing that the major types of bars recognized in the literature include the following: point bar, lateral bar, confluence or convergence bar, mid-channel bars of various types (longitudinal, transverse, linguoidal, crescentic, diamond, convergence), diagonal cross-channel riffles or bars, and perpendicular cross-channel riffles or bars. We then tried to reduce this to a smaller number by lumping together types that were morphologically similar and probably were also similar in terms of hyporheic flow. This resulted in the list of five primary bars that we use in the bar subtype classification above: attached point bars, attached lateral bars (includes attached tributary confluence bars), mid-channel simple bars (includes longitudinal, transverse, linguoidal, crescentic, diamond, and convergence), mid-channel bar complexes, and cross-channel bars. We also recognize that the literature suggests that bars undergo evolution. Many examples of bar complexes result from 1) formation of a large simple bar during high flow and active bedload transport, then 2) dissection during the falling limb of the hydrograph or during subsequent smaller flow events that don’t fully remobilize bedload, resulting in 3) a bar dissected by minor channels and chutes that may carry low flow or be dry during low flow. The minor channels and chutes may be wet or dry depending on river stage on the day of the photos. Dissection is probably the origin of the features Leal has called criss-cross bars. Are there different morphological types of complex bars, or should we just lump them all in one subtype? We opted to go with one subtype for now. Dissection can also happen to attached bars. There will be some point bars that are partially penetrated, or even separated from the bank by a small chute. The chute may be wet or dry depending on river stage on the day of the photos. We suggest keeping these in the attached bar group unless the chute is wide. We also recognize that our bar identification will limited by the 2-D view of aerial photos. Many geomorphologists would define bar morphology irrespective of water level (i.e. visualize a dry channel) – the bars would be the convex forms, including that portion that is submerged at low flow. In this approach, the entire active channel would be divided into bars, or into bars and pools (or interbar areas). Since we’re working from aerial photos, we will identify and digitize only the bar area above the water on the day of the photo. This raises the problem that bar area is dependent on river stage, but we don’t see any way around it. Therefore, we’ll probably have to calculate bar area as percentage rather than area, or normalize bar area by stage in some way. Any ideas? Note: These are more or less analogous to pool vs. riffle vs. cascade. We will not do these in the first round of digitizing (before April); water can be subdivided into these types later. Smooth water surface: even, smooth, usually dark water surface Rough water surface: rough water with a certain proportion of lighter patches?? Very rough/whitewater: rough water with high proportion of white spots or all white? Subdivisions within floodplain area First level is water-related vs. non-water related. Water-related features: includes ponds, perennial wetland vegetation, irrigation channels, ?? Ponds: small isolated bodies of water on the floodplain; distinguished from active channel features in that they do not appear to have formed by flowing water and do not show signs of scour. Should this include both natural ponds and human-made ponds? Do we want to distinguish between perennial and intermittent ponds? Perennial wetland vegetation: This category may include polygons that are channel-like in shape (i.e., old filled channels) but show no sign of recent scour. (Needs input from Scott O'Daniel) Irrigation channels: channels used for delivery (or drainage?) of irrigation water; potentially includes local irrigation ditches and larger irrigation conveyance canals. Non-water related features: includes agricultural land, roads, buildings, etc. Not to be digitized separately at this time. We tested our classification on two short reaches. The results are presented in the following figures. (Note: These were done with scanned and georectified images that are not as good as we wished. We are re-doing scans and georectifying in a different way that should produce better results.) River mile 63 (near Mission), done on a 1971 photo (after the 1965 flood): 1971_RM63_map.pdf -- with mapped features 1971_RM63_photo.pdf -- photo only 1971_RM63_ACB.pdf -- paired photos with ACB on and off River mile 48 (downstream of Pendleton), done on a 1964 photo (before the 1965 flood): 1964_RM48_map.pdf -- with mapped features 1964_RM48_photo.pdf -- photo only 1964_RM48_ACB.pdf -- paired photos with ACB on and off 1964_RM48_fluvialmarkers.pdf -- with relevant fluvial markers highlighted (In the RM48 photos, note the dry active channel area that is not bar.) Go to: General
issues Classification
system figures Classification
system definitions
|
