Late Holocene Floodplain Processes and Post-European Channel
University of Wollongong Thesis Collections University of Wollongong Thesis Collection
University of Wollongong Year
Late holocene floodplain processes and post-European channel dynamics in a partly confined valley of New South Wales Australia Timothy J. Cohen University of Wollongong
Cohen, Timothy J., Late holocene floodplain processes and post-European channel dy- namics in a partly confined valley of New South Wales Australia, Doctor of Philosophy thesis, School of Earth and Environmental Sciences, University of Wollongong, 2003. http://ro.uow.edu.au/theses/1931
This paper is posted at Research Online.
5. Post-European channel dynamics in the Bellinger valley
CHAPTER 5 - POST-EUROPEAN CHANNEL DYNAMICS IN THE BELLINGER VALLEY 5.1 INTRODUCTION
The Bellinger River, like many other systems in south-eastern Australia (see review by Rutherford, 2000), has undergone a series of planform and dimensional changes since European settlement. This chapter investigates the nature and timing of these changes and relates them to both the pre-European state of the channel and floodplain, along witii the rainfall and discharge patterns established in Chapter 3.
The chapter consists of four main sections. Section 5.2 briefly outiines the methodological techniques used in this chapter, while Section 5.3 provides a historical assessment of channel and floodplain character in the Bellinger catchment. Sections 5.4 - 5.5 quantify the magnitude of change and present a summary of the evidence for cyclic channel changes.
5.2 METHODOLOGICAL APPROACH
5.2.1 Historical and anecdotal evidence for the BeUinger valley
In order to determine the character of the Bellinger River and the surrounding floodplain environment at the time of settlement a number of historical data sources have been utilised. Eariy surveyor's notes between 1841 and 1842 provide the first written description of the Bellinger valley. Parish maps, along with portion plans for individual properties within each of the three reaches were used to construct the position of the channel at the time of settlement. Interviews with some of the older residents provide additional anecdotal evidence for the early to mid 20* century, but also includes oral descriptions from their parents. Historical photos supplement both the historical and anecdotal data, while aggregate extraction records supply details of the mid-late 20* century gravel extraction operations that occurred on the Bellinger River.
5.2.2 Palaeochannel analysis
Planform data from parish maps and portion plans revealed evidence for the timing of planform changes within the study reaches. The two lower valley reaches. Wills and Gordonville Reach contain palaeochannels, while additional palaeochannel data has been obtained from an abandoned cutoff upstream of Orama Reach (Darkwood Palaeochannel)
206 5. Post-European channel dynamics in the Bellinger valley and downstream of Gordonville Reach, at Bellingen township (Bellingen Paleochannel). Unlike palaeochannel analysis in previous studies (Erskine et al., 1992; Cohen and Brieriey, 2000; Brooks et al., 2002), only two of the five palaeochannels are completely abandoned, with the other three remaining as secondary channels. Cross-sections were undertaken at cutoff apices and inflection points, with the estimation of prior channel depths obtained from auger holes and historical evidence. The five palaeochannels provide a crude estimate of the dimensions of the channel prior to the major post-European dimensional changes documented later in the chapter. The limitations of such a data set are outiined in Section 5.4.1.
5.2.3 Spatial analysis of channel characteristics
Aerial photographs from a 1943, 1956, 1960, 1973, 1982, 1994 and 2000 time series allowed assessments of the magnitude and timing of planform and dimensional changes in the lati:er half of the 20* century (see Appendix 5.1 for aerial photograph details). Ground control points were collected using real time kinetic (RTK) differential GPS tied to die Australian Height Datum (AHD), with an X,Y accuracy of ± 2 m and a Z accuracy of ± 0.1 m. In order to test the notion that channel geometry adjusts in a cyclical manner associated with the inferred cyclical changes in discharge regime (FDRs and DDRs), four time intervals were chosen for channel geometry analysis. These include the 1943, 1956, 1973 and 2000 aerial photographs, with the 1943 representing the last DDR (1900 - 1947), the 1956 and 1973 representing the early and later stages of the previous FDR (1948 - 1977), with the 2000 aerial photograph/field data representing the current DDR (1978 - present). All historical data sets have been collected using the same map grid co ordinates (based on the differentially derived GPS ground control points) as the current field data, allowing for the most accurate comparison of past channel geometries, with those of today. In order to maintain consistency within the four time intervals (1943, 1956, 1973 and 2000) the ground control points were used to rectify the 2000 aerial photographs, while all previous time series were rectified using the nearest neighbour polynomial image-image warping method within ENVI 3.4. Both current and historical cross-sections occur approximately one channel width apart (125 - 150 m) and include at least one cross-section in a representative pool or riffle location.
Using the (RTK) ground control points photogrammetrically derived cross-sections were obtained from the 1943 and 1973 data series, with each of these data sets providing accurate cross-sectional geometry, but of necessity limited to the channel above the Mow- flow' water level. For direct comparison current field-surveyed cross-sectional geometry
207 5. Post-European channel dynamics in the Bellinger valley has therefore been adjusted to exclude the small areas (i.e. < 5 %) below low-flow. The 1956 rectified aerial photographs provide an additional width data set alone. Each of the photogrammetrically derived data sets has an error associated with observational error, determined by the scale, quality and resolution of the aerial photographs. The 2000 field data provide the most accurate representation of channel geometry, while the 1943 photogrammetrically derived data (from 1:18,500 photography) provide the second most accurate information. The 1973 1:41,000 photography is of poorer resolution thereby resulting in both observational error and errors associated with the setting up of the photogrammetry model. The 1973 photography, however, has more ground control points per photogrammetry model than the 1943 photography due to the fact that it covers a greater length of the valley. Thus, the discrepancy between differences in scale is offset by an increased level of ground control (P.Tuckerman, pers.comm., 2002). Residuals on ground control for the 1943 and 1974 aerial photography range from 0.2 - 0.6 m. However, given the level of accuracy of the ground control points all photogrammetry and field survey measurements are presented to the nearest 0.1 m. The 1956 photography at 1:40,000 is also of poor resolution and is subject to error within the rectification process. In all rectified images root mean squared errors (RMS) were kept below 3 m. This data are therefore subject to a ± 3 m spatial error, in addition to pixel resolution error of ± 1 m, coupled with observational ertor associated with defining on-screen bank tops.
All data sets are small (n = < 30) and therefore normal distributions are not assumed and were tested using Shapiro-Wilk's test for normality. Differences in variances have been assessed using Levene's test for equality of variance, while differences in means have been assessed using paired t-tests with a pooled variance estimate. Where populations displayed a divergence from a normal distribution, non-parametric analysis utilising the Wilcoxon signed-rank sum test has been used.
Riparian and hillslope vegetation changes since 1943 have also been quantified using the rectified data sets, expressing riparian vegetation changes along each of the reaches as an area/unit length index. All vegetation within a 150 m buffer from the mid-point of the channel was digitised for the four selected time intervals. The chosen buffer width represents the maximum of the mean channel widths for the three reaches and provides a consistent approach to assessing temporal changes to riparian vegetation within one channel width's distance from the centre line of the channel. This method, however, sometimes includes vegetation beyond the immediate riparian zone such as where the channel impinges against the bedrock valley margin. This vegetation index is therefore quantifying spatial changes to both riparian communities and hillslope communities within the 150 m buffer.
208 5. Post-European channel dynamics in the Bellinger valley
5.2.4 Reach analysis, channel processes and an assessment of prior bed levels
Cross-sectional and longitudinal profile data from individual channel units was obtained witiiin each reach as part of a broader reach analysis (Chapter 2). While pool and riffle locations were systematically surveyed, additional channel cross-sections were obtained in areas of bedrock constriction and expansion. Riffles and pools were identified based on morphological criteria but compared to the polynomial regression method of Richards (1976). Each cross-sectional data set includes some sections benchmarked in 1992 and resurveyed in 2000. A number of sections were re-surveyed following the 2001 flood which modified channel geometry in selected locations (described and modelled in Chapter 6). Historical, anecdotal and field evidence was used to provide a qualitative assessment of prior bed levels along the lower Bellinger River (Wills and Gordonville Reach). This evidence, where available, has been tied into the longitudinal profile. Evidence for channel incision in the upper reach (Orama Reach) is poor and remains unknown.
Estimation of bankfull is problematic in a landscape with numerous inset alluvial units, a high and a low floodplain, along with varying degrees of post-European incision. The stage of the 1.58-year flood on the annual series has been used in previous studies to identify bankfull (Dury et al., 1963) and linked to the most effective or 'dominant' discharge with regards to formative processes (Wolman and Leopold, 1957; Wolman and Miller, 1960). Additional indices for identifying bankfull include the field identification of bankfull (Gordon et al., 1992) and vegetative indices (Sigafoos, 1964). Riparian vegetation in south-eastern Australia, however is not characterised by annual species that regrow following each bankfull event. The field identification of 'bankfull' based on vegetative characteristics is prevented due to the discharge variability (Chapter 4), large channel dimensions that contain floods beyond the 50 years recurrence interval (Doyle, 2003) and the perennial and fragmented nature of the riparian vegetation.
Geometrical indices such as minimum width-depth ratios utilising either maximum or mean depths (Wolman, 1955; Williams, 1978a) and bench indices (Riley, 1972) have also been utilised to identify bankfull conditions. However, Pickup (1976a) recognised that identifying bankfull conditions using the minimum width-depth ratio may not reflect differences in channel shape, with significantiy differently shaped channels sharing the same width-depth ratio. In addition to the limitations outiined by Pickup (1976a), the presence of low-flow notches, inset within a bench or bar along throughout the study area will often produce the minimum width-depth ratio. Thus, using the minimum width/depth ratio in this situation clearly under estimates a meaningful approximation of bankfull. The
209 5. Post-European channel dynamics in the Bellinger valley small discharges associated with these features do not have a formative role in determining bankfull geometry or overall channel morphology. For consistency, channel top width has been utilised within this thesis to make comparable observations of bankfull conditions between the four time intervals. Channel top width was taken as the top of the levee on the high floodplain. Where the high floodplain was at a lower level on one side of the channel cross-section, this height was used.
5.3 THE BELLINGER RIVER AT THE TIME OF SETTLEMENT: AN HISTORICAL ANALYSIS
Unlike larger valleys to the north and south which were settled eariier in the 19* century (Daley, 1981, Birrel, 1987) the Bellinger valley was first 'discovered' by William Myles, a cedar getter moving north from the Macleay valley in either 1840 or 1841 (Braithwaite and Beard, 1978). With the notion that there was an unsettled valley with good water supply, Clement Hodgkinson led the first expedition from the Macleay valley in 1841. In this expedition, Hodgkinson, who was accompanied by 'tame' blacks who facilitated the joumey through adjoining Aboriginal lands, made a number of comments with regard to the nature of the river and surrounding floodplain forest (Hodginson, 1844).
Like many of the early descriptions of the floodplain forests of the mid-north coast region of NSW, Hodgkinson refers to the 'brush' — a euphemism used to describe dense rainforest — that they often had to cut through. Descriptions of the country-side using terms such as 'choked' or 'impervious brush' provide an indication of the density of some of the vegetation they were travelling through. Weingarth (1925) estimates that Hodgkinson and his party approached the Bellinger valley from the south, entering the main valley in the area of Orama Reach, or potentially just downstream (i.e. mid-upper valley setting). Descriptions of the riparian zone and surrounding floodplain forest vary considerably from 'beautiful grassy forest bank' to 'choked' and ' impervious brush'. The following quotes provide a good description of some of the vegetation communities, along with a brief description of the dimensions of the channel.
" As soon as I awoke I ran down to the river, which I found to be quite a rapid stream, upwards of one hundred feet wide at this point, and flowing over a bed of large shingles we got down into the bed of the river and forded it, entering a dense cedar brush on the other side." (Hodgkinson, 1844, p.40)
210 5. Post-European channel dynamics in the Bellinger valley
This estimate of channel dimensions provides the first written description of the scale of the channel, however it is uncertain as to whether Hodgkinson is referring to the bed of the channel or the 'bankfull' dimensions. Seeing as though the channel bed is currentiy no wider than approximately 20 m (covered in more detail in Section 6.4) it would seem tiiat Hodgkinson was referring to the bankfull width of the channel. He goes on to describe the nature of the Orama region.
" In a straight line often miles, we crossed and re-crossed the river no less than twelve times; this was unavoidable on account of the steep, inaccessible forest banks, which formed tangents to the convex bends of the river on either side. Our course, therefore, lay from necessity along the alluvial land, which consisted of brush, cedar plains, and forest fats. The brush contained the finest cedar and rosewood I had ever seen; the trunks of these trees were often six feet in diameter, and ninety feet high, before they drew out a single branch. The Casuarina also grew to such an uncommon height, and its foliage assumed such an unusual form, that I thought at one time it was a species of Pine. The small clear plains, just mentioned, were covered with coarse broad-bladed grass, growing as high as a man's middle, and having the appearance of small wheat fields; the grassy forest flats were principally wooded by that species of Eucalyptus called Forest Mahogany." (Hodgkinson, 1844, p.40)
Descriptions within the above quote provide a good insight into the nature of the vegetation pattems in Orama Reach. The 'steep, inaccessible forest banks' most certainly refer to the concave bedrock valley margins, which resulted in the party traversing the discontinuous floodplain and terrace assemblage assessed in Chapter 3. It is interesting to note that Hodgkinson identifies at least three alluvial vegetation associations, each with its own diagnostic species and more than likely growing on specific floodplain units.
