SOILSOILSOILSOIL2199-398XCopernicus GmbHGöttingen, Germany10.5194/soil-1-641-2015Assessing the performance of a plastic optical fibre turbidity sensor for
measuring post-fire erosion from plot to catchment scaleKeizerJ. J.jjkeizer@ua.pthttps://orcid.org/0000-0003-4833-0415MartinsM. A. S.PratsS. A.SantosL. F.VieiraD. C. S.https://orcid.org/0000-0003-2213-3798NogueiraR.BilroL.Earth surface processes team, Centre for Environmental and Marine Studies
(CESAM), Dept. Environment and Planning, University of Aveiro, Campus
Universitário de Santiago, 3810-193 Aveiro, PortugalInstituto de Telecomunicações, Aveiro (IT-Aveiro), Campus Universitário de Santiago, 3810-193 Aveiro, PortugalJ. J. Keizer (jjkeizer@ua.pt)24September20151264165013April201511May20151September2015This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://soil.copernicus.org/articles/1/641/2015/soil-1-641-2015.htmlThe full text article is available as a PDF file from https://soil.copernicus.org/articles/1/641/2015/soil-1-641-2015.pdf
This study is the first comprehensive testing of a novel plastic optical
fibre turbidity sensor with runoff samples collected in the field and, more
specifically, with a total of 158 streamflow samples and 925 overland flow
samples from a recently burnt forest area in north-central Portugal,
collected mainly during the first year after the wildfire, as well as with 56
overland flow samples from a nearby long-unburnt study site. Sediment
concentrations differed less between overland flow and streamflow samples
than between study sites and, at one study site, between plots with and
without effective erosion mitigation treatments. Maximum concentrations
ranged from 0.91 to 8.19 g L-1 for the micro-plot overland flow
samples from the six burnt sites, from 1.74 to 8.99 g L-1 for the
slope-scale overland flow samples from these same sites, and amounted to
4.55 g L-1 for the streamflow samples. Power functions provided
(reasonably) good fits to the – expected – relationships of increasing
normalized light loss with increasing sediment concentrations for the
different sample types from individual study sites. The corresponding
adjusted R2 values ranged from 0.64 to 0.81 in the case of the
micro-plot samples from the six burnt sites, from 0.72 to 0.89 in the case of
the slope-scale samples from these same sites, and was 0.85 in the case of
the streamflow samples. While the overall performance of the sensor was thus
rather satisfactory, the results pointed to the need for scale of
site-specific calibrations to maximize the reliability of the predictions of
sediment concentration by the POF (plastic optical
fibre) sensor. This especially applied to the
cases in which sediment concentrations were comparatively low, for example
following mulching with forest residues.
Introduction
Wildfires are now widely recognized as a potential driver of conspicuous
changes in geo-morphological and hydrological processes, through their
direct effects on vegetation, litter layer and topsoil (Shakesby, 2011;
Moody et al., 2013). Studies across the globe have shown strong and
sometimes extreme responses in runoff and erosion in recently burnt areas,
especially during the earlier stages of the so-called “window of
disturbance”
(e.g. Cerdà, 1998; Lane et al., 2006; Robichaud et al., 2007).
Nonetheless, important research gaps remain with respect to wildfire impacts
on runoff and especially soil erosion, in part due to the relatively limited
number of post-fire erosion studies as compared to erosion studies in
agricultural areas (Shakesby, 2011). The latter is well-illustrated by the
four studies that appear to have been carried out in the Mediterranean Basin
on sediment yields from recently burnt catchments (Lavabre and Martin, 1997;
Inbar et al., 1998; Mayor et al., 2007; Keizer et al., 2015). Clearly, more
studies have been published on post-fire erosion at the plot-to-slope scale
in the Mediterranean Basin (e.g. Thomas et al., 1999; Fernández et al.,
2007; Prats et al., 2014). However, they have typically addressed soil
losses with a relatively coarse temporal resolution, i.e. multiple runoff
events, which is hampering further insight in underlying sediment transport
processes.