The associations described by Hodgkinson appear to relate to vegetation communities adapted to a certain flood frequency and associated substrate condition. The higher late Pleistocene terraces inundated less regulariy would more than likely sustain 'drier' communities, potentially the cedar plains and forest flats and potentially conditioned by fire. The coarse bladed-grass possibly refers to either Imperata cylindrica or a Lomandra sp. while the forest mahogany referred to may indicate the presence of Eucalyptus microcorys. In contrast, the more regularly inundated surfaces (the low and high floodplains) would be characterised by the 'brush'. This brush would most likely consist of both riparian-gallery rainforest and subtropical rainforest with emergents such as the red cedar (Toona ciliata), stinging tree (Dendroconide excelsa), river oak
211 5. Post-European channel dynamics in the Bellinger valley
(Casuarina cunninghamiana), white booyong (Heritiera trifoliolata) and the large emergent Moreton Bay Fig (Ficus macrophylla); associated witii suballiance 1 and 3 (Floyd, 1990). The riparian-gallery forest would more than likely have also included the now common (but sparsely distributed) riparian species, weeping myrtle (Waterhousia floribunda) and water gum (Tristaniopsis laurina). Flood disturbance would have ultimately determined the distribution of species that Hodgkinson describes, with areas subject to regular disturbance and coarse substrate having species such as the river oak. The bedrock valley margins would be characterised by dry rainforest with species such as Brush-box (Lophostemon confertus).
Hodgkinson and his party as he travels upstream also identifies the presence of in-channel species in the following quote:
" / continued my course up the river, over the same kind of ground as the day before, until I attained a point where the brush seemed to be almost entirely superseded by narrow, lightly wooded flats; with patches of the Swamp Oak (Casuarina paludosa) growing among the shingles of the stream." (Hodgkinson, 1844, p.41)
It is more than likely Hodgkinson is referring to the area immediately upstream of Orama Reach where valley widths decrease before entering Dardenelles gorge. The presence of in-channel oaks more than likely refers to Casuarina cunninghamiana, as Swamp Oak (Casuarina glauca) is normally found in more saline environments, while Allocasuarina paludosa is restricted to the southern regions of NSW (Fairley and Moore, 1989).
Hodgkinson's first trip in the mid-upper valley was followed in 1842 by a second trip with the aim of discerning whether the lower Bellinger was navigable. Similar descriptions of brush containing cedar, rosewood, fig trees, nettle trees, and plum wood, along witii descriptions of more dense rainforest on the coastal floodplain suggests that this second joumey made its way to the tidal limit. Hodgkinson (1844) describes the non-tidally influenced forests or 'brush' to be finer than that found on the coastal plain. Other than brief descriptions of the dimensions of the tidal channel ('280 feet wide'), and the presence of in-channel vegetation in a non-tidal tributary, Hodgkinson makes little mention of the character of the lower Bellinger River.
Shortly after Hodgkinson's second expedition a successful attempt to cross the bar at the river mouth was made by William Wright in 1842. This time interval saw the arrival of the first domesticated animals in the valley, along with the primary industry of cedar getting
212 5. Post-European channel dynamics in the Bellinger valley
(Braithwaite and Beard, 1978). By 1843 there were 20 pairs of pit sawyers on tiie Bellinger River witii the areas around the coast targeted first. The initial white settiement was between Raleigh and tiie coast, with tiie township of Bellingen originally known as Boat Harbour. Like many coastal catchments of NSW, the search for red cedar (Toona ciliata) induced a rapid expansion into previously unsettied areas. Between 1843 and 1849 cedar getting and 'squatter settlement' was occurring throughout tiie lower valley, witii logs rough dressed, chopped and floated individually down the river (Braitiiwaite and Beard, 1978). While logs may have been floated down the river in the lower reaches, L.Rose (pers.comm., 2002) has suggested tiiat in the mid-upper valley settings, notches were excavated in the alluvial banks where the branded logs were launched into the river where they would await the next flood to be carried downstream (sensu Daley, 1981).
The lower valley experienced more formal settiement witii the first ship built on the lower reaches by 1849, while cedar getting moved further upstream by the end of the 1850s. Once cedar supplies became exhausted other rainforest species were progressively logged, with hoop pine (Araucaria cunninghamii), found in the plateau regions, being the most sought after (Lonie, 2000). The lower Bellinger River, downstream of Boat Harbour (now Bellingen), was used to transport timber with punts, shunting logs and goods from Boat Harbour to the coast, where they were shipped to Sydney. The development of sawmills both in Bellingen in the tablelands region from 1890 onwards resulted in forestry becoming a major industry in the valley.
In 1861 the NSW government passed legislation (The Crown Lands Act) which allowed settlers to take up to 320 acres of crown land, conditional upon the owner living on the land for three years and demonstrating that the land was cleared (Stevenson, 2001). This legislation allowed for the controlling of the 'squatters' while also determining the character of agricultural development in the mid-late 19* century. This time interval within the Bellinger valley saw the development of schools, police stations and post offices, with the name of Boat Harbour changing to Bellingen in 1889 (Braithwaite and Beard, 1978). The mid-late 19* century also saw a succession of floods between 1864 and 1894. Braithwaite and Beard (1978) identified that the lower river was 'shoaling' up in the latter part of the 19* century, resulting in the wharf being moved upstream. Soros-Longworth & McKenzie (1980) suggest that the shoaling of the river mouth at the start of the 20* century may have been caused by channel erosion in the preceding time interval. The exact nature of sedimentation in the lower reaches remains ambiguous with uncertainty as to where the estuary was shoaling and whether it was marine or fluvially derived sediment. Indeed, the relocation of the wharf upstream may indicate that the tidal reaches of the
213 5. Post-European channel dynamics in the Bellinger valley
Bellinger River had undergone increased sedimentation, in contrast to the non-tidal reaches around Bellingen. Section 5.3.3 identifies a series of planform changes along the lower Bellinger River that may have occurred in the mid-late 19* century.
By the time J.H. Maiden journeyed through the valley in 1894 he noted that many of the alluvial flats had been cultivated for maize and that cedar getting was nearing an end (Maiden, 1894). Maiden also made comment on the nature of the channel upstream of Wills Reach before ascending to Dorrigo on the plateau:
"About 10 miles after we leave Bellingen we find the Bellinger here a thin stream, though the great width of cobble stones and the traces of undermined banks very wide apart show clearly that the Bellinger River at this distance from the sea is sometimes a very formidable river."
While Maiden is more than likely talking about the channel in the Thora region, it is difficult to ascertain a good geomorphic interpretation. The first photographic evidence of the riverine environment occurs in the early 20* century and is compiled in Appendix 5.2. These images highlight the shift from sub-tropical rainforest in the 1840s with complete floodplain clearance by 1902. Appendix 5.2 (ii) highlights the nature of the denuded banks at a watering point at Bellingen township indicating some locales had essentially little to no riparian vegetation at the turn of the 19* century. The below-average rainfall in the early part of the 20* century (Chapter 4) resulted in drought conditions that saw a shift from maize to dairy and beef production, placing increased numbers of stock on the valley floor. While it is difficult to ascertain the exact condition of riparian vegetation or channel morphology in the early days of settlement, the photographic evidence suggests that some locations were largely denuded of frees, while others display evidence of sparse and discontinuous bankside vegetation. The onset of dredging to maintain navigation in the lower valley (see Appendix 5.2 iii), the decrease in riparian and floodplain vegetation, along with the historical accounts of downstream aggradation suggest that the Bellinger River underwent a shift in the nature of sediment supply in the first 50 - 60 years of settlement.
Figure 5.1a shows an upstream view towards Wills Reach (lower valley setting), with ringbarked hillslope trees in the mid-ground with what appears as regrowth riparian vegetation on the true left-bank and left-bank bar. A partly vegetated bar exists in tiie centre of the channel which may highlight the location of a riffle (a riffle currently exists at this approximate location). Figure 5.1b-c highlights the presence of young Casuarina
214 5. Post-European channel dynamics in the Bellinger valley
a)
b)
Figure 5.1 a) View upstream towards Wills Reach, 1902, note the regrowth riparian vegetation, vegetated island and ringbarked hillslope vegetation, b) Postcard of the lower Bellinger non-tidal reach looking upstream, 1908, note the cleared floodplains and dominance of Casuarina cunninghamiana. c) Construction of Thora bridge in 1907, highlighting the prevalence of C. cunninghamiana regrowing on the banks (source Bellinger Valley Historical Society).
215 5. Post-European channel dynamics in the Bellinger valley cunninghamiana in two locations up and downstream of Wills Reach. Based on growth rates established from aerial photographic evidence (presented in more detail in Section 5.4) the trees in these photos look < 30 years old. This would make them post-settiement species that have regrown since the initial clearance. Swain (1912), a District Forester for the NSW Department of Forestry, noted that by 1905 53,000 acres of land had been taken up for farming, doubling by 1910 to 113,000 acres. This represents approximately 40% of the total catchment area and of this area Swain (1912) claims that half was cleared and the remainder on the 'verge of denudation'.
What is clear from the historical accounts is the dramatic transformation from sub-tropical and riparian-gallery rainforest and open communities on the higher late Pleistocene tertaces to cedar getting and then quickly through to agricultural practices by the late 19* century. No real assessment of the dimensions of the channel can be made, other than the 'upwards of 100 feet' (~ 33 m) mentioned by Hodgkinson (1844) and by anecdotal evidence by L.Rose (pers.comm., 2002) who suggests the Bellinger River in Wills Reach could be crossed from bank to bank on a fallen tree in the early part of the 20* century. This sort of description is not dissimilar to equivalent evidence presented in the adjacent Nambucca catchment (Doyle, 2003). The shift from diverse vegetation communities, to narrow strips of either remnant or regrowth riparian vegetation, along with the introduction of grass species such as paspalum and kikuyu has resulted in a major shift in the biogeomorphic processes occurring throughout the valley. Many older residents who describe wide-scale post-European channel changes refer to the narrow bands of vegetation that were left from the top of the high bank (probably the high floodplain) to the waters edge. These fragmented vegetation communities were often outflanked in large floods resulting in chute-channel formation behind the narrow bands of vegetation.
Since the mid 20* century land use practices have shifted, with a reduction in rainforest logging, the declaration of National Parks, the decrease in dairy farming associated with major industrial de-regulation, a series of large floods in the 1950s, and a continuance of beef cattle. The reduction of the intensive dairy farming on the valley floor, along with an increase in hobby farming, agro-forestry and increased shared rural residential blocks (multiple occupancy) has resulted in a major shift in land-use patterns for the fluvial environment. Section 5.4 will document the nature of 20* century channel changes, changes to riparian vegetation cover, and the impacts of gravel extraction in the mid-late 20* century, while Chapter 7 further discusses the changes in population, stock numbers and mral holdings.
216 5. Post-European channel dynamics in the Bellinger valley
5.4 CHANNEL GEOMETRY AND PLANFORM CHANGES IN THE UPPER AND LOWER BELLINGER RiVER 1842 - 2000
5.4.1 Channel planform of the mid and lower Bellinger River in the mid 19* century
Utilising eariy parish maps and portion plans of properties that contain river frontage allows a reconstruction of the planform of the lower Bellinger River at the time of settlement. Appendix 5.3 contains the eariiest parish maps of both the lower and upper valley settings (covering the three study reaches). What is apparent from these eariy maps is that the majority of planform changes have occurred in the lower valley settings. Appendix 5.3a shows that the planform of the Bellinger River within Orama Reach to be unchanged due to the dominance of bedrock within this mid-upper valley setting, exhibiting one meander cutoff, upstream of Orama Reach. The nature of channel adjustments within this setting are characterised by increases in channel width on the convex bank, or channel expansion of both alluvial banks at valley inflections (expanded in Section 5.4.3).
Figure 5.2 is a summary diagram illustrating the location of the channel in the lower valley down to the tidal limit at the time of settlement, determined from parish maps, portion plans, historical and anecdotal evidence, and some chrono-stratigraphic data from Chapter 3. Sites of major planform changes are numbered with key sources provided. What is clearly apparent is that all major planform changes, but one, are chute channel cutoffs. Estimated sinuosity of the lower Bellinger River at the time of settlement between the tidal limit at Bellingen and Thora is 1.21, in contrast to the current sinuosity of 1.14. The nine planform adjustments evident in Figure 5.2 represent a reduction in channel length of 1.2 km out of 17.4 km (7%).