Over the past decade, various authors (Ruhl et al., 2001; Campbell et al., 2005; Postolache et al.,
2007) have obtained promising results in measuring turbidity
of aqueous solutions with POF sensors. Nonetheless, in their review
study, Omar and MatJafri (2009) identified the need for more extensive
testing, in particular also with respect to the dependence on particle size.
Therefore, this study aimed at further testing the performance of the POF sensor
developed by Bilro et al. (2010), which had provided promising results for
contrasting suspended materials, including ashes from recently burnt areas
(Bilro et al., 2011). More specifically, this study tried to (i) assess the
performance of this sensor for measuring the sediment concentration in post-fire
runoff generated during the initial stages of the window of disturbance,
when erosion rates are expectedly highest; (ii) evaluate if sensor
performance differed for streamflow and for overland flow from erosion plots
with contrasting runoff areas (micro-plots vs. slope-scale plots) and, thus,
potentially different erosion processes (inter-rill erosion vs. rill/gully
erosion); and (iii) determine if sensor performance depended on land cover,
parent material and site-specific conditions. This study was envisaged as an
important step towards the development of a commercial version of the sensor
designed by Bilro et al. (2010).
Study area and sites
This study was carried out near the hamlet of Ermida in the Sever do Vouga
municipality of north-central Portugal (Fig. 1). The area was burnt by a
wildfire that took place between 26 and 28 July 2010 and that affected
some 300 ha (DUDF, 2011). By the time of the fire, the area was mainly
covered by plantations of eucalypt (Eucalyptus globulus Labill.) but
did include some plantations of maritime pine (Pinus pinaster Ait.).
The severity of the wildfire (sensu Keely, 2009) was assessed in the field
using as indicators ash colour as well as degree of tree crown scorching and
of litter layer consumption, following Shakesby and Doerr (2006) and prior
studies in the region such as Malvar et al. (2011, 2013). At all six study
sites selected within the burnt area (Fig. 1), fire severity was classified
as moderate. During the winter of 2010/11, the central part of the study area
was bench terraced using a bulldozer, affecting three of the study sites
(the terraces are clearly visible in Fig. 1).
Location of the study area, the experimental catchment and the seven
study sites (A: burnt eucalypt plantation on granite; B, D, E and
S: burnt eucalypt plantations on schist; C: burnt pine plantation
on schist; F: long-unburnt eucalypt plantation on schist).
The climate of the study area can be classified as humid meso-thermal (Csb,
according to the Köppen classification), with moderately dry but extended
summers (DRA-Centro, 1998). The parent material in the study area mainly
consisted of pre-Ordovician schists but included Hercynian granites at some
locations, as is typical for the Hesperic Massif (Ferreira, 1978). The soils
were mapped, at a scale of 1 : 1 000 000, as predominantly Humic
Cambisols (Cardoso et al., 1971, 1973). However, field descriptions of soil
profiles at the various study sites suggested a prevalence of Leptosols (WRB,
2006) (see Machado et al., 2015; Martins et al., 2013). Soil texture of the
A horizon was also determined in the field and was slightly coarser for the
soils on granite (sandy loam) than for the soils on schist (sandy clay loam).
The topsoil was very rich in organic matter, amounting to 20–30 % at
0–2 cm depth (Machado et al., 2015) and 8–11 % at 0–5 cm depth
(Prats et al., 2014).
General information about the seven study sites as well as the
numbers of runoff samples from micro-plots and slope-scale plots analysed
from each site and the start and end dates of collecting these samples (in
ddmmyy).