5.4.2 The use and limitations of palaeochannels to determine channel characteristics at the time of settlement
Anecdotal evidence suggests that the parts of lower Bellinger River (i.e. the non-tidal reaches) at the time of settiement were narrow and deep and could be crossed by a fallen ti-ee (L.Rose, pers.comm., 2002). Major changes in channel geometry and planform have all but eroded evidence of prior channel conditions. The four palaeochannels used in die lower valley setting (Wills, Gordonville, CSR and Bellingen palaeochannel) along witii Darkwood palaeochannel, upstream of Orama Reach, all suggest narrow channel
217 5. Post-European channel dynamics in the Bellinger valley
•'m:^ixyyxff D. O 3 O o o -4—' O -o c ii O, ii B c ^ c o T3 C CS X) c3 en C U o D- K Pi to .2S > ^•^ c c^c ^ -o •^u] ^ ^ K (U ^ O O ilW M O c 3 <+H .« C Reac h Da •^T3 5 Gordonvill e floo e2 0 't z. )de r (N -73 -C '^ =,-" OO rt t? a u r7c/5 ° en (3N Cu m — r^i: CO h a r- cut o me n ite d s t a> M-, >- Vi S •^ o ic O •*-' ..„ f 11 o a> 6X) (u 11 \o ^ -a C 00 u^ -ri -^ -a \ (N > ^ B£ Co r~l O o a. r^ ro "O^S •n r- c E ON o H T3 —< r^—o - o r-~ E O O- ON C8 en €•> OO (D C -" « •^ O 03 rv ^C oi -s 2^ ^ 1 //"O 1 — 00 X 1 / / c 1 C3 o o o 1 / J OJ —; — a. a. C ori OJ O- cd .-^ C c a r^ m rt in \ei t> Gc OS Q I ^ O 3 O fa ^ 218 5. Post-European channel dynamics in the Bellinger valley dimensions (Figure 5.3). The data obtained from these abandoned channels provide one of the only mechanisms to assess channel dimensions at the time of settiement. It must be acknowledged, however, that there are limitations to this data. The palaeochannels may only provide a minima of cross-sectional area, channel width and deptii, with post- abandonment infilling reducing the values obtained for palaeochannel geometry. Some palaeochannels have experienced an unquantified amount of infilling via oblique accretion and the influx of bedload. There may well have also been some post-settiement dimensional changes prior to the abandonment that may further obscure tiie representativeness of such features. In addition, the nature of channel abandonment (i.e. chute-channel cutoffs) may suggest that the dimensions of any given palaeochannel may not represent a channel that conveyed tiie entire 'bankfull' discharge. This argument would suggest that the small palaeochannels are not truly indicative of the mean channel dimensions within a reach, and that chute channels always conveyed a large proportion of the flood discharge. It is highly likely that prior to settlement, the chute channels were active at conveying flood discharges, although proportions are essentially unknown. As such, the most probable scenario would be of a densely vegetated riparian zone and a distal floodplain depression vegetated by either a sub-tropical lowland rainforest association or a continuation of the riparian gallery rainforest that conveyed flood discharges. These chute-channel zones, both prior and post abandonment, have been targeted for aggregate extraction. Photographic evidence of the extraction practices, along with photogrammetrically derived palaeochannel/chute channel cross-sections (Figure 5.4) suggest that these areas were high level chutes and active in larger floods. Indeed, Figure 5.4a shows one of the chute-channel zones in 1943 (100 years following settlement), highlighting the cut nature of this unit. This figure demonstrates that by 1943 a proportion of the proximal floodplain associated with the palaeochannel remained intact, while the downstream portion of this feature had already been reworked. The chute channel in this figure on the distal floodplain is distinct and appears to be of a 'cut' morphology. The extent of incision of these chutes by 1943 is unknown, but the eroded morphology would suggest the process of channel straightening was underway by 1943. It still remains unknown, however, what discharge was determining the size and geometry of the palaeochannels. Therefore, despite good agreement with historical and anecdotal evidence, the dimensions of palaeochannels obtained within this section must be taken as minima. 219 5. Post-European channel dynamics in the Bellinger valley Figure 5.3 Palaeochannels used in the study; a) 2000 aerial photograph of Darkwood palaeochannel. b) Oblique aerial photograph of Wills palaeochannel. c) Oblique aerial photograph of Gordonville palaeochannel. d) 2000 aerial photograph of CSR palaeochannel. e) 2000 aerial photograph of Bellingen palaeochannel. 220 5. Post-European channel dynamics in the Bellinger valley . -^> .-I* active channel in 1943 (Wills palaeochannel) 0 50m active channel in 1943 (CSR palaeochannel) 0 lOOm Figure 5.4 a) Wills palaeochannel in 1943 (as the active channel) with photogrammetrically derived palaeochannel/chute channel cross-sections. Note the presence of the cut morphology in the upstream cross-sections (i & ii) and the proximal floodplain remnant adjacent to the palaeochannel. Cross-section (iii) shows that the downstream section of the proximal floodplain/chute channel had been eroded by 1943. Bend was finally abandonedin the early 1950s, b) Roodplain/chute-channel at CSR palaeochannel in 1943.Note the high-level chute. Gravel extraction removed the proximal floodplain from 1970 onwards, with the cutoff occurring by 1973. Table 5.1 presents a summary of surveyed palaeochannel dimensions from the four cutoffs on the lower Bellinger River and the single cutoff, immediately upstream of Orama Reach (Darkwood palaeochannel). For the lower Bellinger River the average palaeochaimel bankfull width (W^^) is 65 m, with an average maximum depth (D,^) of 5.7 m and an average cross-sectional area (XS^) of 168 m^. As mentioned previously. 221 5. Post-European channel dynamics in the Bellinger valley cd CN • • ' V V — cd .- cd Cu -a X T3 c cd ir, CN *- c/1 CN -C .1^ -14 9 ii _3 ir, CJ TD a"H C 3 in 20 9 td D •? oa. Vi « 4- .5 U) o o o -a o o .s 3 c c in cd e o o c > CJ S3 o c d d Id o s _c 3 Cd -o Cu w C/!} o Cd C c X "u D o C LH r? ^ o C k. cd 3 3 ,000 8 CJ SLi B o CJ o o "cd o O o _cd « "cd X -a 00 Cu ^ E 00 S o 1 , 1 o ,000 9 C > cd TD 61 o o I O c cd -1 O _cd cd CT" Cd "cd ON CN ii (N ON 186.2 * 5 OJ 'ii Cd B > o >0 r<^ ^. m ON ON W <- Bii O On £- rn d FH d D < C 1—H 4.1 C X 3 Cd a 3 in CJ 13 en c cd .-S 5. 7 X 6. 9 c u 5.3 * o cd TD JS OoJ o c o .s "s a Tf ON ON CN 0) TD 3 13 OJ CJ ON —. CN in in _cd o TD >-• V5 sS- CN cd C c kH CN OJ « > OJ X) o, ^ C td (N CN 00 ON t-- _i s .2 O" ^ cd 0C3 00 TT o E ge r t-~ in 'x -^ TD -a crt cd OJ c/5 li n cd c -S Xl OJ TD 00 U5 e (3 ) 3 s o C (16 ) (3 ) I^ _w o OJ OJ 3 00 ZeJ. / Tl Vi IJ c " W (3 ) c X c ea CL. ca FS X ^^ OJ j= Darkwoo d Bellinge n U ^ a CS R Lowe r Bellinge r < TD o 222 5. Post-European channel dynamics in the Bellinger valley infilling of gravel bedload into the palaeochannels has most likely reduced the attained D^^ values and subsequent XS^ within this assessment. The estimated average palaeochannel W^f is less susceptible to error than are depths, given that most of the post- abandonment infilling has occurred in the lower part of the cross-section. Therefore, while the estimated cross-sectional areas are probably a slight underestimate of pre-European channel cross-sectional areas, they still provide an indication of the gross (minimum) dimensions of the channel prior to European settlement. The reach-averaged palaeochannel geometry from the Darkwood palaeochannel is very similar to the lower Bellinger values with a W^f of 63 m and XS^ of 158 m^. This contrasts with current W^^ dimensions of 115 - 150 m and XS^ values of 650 - 820 ml Predicted bankfull discharges for three of the five palaeochannels that have not been infilled are presented in Table 5.1. These values present the range of bankfull discharges that would have inundated the adjacent floodplains and their equivalent recurrence interval using the recent annual series, presented in the previous chapter. While not providing an absolute estimate of bankfull discharge on the Bellinger River, the data highlight that the palaeochannels contained a bankfull discharge with a recurrence interval of less than 2 years. This differs markedly from the bankfull recurrence intervals for the three reaches in the 20* century (presented in more detail in the following section). The estimated W^j for palaeochannels is probably larger than mentioned by Hodgkinson (1844) of "upwards of one hundred feet" for one site near Orama Reach. If the anecdotal evidence of channel-spanning trees is accurate, then it would seem improbable for a tree to span approximately 65 m from high bank to high bank. Canopy trees at the time of settlement were said to be up to 90 ft before the first branch (Hodgkinson, 1844; p40) and estimated to be 150 ft in height (Maiden, 1894). This would suggest that trees of these dimensions would only span the channel if growing on an inset unit such as the lower levee. Altematively, fallen trees may have only spanned the channel in the most narrow of locations. The palaeochannel data suggest that there may have been considerable post- European changes to the dimensions of both the lower and upper Bellinger River. Chapter 6 will examine the floodplain recurrence intervals in the three reaches, while the next section will outiine tiie timing of the planform changes and provide an assessment of the timing and magnitude of the dimensional changes. 223 5. Post-European channel dynamics in the Bellinger valley 5.4.3 The timing of planform changes on the mid and lower Bellinger River: 1842-2000 Figure 5.2 includes the dates of cutoffs on the lower Bellinger River obtained from aerial photographs, historical and anecdotal evidence. What is apparent from this diagram is that the Bellinger River, like many other coastally draining systems, has experienced widespread planform and dimensional changes. There has been a 7% reduction in channel length between Bellingen township and Thora, with a subsequent drop in sinuosity from 1.21 to 1.14 between settlement and the 1970s. The scale of planform changes along the Lower Bellinger River is small in contrast to other documented post-European planform changes, reflecting the confined nature of the system and the dominance of bedrock and antecedent controls. The adjacent Nambucca valley, which has a similar confined nature, has also undergone a reduction in channel length of approximately 5-7% between 1942 - 1991 (Nanson and Doyle, 1999; Doyle, 2003). In contrast, more alluvial lowland settings in NSW have undergone reductions in channel length that range from 25 - 44% (Erskine and Warner, 1988; Reinfelds et al., 1995; Brooks et al., 2002). The original hypothesis presented by Warner (1987) and Erskine and Warner (1988) of cyclical channel changes associated with flood and drought dominated regimes for coastal rivers of NSW simplifies the nature of channel responses to generalised width and depth responses. The role of planform changes and subsequent incision in influencing the nature of width response has been generally overlooked. Figure 5.2 demonstrates that there are no defined time intervals for the occurrence of channel cutoffs and that in many instances, planform changes have occurred in periods characterised by below-average rainfall and flood activity (Chapter 4). On the lower Bellinger River there appear to be both planform and dimensional changes, some of which have occurred soon after settlement, while others have been progressive adjustments to new catchment boundary conditions. Cutoffs 4 and 7 in Figure 5.2 occurred in the mid-late 19"" century, and are examples of planform changes in the early period of settiement. A period of high flood activity between 1870 and 1894 coincided with a time interval when catchment clearance was well under way. Indeed, the legislative changes in the late 1860s, which induced a shift in the nature of agricultural development, along with historical evidence of estuarine aggradation in the late 19"" and early 20* century, suggest that the floods in this time interval were geomorphically effective (sensu Brooks and Brierley, 1997, 2000). The 1870 and 1875 flood have been documented, with regard to stage at Bellingen township, as the largest 224 5. Post-European channel dynamics in the Bellinger valley floods on record (Public Works Department, 1980). However, the nature of dimensional, riparian and catchment changes in this time period makes a direct comparison of early settiement and current flood stage difficult (Chapter 4). Any chaimel shortening in this lower valley setting, where there is limited outcropping bedrock on the floor of the channel, would more than likely have induced localised steepening and incision. The time interval between 1900 and 1949 was presented in the FDR-DDR model as a time where channels in south-eastem Australia underwent channel contraction and increases in channel depth associated with a shift in the flood frequency curve (Chapter 1). While Erskine and Warner (1988) have suggested that this period is not necessarily characterised by the absence of major floods, it is suggested that the dominant process is one of contraction via sediment storage, through the formation of inset features. On the Bellinger River, this time interval was characterised by below average rainfall (Chapter 4), but two large floods occurred in May and July of 1921. The earliest aerial photographic evidence in 1943 (i.e. in the inferred DDR) clearly shows erosional scarps in many sections of the upper and lower valley settings (expanded in section 5.4.4). The high level chutes in Figure 5.4 demonstrate that many bends were undergoing channel shortening in the early 20* century. Cutoffs 1 and 3 represent sites that more than likely underwent initial modification or partial cutting in the late 19* century floods, but were finally abandoned by the early 1940s (confirmed by the 1943 aerial photograph). Photographic evidence from one of the 1921 floods demonstrates that immediately upstream of Bellingen township floodwaters inundated the entire valley floor (Figure 5.5). Figure 5.5 1921 flood on the lower Bellinger River, ~1 km upstream of Bellingen palaeochannel (Figure 5.2), upstream view. Note valley-scale inundation and the nature of the hillslopes with tiie extensive disfribution of ringbarked frees. This flood has been estimated to be between the 10 and 20-year flood at Bellingen (PWD, 1980). Channel is on far left of photograph (source: Bellinger Valley Historical Society). 225 5. Post-European channel dynamics in the Bellinger valley The two floods in 1921 appear to have induced a range of morphological responses at a time of large-scale catchment clearance. The Never Never Parish maps (3'", 4* and 8* edition) denote Cutoff 1 as the active channel from settiement up until 1934. However, L.Rose (pers. comm., 2002) suggests the cutoff at Rose's bend (Cutoff 1) occurred before he was born in 1923, implying the two 1921 floods were responsible for the final abandonment, while also shifting the erosional foci inunediately downstream. The 1921 floods also produced a cutoff at Gordonville (Cutoff 3), witii the final abandonment confirmed by the 1943 aerial photograph. An additional morphological response to the 1921 floods included 'floodplain stripping' in the form of another high level chute at chute channel 8 (A.Raymond, pers. comm., 2002). Cutoffs 2, 5, 6 and 9 represent the last phase of planform adjustments on the lower Bellinger River. Wills palaeochannel (Cutoff 2) was pre-conditioned by the floods of the early to mid-20* cenmry and was finally cut in the floods of 1950 - 1954. Cutoff 5 represents a small section that straightened between 1943 and 1956, presumably in the floods of the above time interval. CSR palaeochannel (Cutoff 6) was partially cut from the time of settlement through to 1973. The activities of aggregate extraction from the mid 1970s onwards finally led to the cutoff producing CSR palaeochannel, which is currentiy used as a secondary channel. Bellingen palaeochannel (Cutoff 9), like the CSR cutoff, was also a site which was partially cut throughout the late 19* century and into the mid 20* century. The palaeochannel at Bellingen was the active channel right up until the early 1970s and finally abandoned between 1973 and 1976. As mentioned previously, the upper valley setting did not undergo the same planform changes as the lower valley setting due to the dominance of bedrock. The only location of planform changes in the upper valley is the Darkwood cutoff, which was cut in the 1950 flood and also currentiy acts as a secondary channel. 5.4.4 The timing of channel geometry adjustments on the upper and lower Bellinger River: 1842 - 1943 Table 5.2 presents a comparison of channel top width and cross-sectional area between the palaeochannels and the 1943 photogrammetrically derived data. It shows a consistent increase in mean channel width between the palaeochannels and the 1943 widths in all three reaches. The upper valley setting of Orama Reach shows a 35% increase in channel width using the rectified 1943 series. If the palaeochannel widths are used as minima and a comparison is made between the two data sets using one standard deviation (i.e. a sensitivity check), the predicted change results in a net reduction in W^f of 40% between 226 5. Post-European channel dynamics in the Bellinger valley Table 5.2 Percentage change in Wbf and XS^ from settiement through to 1943 Bankfull width - WH. Reach PCh.\V^f PCh. W^. 1943 W^, 1943 W,, Change Sensitivity (m) SO (m) (m) SD(m) % Analysis average Change % Drama 63(5) 25.2 84.9 (26) 21.5 + 35 -40 Wills 65 (16) 20.9 121.9(20) 16.1 + 89 + 24 Gordonville 65 (16) 20.9 156.1 (26) 70.5 + 141 0 Bankfull cross-sectional area - Reach PCh.XS^ PCh.XS^ 1943 XS^ 1943 XS^ Change Sensitivity SD (m') (m') SD (m^) % Analysis average Change % Drama 158 (5) 121 Wills 168 (16) 83 781.5(20) 149.3 + 367 + 151 Gordonville 168(16) 83 610.9 (26) 246.2 + 265 + 68 Palaeochannel (PCh.) W^r and XS^ are widths derived from the palaeochannel analysis, while 1943 values are taken from the photogrammetrically derived dataset. Sensitivity analysis uses a reduction of one standard deviation. All Wb, and XS^ are taken at Q|,r (defined as high floodplain to high floodplain). Bracketed values represent number of x-sections. (-) represents no data available. settiement and 1943. In other words, the accuracy of the derived width series from the aerial photograph and the large standard deviation of the Darkwood palaeochannel data makes difficult any comparison between the two periods. Wills Reach, in tiie lower valley setting, shows an 89% increase in W^^ between the palaeochannels and 1943, whereas XS^ is estimated to have increased by 367% (up to 3.67 times). The sensitivity analysis reduces the increase in W^f to 24% and XS^ to 151%. Table 5.3 presents the attributes of the 1943 channel geometry for Wills Reach along with the predicted bankfull discharge and associated recurrence interval. What is clear from this table is the increase in channel capacity by 1943, such that Q^f was estimated to be 1820 - 3030 m^/s relating to a recurrence interval of > 50 years on the annual series and > 20 years on the partial series. Similar ti-ends are evident for Gordonville Reach with a predicted increase in W^f of 141% and an increase of 265% in XS^. As documented in Chapter 2 and 3, Gordonville Reach is characterised by a greater diversity of inset alluvial units, including the development of both the high and low floodplain in addition to in-channel benches and elevated flood- channels. As such, XSA varies considerably within the reach (Table 2.4, Chapter 2). The predicted percentage change with the sensitivity analysis reduces the increase of W^f to no change, and a predicted increase in XS^ of 68%. Table 5.3 shows that by 1943, Q^.^ varied considerably witiiin Gordonville Reach, witii a mean Q^f of 1120 - 1867m^/s. This represents a range in recurrence intervals of 5 - 50 years on the annual and partial series; considerably less frequent than the predicted recurrence interval for the palaeochannels. Unlike Wills Reach, Table 5.3 highlights that in 1943 estimated XS^ between pools and riffles differed markedly, producing the 40% standard deviation, with riffles 227 5. Post-European channel dynamics in the Bellinger valley o _ o c CO CO u o o o CM >0 ii a> u I I I CM o 2 >n rJ O CM 3 fi A A A >5 0 "^ ^ 10- 2 X T3 U a| al u cd u. >a > > i; cd Q c 3 X. CO 2. •* CM 00 "" I I I fsi o\ \o »o en C\ r^ o CM Vi 25 —I r~ Tt 00 00 O —' Vi Vi ^ o 00 o oo c^ f~- 00 —' >n o CM cn .S 1' tT) -H CO -H cv| —c « OS ri -H X o cd nl y 1 ^ ro "o c«^ o o f ^ 1 d O T3 low - 0.00 1 0.00 1 0.00 1 gt h C Q = cd 1) >» CO ov e —' X X cd >> -o o o o 2; 2 2; o o o o o o Wi l M •a^ cidi o o o o o o » OS OS 0\ 00 U E 1 I I 00 tr « T3 la b in «o io rn ro en C 3 Cd es e "cd va i cd \q en .—1 X) Tt so m ep r I I I al u C reduc i «n o CN 0^ Tt en ii flood p > T3 ^ I I I OS CN CN in CO P^ J3 O 1—1 o Tl- en >o cd T3 X 00 to CN ii i C M cd T3 2 CO > o Cl in OS cd ;-• 61 I I I ^H ^ - 2. 4 CN ? 1 ^ d co^ a n X d e a l plai n Cu e , Dashe s Tt- rn I I I 4 . 4. 3 4. 2 6. 4 ectio n floo d 3.9 8 otogr a r CO JS aeoslo p ; h n> .i5 OO Cd •=; OS CN OS CN Tt rial p m cross - T3 E Chap l Tt 0C3 c/3 .s ^ o\ cd "o G 1—1 t o fro i Po i a s CO in OS >o o\ estimat e r-H 0. 2 0. 9 2 < ectifie d ons . i s ne d T—t •^ o 00 ?^ lo resente d 77 7 78 3 SO Tt ^ m % cd r- E o 0 "^ ps p -see l CO a OS 00 Tt 00 00 Vi •.s ii o c CO s5 O ^ o ^ d cr o ive d ion s ts 1—1 der i OS OS o cn ph o Tt ne d f r OS m sdon a (6 ) <>^ fr o so ^ bas e (N 00 (26 ) ^ mea n dept h a l s obta i Ci le d X ii Xo —^. cd O S iffl e eac h btai r tilis e 'alu e ool ( u o a o Ct-( sdmate d fro m 3 U inter v 5 0^ o* (5 OS Cu OS ?• o 228 5. Post-European channel dynamics in the Bellinger valley having larger bankfull capacities and greater recurrence intervals than the smaller capacity pool locations. In summary, the results for this section suggest that there had been an increase in both W^f and XS^ from settiement to 1943. Sensitivity checks, however, identify the need for caution in stating the magnitude of the change or that the change necessarily occurred at all sites. 5.4.5 The timing of channel geometry adjustments on the upper and lower Bellinger River: 1943 - 2000 The upper valley setting - Orama Reach Table 5.4 presents a summary of channel attributes for the four time periods in the three study reaches. Three sets of width data were measured from the rectified 1943, 1956 and 1973 aerial photographs in Orama Reach from at the same grid co-ordinates used for the current field data. Table 5.5 presents the statistical summary of both t-tests and the non- parametric Wilcoxon signed rank sum test for the three reaches, while Table 5.6 presents a summary of percentage changes in hydraulic variables between the major time intervals. The width data sets show a consistent increase in W^^ from 85 m in 1943 to 90 m in 1956 and 1973 and a further increase to 115 m in 2000. The data within Table 5.5 show that these changes are all statistically significant (p < 0.05), with the exception of width changes between 1956 and 1973. Between 1973 and 2000 the W^f of riffles increased more than at pools (Table 5.5). While the data suggest a net increase of 35% in W^f between 1943 and 2000, these differences could be influenced by the level of accuracy between data types, with the field data being more accurate than data from the rectified images. Figure 5.6 demonstrates the nature of channel changes in Orama Reach between 1943 and 2000. As mentioned in Section 5.4.3, there have been only limited planform changes in Orama Reach due to the bedrock control and the incised nature of this upper valley setting. Figure 5.6 displays erosional scarps and bare gravel bars in the 1943 aerial photograph. The 1956 aerial photograph (Figure 5.6b) identifies that many of the vegetated bars and inset sedimentary units were reworked in the floods between 1943 and 1956. Older residents within this middle-upper valley setting have remarked that the bed of the channel doubled its width in the floods of the early 1950s (T. Joyce, pers.comm., 2001). 229 5. Post-European channel dynamics in the Bellinger valley a w Tt- — _ tn CM _3 cd vi Tf v a 197 3 wit h , an d en o\ OC} Ov o 7. 4 7. 3 o ed . 194 3 rsi CJ\ s-secdon s PJ — 00 CM V\ OS Vi < -H en r-: OS OS 'kn_>l >o en — kpH kO CM — 00 lO rs) m O 0 X r- r-i o TD —* C4-I 0 0 C 1) X 4> •d- -J "^ 52. 8 148. 2 204. 9 na m hen c = c d Z cd c cd en c^ VD 0c g u. so o ^ — vi eta t aris i a. 00 hoto g E >a D. 0 r^ C l 5. 7 7. 1 b y aeri a fo r e d •a kH so Os 3 0 o\ tj CO t^ — C3S Ul c^ ii X TD < CM S kH — CM 0 CD 00 — ^ 0 X sC- tCM r^ 0 kH <+3 > las - re c kH CO dat a 56. 3 174. 1 _a — c^ — 217. 3 Ctio n yth e TD 0 X Cfl "S U 1 TD < 0 OS S a 0 kH X 0 X•5 X fM TD 0 C Os ii ar e X riv e o CO <=> "rk <> d e 59. 6 Qs en OS 163. 8 ii 212. 3 X wit h o whic h cd y lall y ii CJ CO c kH -fl o cd X OS 1-H Og r 0 Ov OS CM •* _ 0 Vi w~^ oci o K so X •kH > OH CO o 0 Cd X c kH Cd c Xl U cd Os .^ 0 Vi en 0\ CM 00 DH c -; OS d VD 00 D Vi 00 Tt rr vo TD "cd X N—O eM 00 O kH TD a 0 >kH > s SO re a o 5 cd c TD en '-H a <^ -H OS lO vo •*-» u 00 vi z; VO d d >> 15 6 ^ .-H sn 24 4 cu alue s varyi ; i V ^ 53 ft; onl y photogra m E5 1^ OS Mean Rang Al l 230 5. Post-European channel dynamics in die Bellinger valley T3 *» on CO CO on * c/3 C * * Z Z Z * * Z * Z Cd o Si I I 1 on on on CZ3 ^ CO 00 CO on on F Z Z Z Z * e Qn Q z z z z z CO 03 f^ « CO CO on CO c/3 CO CO „ CO c^ 9s < 1—1 e C« Z Z Z Z Z Z Z * Z i- CO * * ^ * * * * on * on C . gg* * Cd * * z * * * Z * Z > e U e Vi M Qa X CA > > se 0! »s * C/J c/3 C/5 on . on on on on ^ on CO on T.H g * Z Z Z ^^ * Z Z Z z z z z z o TD rr) B« U t-- Q en TD *-( b X Cd CA CA 43 ve B< 00 oo CO on c/3 CO CO CO ^ ^ CO C/3 on ^ * * Vi 19 5 g * * Z Z Z * * Z Z Z c z z z z z o cd > 5: kH o ii on c/3 on c/3 on * * CO * c^ Z Z Z cd _c Z z * * Z * Z TD ii O S C * * on on on czi oo ^ on oo > * Z Z Z o * Z Z * Z Z X c CO cd Cd o u in 2 Q CO on on CO c/3 * * on on on 'S X 2 Q Vi 00 X z z z z z * * Z Z Z cd o Vi * CO c/3 CO * CO CO * ii CO CO ^ CO c/3 X ^^ * Z Z Z * Z Z * Z Z * Z Z c Z ii * * CO „ on oo oo CO c/3 c/3 * * Z * Z Z Z Z Z Z O o (t^~ «Q CO CO c/3 GO c/3 * o *V [^ * * ii * ov z z z z * Vi * * eo « 13 Q on t/3 O 19 4 on Q X Z Z V Cu * on * on * « * * CO * c/3 CO Z * Z * * * Z * "5 * * z * z TD ii eo •4-t ed Vi C/5 td ve TD V) X OS TD 1-H FD R CA 3 cd $ > CO 00 on CO CO c/3 on on on CO on CO E 9 V. E DD R >" z z z z z z z z z z z TD 3 _>;v C/5 TD 15 o CJ E Oi o ftj o kH TD B c c oe: o U .2 2 cd c cd TD TD TD B Cd 'kH cd O Cu ^ > £c 5: 0H Cd C ^ > £ E £1. ^ > = E B 231 5. Post-European channel dynamics in the Bellinger valley caption over page 1km 232 5. Post-European channel dynamics in the Bellinger valley Please see print copy for image Please see print copy for image Figure 5.6 Unrectified time series of channel changes in Orama Reach 1943 - 2000. a) In 1943, with active bars and partly vegetated bench and channel margins (denoted with black and white arrows respectively), b) In 1956 following the floods of the early 1950s demonstrating, no change in planform but some changes in bar and bench morphology, b) - d) Highlights the revegetation in Orama Reach between 1956 - 2000 with both bars and benches becoming colonised with Casuarina cunninghamiana and exotic riparian species. Flow is from left to right. 233 5. Post-European channel dynamics in the Bellinger valley While the 1956 aerial photograph certainly reveals an increased distribution of bare bar surfaces it is not possible to quantify the extent of these bed-width changes. Despite die persistence of above-average flood activity into the 1970s, Figure 5.6c shows that many of the active bar surfaces seen in the 1956 aerial photograph had become partly vegetated by 1973, with most being fully stabilised by the 2000 aerial photograph (Figure 5.6d). Channel geometry changes in the lower valley setting - Wills Reach On initial inspection the aerial photographs for Wills Reach from 1943 to the present show litde change. Indeed the majority of the reach appears to have undergone some change in channel geometry not visually obvious from the photos, but which become apparent in analysis of the photogrammetrically derived cross-sections. However, what is apparent from the photos are shifts in bar and riffle location, adjustments to the morphology of the inset sedimentary units, and isolated examples of major planform and width adjustment following the floods of the early 1950s. The 1943 cross-sectional data suggest that much of Wills Reach could have contained a Q^f of 1820 - 3030 m^/s (equating to > 50 years on the annual series and > 20 years on the partial series). Figure 5.7 highlights the entrenched nature of Wills Reach by 1943 and the subsequent channel changes in the following 57 years. The aerial photographs in Figure 5.7 demonstrate that one of the responses in the lower half of Wills Reach has been an adjustment of the riffle- pool spacing, and a shift of the erosional foci in the low amplitude bends in a northward direction between 1943 and 1956 (presumably as a result of floods between 1950 and 1955). The series of large floods in the 1950s appears to have eroded the riffle opposite the unnamed tributary confluence, forming a long pool (still current in today's longitudinal profile). The 1964 - 2000 aerial photographs show that this shift in the erosional foci has been accompanied by a re-positioning of altemate bars, with the bar platform between W13 and W14 becoming progressively vegetated by Casuarina cunninghamiana from 1973 onwards (Figure 5.7). Figure 5.7 also highlights temporal changes to a large inset bench between W5 and W8. This unit, whose sedimentological attributes are examined in Chapter 6, is comprised of an active proximal bar, an elevated dissected bench, and a distal flood-channel. The 1943 aerial photograph clearly shows the presence of an erosional scarp on the left-bank, with a predominantly unvegetated bench unit. The subsequent aerial photographs present evidence of reworking of the surface of the unit, part revegetation by 1964, followed by the removal of the vegetation by 1973 up until today. The 1943 and the 1973 photogrammetrically derived cross-sections for this bench show a cross-sectional form 234 5. Post-European channel dynamics in the Bellinger valley X a -a c i:;, i (J X C ii OJ ^^m^ Please see ca -^ en print copy for print copy for an d bet w n W1 2 li s image image image >.2 c e-c1 ^fro m ;> c^ k- k- > c/5 > •7 1^ £ — o F l b e h . a r riffl e sion a ange s x^^ 0 ^ ^ 0 i^ S s ii ^ <^ c -^ C 0 C rt .k- of ; cat i ns e ha n Ite r (J 0 '^ a • — <*., ~~ ii '^ a 0 H x: ^, X Vi X^-| ,-^ aVi ^-^ S 0 c ^c c/3 C 0 (U 0 OJ «J -z: ^ -2 .| - ^ ^ 5 •" ^ 00 0 0 •q 0 a (o • — ^^ ^j S ^i2 Cd « = 50 0 B U i 5. 7 an d W 8 14 . Arrow s right . 5 gur e 43-2000 . positionin g E2: 2^^2 ^-^ L © Please see Please see print print copy for copy for image image 235 5. Post-European channel dynamics in the Bellinger valley not dissimilar to that of today (expanded in Chapter 6). The channel geometry obtained for the four time periods in Wills Reach are shown in Table 5.4 and Figure 5.8. Differences between time intervals show a divergence from a normal distribution in four of the six comparisons, resulting in the use of both a paired t- test and the non-parametric Wilcoxon test (Table 5.5). The estimated changes in W^^ are significant (p < 0.1) between 1943 and 1956 and between 1943 and 1973 (p < 0.05). This suggests Wills reach underwent a statistically significant increase in W(,j. of 12% between 1943 and 1973. This was followed by a statistically significant decrease in W^^f of 6% between 1973 and 2000 (p < 0.1). Despite the net increase in W^^ between 1943 and 2000, variance has not changed significantly. The response of either pool or riffle units is also shown in Tables 5.5 - 5.6 and Figure 5.8, highlighting that there has been a significant difference in the rate of change in Wj,^ between pool and riffle locations between 1943 - 2000 and 1973 - 2000. Within Wills Reach the evidence suggests that W^f of pools have preferentially eroded in contrast to riffles, with pools increasing by 8%, in contrast to negligible net change at riffle locales. The decrease in W;,^ between 1973 and 2000 also appears to be characterised by a preferential decrease in riffle locations in contrast to pools, with Wj,f at riffles decreasing by 12% in contrast to a 2% reduction at pools. Thus, the data suggest that pools have undergone a net increase in bankfull width, in contrast to riffles, which experienced an increase in the FDR but have subsequendy decreased. The trends in W^f are also replicated for XS^. However, rather than a decrease since 1973, reach-averaged channel capacity has remained high (Figure 5.8). Tables 5.5 - 5.6 demonstrate a statistically significant increase in XS^ of 4% between 1943 and 1973 (p < 0.1), and a 5% increase between 1943 and the present (p < 0.1), with no significant difference between 1973 and 2000. Tables 5.5 - 5.6 also demonstrate that the increase in XS^ between 1943 and 1973 was not a function of differential erosion at pool or riffle locales, with both units undergoing an increase. Interestingly, the data suggest that there was a significant change in the variance of XS^ between 1943 and 1973, suggesting diat channel dimensions became more uniformly large with a reduction in the standard deviation in the last FDR (1948 - 1977). The lack of significant changes in reach-averaged XS^ between 1973 and 2000 is a function of continued increases in XS^ at pools, with an equivalent decrease at riffles. This suggests that while reach-averaged Wj,f may have reduced, the nature of the XS^ response has been for the continued maintenance of a large capacity channel. In addition, it appears that in-channel sediment storage in the last 30 years has reduced cross-sectional areas at riffles in contrast to pools, which have continued to increase by 6%. Since 1973 there has also been a statistically non-significant 236 5. Post-European channel dynamics in the Bellinger valley ^ So ^•• tf Q e Q ^3 c — cd O (U o ei e <^ e< \ -a != Q \ % (U \ Xiu a ^ ^ \ ii-^ !< •- a Q \™ Vi o Q **** 3 2 cd 'C > cd QO II fi 0) ON tn ec CS CM !>O)• cd " es i5 rs .~ >, Cd S.I Si r< Cd c S «= 04 -si U) (U &0 (8 42 Vi xi o 0M c — ed T3 Xi (U O T3 -" c S ^ JQ If 3 on Cd s. 00 -^ «3* 3.5 BO 237 5. Post-European channel dynamics in the Bellinger valley * !> W) 00 o^ -; SO OS tS — s ^ + + oCS : + 00 * * OS o) OS o OS r- CO -H vd o rj in + + + * x~- rn \o * CS d ^ \o CO Vi + + < ^—» ii o o c a cd * « o * * * * * * vo CS OS •^ Tt n in 0\ t< CO 1—t u o CA 00 00 s + + ca cd o o * ii Tt -; 60 * \D rn r-i rn CS cd t~: so ^ C so I i + + tu CO o ;-i "53 o 00 ii X a •^ + + CO cd Os CO + + -) C/l C I I I H cd ^ X J3 Os O ii 0J3 * * o Tt OS 00 1-H Ct.H Cd * m 0\ CO 1—I •k.1 vo '^ -^ Di c O i£) lo 1 oii so + + + + + + + kH + o CL d V ^ ?^ iI*3: cd 'C o s< POOL RIFFLE REACH POOL RIFFLE REACH * e S p fci * e a? a, as 3 o' C/5 > d a o V ••...4 kH cu -Cl o E3 O * 238 5. Post-European channel dynamics in the Bellinger valley increase in the standard deviation of XS^ (Table 5.5). The channel expansion that occurred between 1943 and 1973 also resulted in a statistically significant increase of 0.5 m in maximum depth (D^^J at the reach-scale and at both pool and riffles, equating to a 6% increase (Table 5.6). Figure 5.8 shows that there has also been a slight reduction in reach-scale D„^ between 1973 and 2000, in addition to the documented reduction in W^f. This slight reduction in D^^^ comprises a variable response at pools and riffles, with riffles reducing in D^^ by 5% in contrast to continued increases at pools (statistically significant at p < 0.1 - see Tables 5.5 - 5.6). In addition to the changes in D^^^ are the changes in mean depth (D^^J. The larger increases in W^f in contrast to die increases in XS^ resulted in a statistically significant decrease in D^^^^ of 7% between 1943 and 1973 at the reach-scale and at both pool and riffles. The larger decrease in bankfull width and negligible changes in cross-sectional area since 1973 have yielded a reduction in D^^^^ at both the reach-scale and at pool and riffles (Table 5.6 and Figure 5.8). The analysis of the channel geometry data demonstrate the maintenance of a large capacity channel throughout the previous FDR and into the current DDR. While reach-averaged conditions display the presence of a large channel capacity, other channel geometry variables do display some tendency to be undergoing a reduction, with variable response between pool and riffle locales (e.g. bankfull width). In addition to the documented changes in channel geometry for Wills Reach, there are also sites of major planform change. Figure 5.9 presents a site immediately downstream of Wills Reach where there have been planform and width changes between 1943 and 1973. The 1956 aerial photograph shows that while Wills palaeochannel was effectively cutoff in the floods between 1950 and 1955, additional planform changes occurred immediately downstream of Wills Reach in the form of a chute-channel cutoff. The series of floods between 1950 and 1955 effectively shortened the channel by cutting off Wills palaeochannel (Figure 5.9i) and eroding the left-bank proximal floodplain, leaving an isolated channel marginal remnant as an island (see Figure 5.9(ii) for comparison between 1943, 1973 and today). Figure 5.9 (iii) also demonstrates the nature of convex bank erosion at the structural pool immediately downstream. At this pool location the 1943 aerial photograph shows the presence of some remnant or post-settlement riparian vegetation on the alluvial left-bank and the presence of an erosional scarp on the upper left-bank margin. The 1956 aerial photograph demonstrates the eroded form of the alluvial convex bank, while the 1964 and 1973 aerial photographs highlight the continued usage 239 Please see print copy Please see print copy for image for image ^1982 ') 25 Please see print copy Ij for image 0 100 200 300 400 500 0 100 200 300 400 500 600 Please see print copy Please see print copy for image for image Figure 5.9 Unrectified time series of channel changes downstream of Wills Reach at Gordonville cutting 1943-2000. Note the progressive cutting of Wills palaeochannel (i) from 0m 1943 to 1973. This Page is Blank 5. Post-European channel dynamics in the Bellinger valley of the chute-channel as the primary channel, witii the 1982 - 2000 aerial photographs demonstrating the progressive revegetation of the enlarged channel. Since 1973, both the left and the right-bank bars underwent aggregate extraction that had ceased by 1996. The current channel-form in the vicinity of the chute channel is a widened channel witii inset benches and a dissected proximal right-bank bar-platform (see Figure 5.9(ii) for current cross-sectional form). The left-bank bench has been progressively colonised by Casuarina cunninghamiana since 1973 resulting in an even-aged stand of river oaks < 30 years old. Since aggregate extraction activities have ceased, the post extraction left-bank bar has become dissected and revegetated by river oaks less than six years old. The right-bank bar-platform is a dissected unit with both remnant river oaks > 59 years old and younger individuals that are both post-1956 and post-1973. Additional early successional species such as Ficus cororuita and exotic shrubs and climbers such as Ligustrum sinense and Cardiospermum grandiflorum have stabilised the inset features. The alluvial convex bank at the structural pool has also been partly revegetated by post- 1973 river oaks, but is actively grazed preventing any additional further regeneration of riparian species. While falling outside Wills Reach, the example presented in Figure 5.9 highlights the localised width and planform changes that occurred in the floods of the 1950s. Channel geometry changes in the lower valley setting — Gordonville Reach The rectified aerial photographs and photogrammetrically derived cross-sections show areas of considerable planform and channel change to the low floodplain and inset sedimentary units within Gordonville Reach. Figure 5.10 presents a time series of non- rectified images in the middle part of Gordonville Reach. This figure demonstrates that the series of floods between 1943 and 1956 resulted in the downstream migration of the bend (i), the reworking of previously vegetated bars and the removal of riparian vegetation. These responses are in addition to a cutoff/avulsion that occurred in the early part of the 20* century (Cutoff 3 in Figure 5.2). The channel in this mid-section of Gordonville Reach continues to rework bars and erode the low floodplain/flood channel complex on the right-bank (denoted as (ii) in Figure 5.10). Figure 5.11 presents a summary of planform and cross-sectional changes in Gordonville Reach between 1943 and 2000 derived from the rectified images and the photogrammetrically derived cross-sections. What is apparent from Figure 5.10 and 5.11 241 5. Post-European channel dynamics in the Bellinger valley Z^ Please see print copy for image s o L © 242 5. Post-European channel dynamics in the Bellinger valley t^ +J Please see print copy for image Please see print copy for image ;-Li O 3 o a c T3 n -o ed 6 "EH c T3 o CO o ii c o ha n n J4 Ti) c cd C X) cd Xi J3 o 00 'S ;-c y^^ Please see print copy for image OS 1—1 fi •n c cd -(io> f^ •-M -t .4-1 OS di ^^ ^ c pc 243 5. Post-European channel dynamics in the Bellinger valley i) alluvial pool (currently infilled) NEVER NEVER RIVER ii) alluvial pool Gordonville palaeochannel iii) alluvial riffle iv) structural pool 0 100m v) alluvial pool Boggy Creek Gordonville palaeochannel abandoned 1901-1921 1943 channel margin 0 1km 1956 channel margin L. 1973 channel margin 2000 channel margin Figure 5.11 Planform and cross-sectional changes in Gordonville Reach from 1943 - 2000. Hanform changes are based on the rectified 1943,1956,1973 and 2000 images. Note the low floodplain erosion and the 1901 -1921 cutoff in the centime of the reach. Cross-sectional changes are based on the 1943 and 1973 photogrammetry data and represent areas above low-flow only, while the 2000 sections represent field data and include the whole cross- section. Representative sections for pool and riffle units highlight the nature of channel expansion in Gordonville Reach. 244 5. Post-European channel dynamics in the Bellinger valley is that this middle section of Gordonville Reach, which falls at a point of reduced effective valley widtii, has undergone considerable shifts in channel position since settlement. The first avulsion was from the right-bank to the left-bank valley margin between 1901 and 1921. The channel remained in this location through to at least 1973, with further lengthening of bend (i) in a downstream direction during the floods of 1950 - 1973 (Figure 5.10). In 1976 aggregate extraction commenced on the right-bank bar-platform in the form of 'bar harvesting' (Section 5.4.6). This continued up until 1991, resulting in further straightening and the progressive abandonment of the 1943 channel position. By 1994 this middle section of Gordonville Reach had undergone partial infilling of the 1943 channel and partial revegetation of the now artificially widened channel, with a locally braided planform (Figure 5.10). This figure shows the continual revegetation of the locally braided section with bars and islands becoming colonised primarily with Casuarina cunninghamiana, Ficus coronata and some Waterhousia floribunda. Islands have become bank attached and are currentiy accreting gravel and fines, with continued migration and accelerated erosion of the partially vegetated right-bank flood channel/low floodplain (denoted as (ii) in Figure 5.10). Figure 5.12 presents the nature of channel changes in the lower half of Gordonville Reach. The series of floods in the early 1950s that reworked the middle section of Gordonville Reach also reworked the partly vegetated inset sedimentary units at two locations (denoted as (i) and (ii) in Figure 5.12). This figure highlights the reworking of the large inset low floodplain/flood-channel on the right-bank between 1943 and 1956. The 1943 aerial photograph also shows the unit denoted as (ii) to be a partly vegetated tributary confluence-bar and riffle associated with Boggy Creek. The 1950s floods effectively removed the vegetation from this bar, and like other locations shifted the erosional foci within the channel. Associated with this shift in the erosional foci has been the formation of a large point bar opposite Boggy Creek (denoted as (iii) in Figure 5.12). Despite the reworking of the tributary confluence bar, the 1956 aerial photograph demonstrates the continued presence of the reworked bar and associated riffle. However, the 1964 and 1973 aerial photographs show the removal of the tributary confluence- bar/riffle and the formation of a long pool (still current in today's longitudinal profile). The formation of the point bar opposite the bedrock concave bank was presumably a function of the shift in the erosional foci, coupled with an increase in sediment supply from eroding banks upstream. The planform of this location has remained constant since 1973 with both the inset sedimentary unit denoted as (i) and (iii) becoming partly vegetated. 245 5. Post-European channel dynamics in the Bellinger valley 1943"^ 5*«:' '^•4i''^^'"*~ tAfe 1982 '^ Please see print copy for Please see print copy for image image I prK«--'«'» 2000 ^ Please see print copy for Please see print copy for image image N A 0 500m 246 5. Post-European channel dynamics in the Bellinger valley Facing page - Figure 5.12 Changes to lower Gordonville Reach from 1943 - 2000; i) represents an inset alluvial unit that was reworked between 1943 and 1973; ii) Tributary confluence bar/riffle that was reworked and removed between 1943 and 1964; iii) Point bar formation from 1956 onwards. As shown previously in Table 5.3 predicted bankfull capacity for Gordonville Reach in 1943 could contain discharges that ranged between the 2 and > 50-year recurrence interval, witii riffles having a larger cross-sectional capacity than pools. The current channel geometiy witiiin Gordonville Reach still displays a large variability in bankfull capacity witii some pools still having smaller channel dimensions. The general post-settlement trend, however, appears to characterised by a decrease in the variability in dimensions ofthe channel throughout the reach. This decrease in variability is associated with increases in channel capacity at pool locations. The geomorphic responses of pools and riffles within Gordonville Reach to large floods within the 20* century, however shows more spatial variability than Wills Reach, and will be discussed in more detail with the analysis of the channel geometry data. Table 5.4 presents the summary of the channel geometry for Gordonville Reach, while Tables 5.5 - 5.6 present the summary of the statistical analysis and the percentage changes, with Figure 5.13 graphically representing the changes. W^^ in Gordonville Reach shows a statistically non-significant increase of 5% between 1943 and 1956, with a further 6% increase between 1956 and 1973 (statistically significant at p < 0.1), while the ~ 12% increase in W^^ between 1943 and 1973 is significant (p < 0.05). Like Wills Reach, Gordonville Reach has also undergone a statistically significant decrease in W^.^ (15%) between 1973 and 2000. Table 5.5 also highlights a statistically significant difference in variance in W^f between 1943 and 1973. This reduction in standard deviation between 1943 and 1973 suggests that the variability in W^^ decreased between the end of the last DDR (1900 - 1947) and the last FDR (1948 - 1977), with a change from a 45% standard deviation in 1943 to 32% in 1973. Since 1973, while W^^ has undergone a reduction, variance has remained lower than 1943, with a statistically non-significant change from 32% in 1973 to 36% in 2000. The documented changes in W^f are shown to be significantiy different at pool locations over riffles between 1943 and 1973 (p < 0.1 only). The 12% increase in Wj,f between this interval is a function of pools preferentially expanding over riffle locations, with a 25% increase at pools in contrast to a 5% increase at riffles. 247 5. Post-European channel dynamics in the Bellinger valley l/^ ^ u^ u^ vo ^C^OOOVC'^CMOOO 00 r- »£> >ri vd vivivi-^rfxt-^-^tn ed 248 5. Post-European channel dynamics in the Bellinger valley Tables 5.5 - 5.6 also show that in Gordonville Reach there has been a statistically significant increase in XS^ of 15% between 1943 and 1973. This increase comprises differential erosion at pools over riffles witii a 25% and 10% increase respectively (statistically non-signiflcant). XS^ within the reach has continued to expand with a further statistically non-significant 3% increase in XS^ between 1973 and 2000. This suggests that XS^ has progressively increased from 1943 to the present. Like Wills Reach, variance of XSA area shows ^ statistically significant decrease between 1943 and 1973, suggesting that the floods in this time interval were effective at reducing the variability of reach- averaged channel capacity, lowering the standard deviation from 40 to 29%. Tables 5.5 - 5.6 also highlight that in addition to the increase in channel capacity since 1973, variance has increased (p < 0.1 only). The documented changes in XS^ are shown to be significantiy different at pool locations over riffles between 1973 and 2000 only, with pools decreasing their cross-sectional area marginally, while riffles have continued to increase by 15% (p < 0.1). Figure 5.13 and Table 5.6 present the changes to both D^^^ and D^^^^. What can be ascertained from Figure 5.13 is the progressive increase at the reach-scale in both D^^ and Dmean f^^m 1943 to 2000. Deviations from this pattem are shown at riffle locations with a predicted 6% decrease in D^^^ from 1943 - 1973 and then a further 6% increase from 1973 - 2000. These changes in D,^ are statistically non-significant and equate to < 0.2 m of difference. This contrasts to D^^^^^ which has increased by a statistically significant 0.95 m (p < 0.05), and is in keeping with the larger net increases in channel XS^ and the slight reduction in predicted bankfull width from 1943 to 2000. The predicted preferential erosion of pools in Gordonville Reach between 1943 - 1973, and of riffles between 1973 and 2000, along with the predicted decrease in reach-averaged W^f warrants further attention. Figure 5.11 (i) and (v) highlight the nature of channel expansion at two alluvial pool locations demonstrating the magnitude of the changes between 1943 - 1973 with XS^ and W^f cleariy increasing. Cross-section (ii) in Figure 5.11 also demonstrates the nature of W^^ reduction in Gordonville Reach since 1943. While post-European changes to floodplain morphology have not been the focus of this section, cross-section (ii) highlights the nature of post-settlement modification to the low floodplain in the upper part of Gordonville Reach. The reduction in Wj,^ in both Wills and Gordonville Reach is often a function of proximal floodplain deposition on previously eroded surfaces (see cross-section (ii) Figure 5.11). Proximal levee and large upper-bar deposits, such as that presented in cross-section (ii) have effectively reduced W^^ by up to 100 m. Indeed, this location shows considerable floodplain dissection and accretion since 249 5. Post-European channel dynamics in the Bellinger valley 1943. This style of W^f reduction has been previously incorporated into the FDR-DDR model (Warner, 1994a), but for DDR intervals. Hence, it would seem that W^f actually decreased in some locations in the FDR and continues to experience proximal deposition, resulting in reduced W^f values. While some cross-sections in Gordonville Reach have shown a reduction in both W^f and XSA since 1973, the reach-averaged conditions have been dominated by continued channel expansion, such as presented in Figure 5.11 (iii, iv and v). Indeed, the 14% increase in cross-sectional area at riffle locations is dominated by the continued erosion of the right- bank low floodplain in the middle of the reach (see bend (ii) in Figure 5.10). This riffle locale is still undergoing channel expansion with the 2001 floods further increasing channel capacity . A Summary of Channel Geometry Changes in the Lower Valley Setting This section has presented a detailed analysis of the nature of changes at the reach-scale and at the unit-scale. The channel geometry analysis presented for Wills and Gordonville Reach has shown that the evidence for cyclical channel changes associated with the inferred climatic regimes is equivocal. There is evidence of continued increases in reach- averaged XSA since 1943, a reduction of W^.^ since 1973, with the most pronounced changes being a re-configuration of in-channel units and a net increase in pool geometries. What this section has identified is the presence of unit-unit variability in the response of the two reaches to the previous FDR and the current DDR. The following section addresses the nature of aggregate extraction on the Bellinger River and its superimposed role on channel morphology. 5.4.6 Aggregate extraction in the Bellinger valley: Evidence for cliannel incision Like many valleys in coastal NSW, the Bellinger valley has undergone various intensities of in-channel and floodplain sand, soil and gravel extraction since settlement. In addition to the adverse ecological impacts of gravel extraction, the geomorphic implications of such practices have resulted in major channel incision and channel expansion in many river systems of New South Wales (Nanson and Doyle, 1999). The Bellinger valley has seen a progressive phasing out of in-channel exti-active practices in non-tidal reaches since 1993 due to tiie adverse geomorphic impacts. This section addresses the location of past gravel exttaction operations, attempts to quantify the volumes removed after records were started and assesses evidence for channel incision. 250 5. Post-European channel dynamics in the Bellinger valley Informal aggregate extraction on the Bellinger River has been in operation since tiie settiement of the valley with the utilisation of sand and gravel for building purposes and road construction. Indeed, Section 5.2.2 highlighted the practice of dredging in the tidal zone from the late 19* century onwards. Quantifying volumes of material removed and tiie geomorphic impact of eariy informal extractive practices is not possible. By 1930 - 1934 two mechanised aggregate crushers were in operation, with loading and carting still being carried out manually with pick and shovel, and horse and cart (L. Rose, pers.comm., 2002). By 1938 early model Ford trucks were in operation which were being used to cart gravel, with loading still occurring by hand. The era of the second worid war saw the introduction of more mechanised removal and cartage techniques, with the post-1950 flood debris being cleared by eariy model trucks and dozers with front-end blades (L.Rose, pers.comm., 2002). It is suggested that the initial removal of alluvium from bars using non-mechanised techniques such as carts would have constituted a very small proportion ofthe total sediment load of the system. Formalised extraction appears to have started in the valley by 1949, with documented evidence occurring between 1970 - 1994. Aggregate extraction records are sporadic and vary depending on whether the land is subject to government royalty payments. Extraction activities on govemment land have resulted in a fixed royalty payment with littie bearing on the actual volume of material removed, while extraction activities on private land have only been subject to regulatory permits and hence there was no monitoring of the volumes of material removed. Resource Planning (1989) when commenting on the nature of records and monitoring of gravel extraction practices, state: " In general, it can be assumed that gravel has been extracted from virtually every bar in the river at some time over the last 30 years." Appendix 5.4 is a compilation of estimated volumes of material removed from the Bellinger River alone, along with a figure highlighting the distribution of past gravel extraction operations for the lower valley setting, incorporating both Wills and Gordonville Reach. Estimates made within the records indicate a total volume of gravel extracted since 1949 of ~ 1,100,050 m^. This excludes both undocumented extraction activities and tidal dredging, and represents an annual extraction rate of ~ 25,600 mVyear. This may cleariy under-estimate a true indication ofthe volumes removed, as records pertaining to Bellingen Shire Council are extremely poor. Indeed, a worker for Bellingen Shire Council who worked in gravel extraction operations between 1949 - 1986 stated that average daily extraction rates were approximately 200 m^/day (Lee, 1989). This is not 251 5. Post-European channel dynamics in the Bellinger valley dissimilar to values quoted for the main extraction operation located at the downstream end of Gordonville Reach (see extraction Site B8, Appendix 5.4). If this were the case, tiien estimated volumes removed by Bellingen Shire Council may be largely underestimated, depending on the number of days per year tiiat material was exti-acted. The estimated annual exttaction rate is not that dissimilar to the adjacent Nambucca valley, which was experiencing an estimated extraction rate of 20,000 - 30,000 m^/year. This value, however, once compared to Bureau of Mineral Resources documents was found to be a third ofthe volume of material that was being removed between 1987 - 1993 (Doyle, 2003). Therefore the figures in Appendix 5.4 can really only provide a conservative estimate of gravel and sand that has been extracted from the Bellinger River. The recharge rates and the geomorphic implications of the legacy of gravel extraction will be addressed in the following section and later in Chapter 6. Channel Incision Resource Planning (1989), Water Resources (1994) along with govemment departmental files present both anecdotal and morphological evidence of bed-level incision on the lower Bellinger River. The evidence is coupled with complaints with regard to gravel extraction operations occurring below the low-flow level. However, quantifying the extent of bed- level incision is problematic due to limited physical evidence and a lack of historical survey data. The sources of data presented in this section include photographic evidence of active knickpoints, palaeo bed levels derived from stratigraphic data presented in Chapter 3, prior riffle bed levels derived from the 1943 and 1973 photogrammetrically derived cross sections, and other anecdotal evidence. In general bed-level incision has been limited throughout the Bellinger River due to the presence of outcropping bedrock in the bed of the channel, limiting the migration of knickpoints. Data for the upper valley setting for Orama Reach are non-existent and the structural control of both riffles and pools, along with the very littie, if any post-European planform adjustment, suggest a limited potential for bed-level incision. Appendix 5.4 presented the locations of past gravel extraction operations in the lower valley setting from Thora downstream. As mentioned in Chapter 2, Thora represents a downstream shift in valley morphology, witii the valley becoming wider and straighter associated with less resistant lithology and or variations in the dip and strike of the Bellingen Slate. Outcropping bedrock becomes less frequent downstream of Thora resulting in an increased susceptibility to bed-level incision. This bed-level incision has been associated 252 5. Post-European channel dynamics in the Bellinger valley with the 7% decrease in channel length and the direct effects of gravel extraction, with the most pronounced incision having occurred in the downstream section of Gordonville Reach. Wills Reach The evidence for bed-level incision in Wills Reach is derived from anecdotal evidence of prior low-flow pool levels associated with outcropping bedrock or remnant riparian species (L.Rose, pers.comm., 2002) and the photogrammetrically derived cross-sections. The presence of the rarely-inundated high floodplain within this reach, along with the relatively small increase in cross-sectional area of 5% between 1943 and 2000 (documented in the previous section); would suggest that channel expansion has been coupled with some bed-level incision, probably prior to 1943. Figure 5.14a presents the prior bed-level data for Wills Reach and highlights that 75% of the 1943 riffle levels along with prior low-flow pool levels are all well above the current thalweg elevations. A reach- averaged estimate yields a mean of 0.4 m above the thalweg profile, while the 1973 data suggest that there have been negligible changes (0.05m) in thalweg elevation between 1973 and 2000. While the prior pool levels may represent an estimate of bed-level lowering, any coeval channel expansion at pool locations may actually have produced the lowered pool levels rather than absolute changes to bed level. The reduction in channel length, however, of approximately 315 m within the reach, in addition to the reach-averaged estimate of riffle lowering would suggest that bed levels throughout the reach have dropped by up to 0.4 m. It is unknown whether this incision was a function of knickpoints migrating either up or downstream. The net result of the mid-late 20* century incision, coupled with the documented channel expansion has been the abandonment of the majority of the high floodplain within Wills Reach to form what is now essentially a terrace. Gordonville Reach While Gordonville Reach has undergone documented channel expansion it has only shortened its length by 75 m. It is, however, immediately upstream of a segment of channel that has experienced a series of cutoffs, shortening its length by 650 m (see Sites 5 - 7 in Figure 5.2). This straightened section of the lower Bellinger River (i.e. immediately downstream of Gordonville Reach) has little to no outcropping bedrock on the floor of the channel making it the most susceptible to bed-level incision. In addition, this steepened segment has experienced extensive aggregate extraction between 1970 and 1990. 253 5. Post-European chaimel dynamics in the Bellinger valley a) Wills Reach 0 1000 2000 3000 4000 5000 Distance downstream (m) b) Gordonville Reach bar harvesting In-channel extraction I 1 GVT7 0 1000 2000 3000 4000 5000 Distance downstream (m) — Current thalweg level m 1943 rifflebe d level H bedrock outcropping on floor of channel A 1973 rifflebe d level 0 bedrock outcropping on outer channel margin X Prior pool level Current pool level Figure 5.14 Prior bed level indicators for Wills and Gordonville Reach based on photogrammetrically derived cross-sections with levels taken at riffle locations to minimise error, a) Wills Reach with 1943, 1973 bed level indicators with 2000 longitudinal profile and two anecdotal indicators of prior pool levels (L. Rose, pers.comm. 20(X)). b) 1943, 1973 bed level indicators with 2000 longitudinal profile for Gordonville Reach. Note the pronounced incision at the downstream end of the reach. 254 5. Post-European channel dynamics in the Bellinger valley Figure 5.14b presents the longitudinal profile of Gordonville Reach with prior bed levels. The downstream end of this reach has been repeatedly extracted to a depth of 4 m below low-flow level since 1970, extracting bars, benches and any in-channel sediment within pools (Resource Planning, 1989; CSR, pers.comm., 2001). This has reportedly reduced bar recharge rates immediately downstream and induced knickpoints that has progressively moved upstream into Gordonville Reach. To date, the quantification of bar recharge rates and the extent of bed-level incision has been poor, although survey data presented by Water Resources (1994) have been used to suggest 3.6 m of degradation between 1958 and 1989, followed by 1.2 m of aggradation between 1989 and 1992. This however, is a gross over-estimate as will be shown in the following section. The prior bed-level data presented in Figure 5.14b incorporates the bed elevation from the stratigraphic exposure in Chapter 3 (refer to Figure 3.40 for location of GVT7) and the photogrammetrically derived 1943 and 1973 riffle thalweg elevations. The prior bed-level data in Gordonville Reach suggest a reach-averaged estimate of 0.25 m of incision between 1943 and 1973, with a further 0.48 m of incision since. While the reach-averaged estimate of bed-level lowering in Gordonville Reach is 0.75 m, some prior bed-level indicators would suggest an estimate of 3 - 4 m of incision, which is not dissimilar to that presented by Resource Planning (1989) and Water Resources (1994) (Figure 5.14b). This estimate may not be representative of the degree of incision due to the fact that the stiatigraphic section assessed in GVT7 may simply represent a palaeo-riffle location in contrast to the current pool (Figure 5.14b). In addition, this pool in the lower section of Gordonville Reach was actively dredged by CSR Readymix for 20 years. Resource Planning (1989) identified that the estimated 4 m of bed-level lowering was a function of thalweg migration and the formation of a scour pool at the bedrock bend at Boggy Creek (see location (ii) Figure 5.12). It is clear that the eariy estimates by Resource Planning (1989) and Water Resources (1994) of 3.6 m of degradation at Boggy Creek are inaccurate, and like the palaeo bed-level indicator at GVT7 may simply represent localised pool formation rather than the extent of absolute bed-level lowering. Additional photographic evidence presented in Appendix 5.5 cleariy shows a migrating knickpoint in 1989 at the extracted chute channel immediately downstream of Gordonville Reach. This is supported by Figure 5.14b, which demonstrates that the greatest extent of degradation has occurred in the lower half of Gordonville Reach, which was subjected to extraction rates of 50,000 m^/year between 1970 - 1988. Thus, there is strong evidence for upstream migrating knickpoints in Gordonville Reach, producing a reach-averaged estimate of 0.7 m of incision since the mid 20* century. Like Wills Reach, many of the 255 5. Post-European channel dynamics in the Bellinger valley high floodplains in Gordonville Reach are rarely inundated (> 25-year recurrence interval). The documented post-European reduction in channel length immediately downstteam of Gordonville Reach, along with more recent aggregate exttaction practices have cleariy resulted in bed-level adjustments and the partial abandonment of the high floodplain. The small extent of outcropping bedrock on the floor of the channel in Gordonville Reach has resulted in more extensive incision than in Wills Reach. The presence however, of outcropping bedrock on channel margins, and/or at the base of sttuctural pools, along witii bed material that armours, has prevented the wholesale migration of knickpoints within the system and potentially resulted in localised zones of bed-level lowering. The documented level of incision coupled with the planform changes and late 20* century aggregate exttaction practices in this lower valley setting have resulted in continued increases in channel capacity, while also displaying the most dynamic channel behaviour of the three reaches examined. 5.4.7 Nature of bar mobility and riparian zone changes: 1943 - 2000 The Bellinger catchment like most, if not all, coastal catchments of NSW has undergone significant changes in land use and vegetation cover since the mid 20* century. Lyall & Macoun (1999) estimated an approximate 50% increase in regrowth forest between 1956 and 1991 in the adjacent Nambucca catchment. A decrease in the area of ringbarked frees, along with an increase in regenerating wet and dry sclerophyll forest can also be documented for the Bellinger catchment, supporting both historical photographic records and anecdotal evidence. Aerial photographs for the Bellinger catchment highlight that the majority of forest regrowth has occurred on hillslopes adjacent to the valley floor of both the trunk stream and tributaries. This shift in vegetation distribution has occurred during a period characterised by less intensive agricultural activities associated with a decline in dairy fanning and a shift to smaller scale hobby and residential rural properties. As outiined in Section 5.2.3 a 300 m buffer (150 m either side of the channel centre-line) was designated in which all shrubs and trees were mapped for each of the four time intervals. The mapping presented within this section does not distinguish between vegetation types, but simply incorporates all trees and shrubs (i.e. woody vegetation) and is referred to as riparian vegetation from herein. In addition, the aerial extent of bare gravel bars has also been quantified for each of the four time periods and is referred to as the bar index. The spatial extent of riparian vegetation (i.e. trees and shrubs) is expressed as an indice per unit length of channel. Thus, total vegetation coverage for one metre of channel would be 300 m^m"', minus the width ofthe low-flow channel (25 - 50 m). 256 5. Post-European channel dynamics in the Bellinger valley Table 5.7 presents the vegetation and bar indices for the three reaches from 1943 - 2000. Orama Reach in the upper valley setting shows the highest vegetation index of the three reaches of 169 m^m"\ Hence, the upper valley setting has 60% riparian cover, due largely to tiie continuously vegetated bedrock banks. The two lower reaches. Wills Reach and Gordonville Reach have a riparian zone index of 123 m^m"^ and 86 m^m"' (25 and 40% cover respectively). Most of the vegetation along the riparian zone within the Bellinger catchment is fragmentary, reflecting land use history and the location of bedrock. Where the channel impinges against outcropping bedrock, riparian vegetation tends to be more continuous than the alluvial banks, due to the unsuitability for grazing. Table 5.7 presents the mean area for 'patches' of riparian vegetation within the three reaches. This is simply a measure of the average size of a patch (i.e. polygon) of mapped riparian vegetation. Both Orama and Wills Reach currentiy have an average 'patch' size > 2000 m^, in conttast to Gordonville Reach with an average of < 1500m^. This index of riparian vegetation continuity has both geomorphic and ecological implications, with Table 5.7 Changes to riparian vegetation and bar areas on the Bellinger River 1943-2000 ~ River Mean Rip. Veg % of No. of Mean bar Bar length patch Index buffer bars size Index (m) area(m ) m'm' (mO mm ORAMA REACH 1943 4500 2116 102 34 46 1113 11 1956 6400 2848 110 37 73 1951 21 1973 6400 3608 139 47 80 655 7 2000 6400 2229 169 56 15 218 0.5 WILLS REACH 1943 6000 565 36 12 55 2167 19 1956 6000 958 34 12 46 5587 42 1973 6000 1489 42 14 51 3902 33 2000 6000 2668 123 39 33 665 5 GORDONVILLE REACH 1943 4300 400 36 11 72 917 16 1956 4300 1716 41 13 39 4373 41 1973 4300 1242 52 16 40 3905 36 2000 4300 1487 86 26 48 980 11 Mean patch area refers to the mean dimension of a polygon of riparian vegetation. % buffer is calculated as total aerial extent of mapped trees and shrubs divided by total buffer area Riparian veg. index is the total aerial extent of riparian vegetadon divided by length of reach assessed - per unit length of channel- m-m' Bar index is the total aerial extent of unvegetated bar surfaces divided by length of reach assessed - per unit length of channel - m-m"' 257 5. Post-European channel dynamics in the Bellinger valley smaller areas of vegetation being more susceptible to edge effects and the subsequent invasion by exotics. In addition, the discontinuous vegetation pattern seen within the reaches results in some sections of alluvium effectively reinforced by large areas of bankside vegetation, while others have little to no riparian cover, other than agricultural pasture. Despite Orama Reach having the largest extent of riparian vegetation, the greatest percentage increase from 1943 - 2000 occurred in the lower valley setting of Wills and Gordonville Reach (240 and 140% increase respectively). All three reaches display similar temporal ttends with tiie 1943 interval having the lowest riparian vegetation coverage in all but Wills Reach, where it was slightiy higher than the 1956. Table 5.8 presents the percentage changes for both riparian vegetation and bar index data. Percentage changes between 1943 and 1956 ranged from 7 - 15% increases in Orama and Gordonville Reach, while Wills Reach experienced a net loss of riparian vegetation of 5%. The next time interval of 1956 - 1973, which represents the eariy to latter stages of the last FDR, exhibits an additional 21 - 27% increase in riparian vegetation for the three reaches (Table 5.8). The last time interval of 1973 - 2000 was characterised by the most pronounced percentage increase in riparian vegetation in any of the time intervals, in all but Orama Reach. These increases, of up 193%, have occurred in the latter stages of the previous FDR and the more recent period of low flood activity (i.e. the current DDR). Figures 5.15 - 5.17 present the riparian vegetation changes for the three study reaches in the four time intervals. Orama Reach has undergone little if any change in the extent of vegetation on the concave banks, with each of the time intervals displaying continuous tteecover on the bedrock valley margins (see Figure 5.15). The greatest change in this upper valley setting has been the increase in the aerial extent of patches of vegetation on the alluvial channel margin and the progressive revegetation of inset alluvial units such as bars and benches from 1956 onwards. Table 5.8 Percentage changes to riparian vegetation and 'active bar' surfaces 1943 - 1956 1956 - 1973 1973 - 2000 Rip. Veg. Bar Index Rip. Veg. Bar Index Rip. Veg. Bar Index (%) change (%) change (%) change (%) change (%) change (%) change ORAMA + 7 + 82 + 26 -64 + 21 -93 mi±s -5 + 120 + 21 -22 + 193 -86 GORDONVUXE + 15 + 160 + 27 - 11 + 63 -70 258 5. Post-European channel dynamics in the Bellinger valley 1943 1956 1973 2000 2km npanan vegetation dominated by Casuarina cunninghamiana and exotics such as Ligustrum lucidum < 45 yrs Figure 5.15 Riparian vegetation changes in Orama Reach from 1943 to 2000. All tteesan d shrubs within a 300 m buffer have been mapped from the rectified 1943, 1956, 1973 and 2000 aerial photographs. Flow is from left to right. 259 5. Post-European channel dynamics in the Bellinger valley riparian vegetation 0 2 km riparian vegetation dominated by Casuarina cunninghamiana < 30 yrs Figure 5.16 Riparian vegetation changes in Wills Reach from 1943 to 2000. All ttees and shrubs within a 300 m buffer have been mapped from the rectified 1943,1956,1973 and 2000 aerial photographs. Flow is from left to right. 260 5. Post-European channel dynamics in the Bellinger valley 261 5. Post-European channel dynamics in the Bellinger valley Table 5.8 demonstrates that Wills Reach has experienced the largest increase in riparian vegetation coverage since 1973. This has been through the colonisation of benches, the re vegetation of the bedrock valley margin at the downstream end of the reach and the development of mature orchard trees within the 300 m buffer. Areas that have undergone channel cutoffs, chute-channel stripping or localised channel expansion have become colonised by C. cunninghamiana since 1973 (Figure 5.16). Two inset units in the middle and upper part of the reach have also become stabilised by C. cunninghamiana, while the rest of the reach has not undergone a marked change in the extent of coverage of riparian vegetation since 1943. Figure 5.17 presents the changes to riparian zone coverage for Gordonville Reach. This reach, unlike the two other reaches, experienced a 15% increase in coverage between 1943 and 1956 and another 27% increase between 1956 and 1973, witii the largest change occurring between 1973 and 2000. The area of greatest change has been hillslopes adjacent to the channel in the upper part of the reach, along with the recolonisation of riparian vegetation on inset sedimentary units and mid-channel bars (Figure 5.17). This figure shows that the 1943 time period was characterised by areas of in-channel vegetation covering bars, riffles and benches, which were effectively removed/stripped by 1956 (i.e. in the floods between 1950 - 1955). Since 1973, areas that have undergone both localised expansion and aggregate exttaction have become progressively re-vegetated, despite the continued increases in channel capacity. The broad trend for Gordonville Reach, like the two other reaches, is one of an increase in both the extent of riparian vegetation since 1943, and an increase in the average size of any given patch of riparian vegetation. The bar index data represent areas of unvegetated gravel bars visible in the rectified aerial photographs. Thus, this index may suggest a level of bar mobility, with total vegetation coverage most likely representing reduced rates of bar reworking and bare bars representing increased rates of bar reworking. For ease of terminology, this bar index is inferred to represent 'active' bars. This is obviously an over-simplification and it is acknowledged that sediment on vegetated gravel bars are still susceptible to reworking, albeit at different thresholds. Despite this limitation, this index provides insight into how much of the in-channel sediment in the three study reaches was stabilised by vegetation in the four time periods. Table 5.8 presents the percentage change in unvegetated bar areas in the four time periods, while Figure 5.18 shows the changes in the areal extent of unvegetated bar surfaces and the associated changes in riparian vegetation. What is evident from Table 5.8 is that the 262 5. Post-European channel dynamics in the Bellinger valley J2DR FDR DDR T3 P 20 • active bar area H ft) fj Ciq (m^m-^) 3 a 10 . SB ^^ o' 3 a) Orama^eafcte 3" a. a> X 40 30 o active bar area (m^m-^) 20 10 0 b) Wills Reach CO 40 5' 3 30 1 ^ • active bar area -s CTQ 3 a (m^ m"^) 20 10 3' O. c) Gordonville Reach a> 0 0 X 1940 1950 1960 1970 1980 1990 2000 Year Figure 5.18 Vegetation and 'active' bar surface areas in the three reaches based on the 1943, 1956, 1973 and 2000 rectified aerial photographs, where riparian vegetation is calculated as area of vegetation within the designated 300 m riparian buffer per unit metre of channel. Active bar area for a) Orama Reach, b) Wills Reach and c) Gordonville Reach has also been calculated from bare 'active' gravel bars per unit length of channel. Curves are inferential only. 263 5. Post-European channel dynamics in the Bellinger valley time interval between 1943 and 1956 was characterised by large increases in titie extent of active bar surfaces as a response to the floods of the early 1950s, with the greatest change (160% increase) occurring in the most downstteam reach. Between 1956 and 1973 all three reaches exhibited a reduction in the area of active bar surfaces, with the greatest reduction (64% decrease) occurring in the most upstream reach. The relatively small reduction of 11% in Gordonville Reach indicates that this most downstream reach retained a greater proportion of unvegetated bars throughout the last FDR, in contrast to Orama and Wills Reach, which reduced their active-bar surfaces by 64 and 22%, respectively. This shows that while the floods in the eariy 1950s mobilised many bar surfaces in these two reaches, the remaining part ofthe FDR saw a progressive reduction in bar activity. Figure 5.18 and Table 5.8 also show that the largest decrease in active-bar surface areas was between 1973 and 2000, with the greatest decrease occurring in the most upstream reach. All reaches displayed a reduction in the extent of active-bar surfaces between 1973 and 2000 (93-70% reduction), with bar indices returning to lower than 1943 levels, with Gordonville Reach (i.e. the most downstream reach) experiencing the smallest reduction (see Table 5.8). In summary, all three reaches exhibited large increases in the areal extent of unvegetated bars between 1943 and 1956, associated with the large floods of the early 1950s. Since then, bar surfaces have become progressively revegetated with Casuarina cunninghamiana, with the greatest rates of change in the areal extent of active-bar surfaces occurring in the upper valley setting. This suggests that all three reaches experienced reworking of bar surfaces in the eariy part of the FDR, but with bars in the upper valley setting experiencing rapid rates of colonisation during the remaining FDR. In contrast, the most downstream reach, experienced the greatest increase in the areal extent of active-bar surfaces following the 1950 floods and has since exhibited the slowest rates of bar colonisation. 5.5 SUMMARY OF POST-EUROPEAN CHANNEL CHANGES: EVIDENCE FOR CYCLIC BEHAVIOUR? As outiined in Chapter 1, the conceptual model presented by Warner (1987, 1992, 1993, 1994a, 1995) and Erskine and Warner (1988, 1998, 1999) invokes a process of cyclical channel response to altemating periods of flood and drought activity, termed FDRs and DDRs. In brief, the predicted morphological response to the FDR-DDR model infers increases in bankfull width and decreases in depth in FDRs and the reverse in DDRs. In addition, bankfull capacity (i.e. cross-sectional area) is suggested to increase by as much as two to four times between a DDR and an FDR (Wamer, 1994a). This chapter has set 264 5. Post-European channel dynamics in the Bellinger valley out to examine the nature of post-European channel processes in the three study reaches, to quantify tiie size of the pre-European Bellinger River and test the applicability of the above climate-based hypotheses. Sections 5.4.1 - 5.4.3 presented an analysis of planform change along the Bellinger River, demonstrating that the lower valley reaches have experienced the greatest extent of planform change, witii the bedrock-dominated Orama Reach experiencing little to no planform change. The evidence for the late 19"^ and the eariy 20"^ century is sparse, but historical evidence would suggest that both periods experienced planform changes, witii cutoffs occurring in the 1890s and during the 1921 floods. The next period of above- average flood activity (i.e. the 1948 - 1977 FDR) brought about further channel shortening. The net result has been a 7% reduction in channel length between settlement and the late 1970s; all forming due to chute-channel cutoffs. Thus, planform changes have not been restricted to periods of either above or below-average flood activity. In addition, the photogrammetry data have shown that the documented chute-channel cutoffs that occurred in tiie 1948 - 1977 FDR were pre-conditioned by floods in the 1900 - 1947 DDR (i.e. partially cut). Hence, the cutoffs were a progressive adjustment to changing catchment boundary conditions, with the majority of cutoffs occurring in a time when riparian vegetation was at an historical low. The dimensions of the abandoned palaeochannels would suggest that pre-European channel dimensions were at least half the width and a third in cross-sectional area. This obviously has implications for floodplain inundation, with the estimated bankfull capacity of the narrow palaeochannels being close to the current mean annual flood. This obviously contrasts to the current channel capacity in many reaches which can contain > 50-year flood. The 1943 photogrammetry data provide the first quantified estimate of channel geometry in the previous DDR for the two reaches in the lower valley setting. The data suggest that Wills Reach had already undergone channel enlargement, in contrast to Gordonville Reach, which still exhibited some channel sections with narrow channel dimensions. By 1943 both pool and riffle locations in Wills Reach could already contain bankfull frequencies of > 25 years. Indeed the presence of erosional scarps presented in Figure 5.7 and 5.9 would suggest Wills Reach had already experienced some channel expansion by 1943 (most likely by the 1921 floods). Thus, the eariy premise by Warner (1987) and Erskine and Warner (1988, 1998, 1999) that DDRs are characterised by channel contraction and increases in depth cannot be substantiated for the Bellinger River. An altemative interpretation would suggest that by 1943 some, but not all, of the lower Bellinger River had both experienced channel shortening and channel expansion (i.e. such 265 5. Post-European channel dynamics in the Bellinger valley as Wills Reach). The timing of this is unknown, but is most likely to have occurred in the 1921 floods or in the late 1890s. The lack of riparian vegetation and the partial cutting of chute-channels in the 1900- 1947 DDR, made the other sections of the channel that had not already expanded by 1943, inherentiy more susceptible to the floods of the eariy 1950s. All three reaches examined underwent statistically significant increases in bankfull width between 1943 and 1973 of between 7 - 12%, with the greatest change occurring in the lower valley (Wills and Gordonville Reaches). In addition, this time interval saw an increase in cross-sectional area at both pools and riffles of between 4 - 15%. Again, tiie geomorphic responses to this FDR varied between reaches with some locations (i.e. Wills Reach) demonstrating preferential erosion at pools, in contrast to riffles. Both lower valley reaches displayed a statistically significant reduction in variance of cross-sectional area, highlighting that the floods ofthe 1948 - 1977 FDR elfectively made the channel in both reaches more uniformly large. In addition to the reach-averaged increases in channel dimensions, this period of above- average flood activity also produced areas of major, but localised channel expansion, a series of cutoffs and reworked a large number of bars and inset features. The exact timing of the increases in channel dimensions within this FDR is unknown, but constrained by the 1943 and 1973 data sets. The increases are most likely a function of the 1950 flood (the 3"^ largest on record) and a series of moderate-large events between 1950 and 1955. This eariy part of the 1948 - 1977 FDR was characterised by large increases in the presence of 'active' bar surfaces, with all three reaches demonstrating large increases between 1943 and 1956. The middle to latter part of the 1948 - 1977 FDR saw a reduction in the areal extent of 'active' bar surfaces but the maintenance of large channel dimensions. The evidence for a 1978 - 2000 DDR channel response (i.e. a reduction in width and a subsequent increase in depth) is poor. In general, the geomorphic response to the DDR, like other time periods varies between reaches. Reductions in reach-averaged bankfull width of between 6 - 15 % have been documented for the lower reaches, however, these reductions are skewed by large decreases in width in a few locations, as a result of proximal levee and floodplain deposition. The 1978 - 2000 DDR has also been characterised by large increases in riparian vegetation and a progressive reduction in the area of active bar surfaces. Thus, this last DDR has been characterised by the revegetation of inset features, but the maintenance of a large capacity channel in all reaches. In addition, 266 5. Post-European channel dynamics in the Bellinger valley the progressive revegetation of bars and benches has also been accompanied by bed-level incision of 0.5 m in the two lower reaches, induced by aggregate extraction practices. Post-European in-channel processes on the Bellinger River reflect a re-configuration of riffle-pool units. As suggested by Water Resources (1994), riffle spacing has increased witii some areas experiencing a loss of a riffle through localised channel expansion forming longer, partly infilled pools. The loss of in-channel vegetation between 1943 and 1956, and indeed since settiement, may represent the loss of a vegetative control on the presence of particular riffles. The revegetation of these reworked inset units and bars in tiie current DDR, however, has not been accompanied with a decrease in channel capacity. The current enlarged channel dimensions of all three reaches is a reflection of post-European channel enlargement processes occurring pre-1943 and continuing between 1943 and tiie present. This has been compounded by post-1950 aggregate extraction activities that have substantially reduced sediment recharge rates. The current DDR channel capacity, despite being more vegetated than the 1948 - 1977 FDR is not half the FDR bankfull capacity as suggested by Warner (1994a). The chapter actually presents convincing evidence that channel capacity is larger now than it was in the 1948 - 1977 FDR. This study shows that the evidence for cyclical channel changes in channel capacity in the Bellinger River is poor, suggesting that the FDR-DDR is a simplistic model of sediment storage and sediment evacuation. This chapter has highlighted that predicted morphological responses to any given discharge regime are by no means uniform. Indeed, the data suggest that there is both inter- and intra- reach variability in the nature of post- European channel response. This variability of post-European response between geomorphic units has been previously overiooked within the FDR-DDR model. The fact that reach-averaged channel capacity has continued to increase throughout the 20"" century suggests a more complex form of post-settiement channel adjustment than originally suggested. The data within this chapter suggest that these channels have undergone extensive post-European modification that greatiy overshadows any significant climatic change. They display non-equilibrium conditions and are still adjusting their capacity and channel geometry to changes in sediment supply and bank resistance. An altemative hypothesis is therefore required to explain the current channel form and the post- settlement behaviour of this and other coastal river systems of NSW. The following chapter will examine stage-discharge relationships of the abandoned floodplains and inset sedimentary units, the stratigraphic characteristics of these benches and changes in the sediment storage volumes between the inferred climatic regimes. 267