Within the burnt area, a total of six study sites were selected to study
post-fire runoff and erosion (Fig. 1; Table 1). They consisted of four
eucalypt plantations on schist (sites B, D, E, and S), one eucalypt plantation on
granite (site A) and one pine plantation on schist (site C), basically
following the incidence of these land cover–parent material combinations in
the burnt area. In addition, a long-unburnt eucalypt plantation was selected
in the immediate vicinity of the burnt area (site F). Furthermore, one of the
catchments within the burnt area was selected for studying the hydrological and
erosion response at the catchment scale.
Materials and methodsExperimental set-up and collection of runoff samples
Five of the six study sites within the burnt area – i.e. except site S –
were divided in three adjacent strips running from the base to the top of the
slope (section) (Machado et al., 2015; Martins et al., 2013). In one of these
strips, either three bounded micro-plots (0.25–0.30 m2) were installed
at the slope's base (sites A, B and C, for being located within the
catchments and therefore to minimize disturbance) or two pairs of such
micro-plots were installed at the base and halfway up the slope (sites D and E).
In another strip, one (un-)bounded slope-scale plot with a width of
approximately 2 m and contributing areas exceeding 50 m2, depending on
slope length, was installed. Each slope-scale plot, however, comprised four
outlets that were connected to different runoff-collecting tanks. Site S
involved a more elaborate experimental design, as it had been selected to
assess the effectiveness of two treatments to reduce soil erosion, i.e.
mulching with forest residues and application of a dry granular anionic
polyacrylamide (PAM; Prats et al., 2014, 2015). Polyacrylamides have been
found to markedly reduce soil losses from agricultural fields and road
embankments (Ben-Hur, 2006). Four triplets of the above-mentioned micro-plots
were installed from the base to the top of slope S to assess the
effectiveness of both treatments. Furthermore, two bounded slope-scale runoff
plots of 4 m width by 20–25 m length were installed to assess the
effectiveness of mulching with forest residues. The unburnt site, on the
other hand, involved a simpler experimental design as it was relatively
narrow and could only be divided in two strips. Therefore, it was only
instrumented with an unbounded slope-scale plot as described above.
The runoff from the micro-plot and the individual outlets of the slope-scale
plots was collected in tanks of 30 and 80–500 L, respectively. Runoff
volume in the tanks was measured and runoff samples were collected in 1.5 L
bottles, following intensive stirring of the water in the tanks. This was
done at 1–2-weekly intervals, depending on rainfall, starting at the end
of August 2010 when the site instrumentation had been completed.
The outlet of the experimental catchment was instrumented with a
hydrological station comprising two flumes, two water level recorders and an
automatic sampler that was triggered by a data logger based on the readings
of the two water level recorders.
Laboratory analysis of runoff samples
For this study, a total of 1139 runoff samples were analysed, of which 158
concerned streamflow, and 565 and 416 overland flow at the slope and
micro-plot scale, respectively. The distribution of the latter samples over
the different sites is given in Table 1. The samples were collected during
the first year after the wildfire, as further detailed in Table 1, except for
36 micro-plot samples that were collected at the S site between the end of
October 2011 and early January 2011.
Relationships of sediment concentration with normalized light loss
at the micro-plot scale for three treatments at study site S (left plot) and
the corresponding best-fitting power functions (right plot; see Table 2).
S_PAM: polyacrylamide; S_CTRL: untreated;
S_MLCH: mulching with forest residues.
The sediment concentration of these samples was determined in the laboratory
using the classic filtration method (APHA, 1998), employing filter paper with
a pore diameter of 12–14 µm and drying the filters in an oven at
105 ∘C for 24 h. Furthermore, the organic matter content of the
filtered sediments was determined using the loss-on-ignition method, placing
the filters in a muffle for 4 h at 550 ∘C.
For each of the runoff samples, the normalized loss of the transmitted light
– i.e. the ratio of the loss of light transmitted through a runoff sample
and transmitted through a reference sample of bi-distilled water – was
determined using the POF turbidity sensor presented
by Bilro et al. (2010) but with a slightly modified design of the sensor
head. To this end, the sensor head was first placed within a plastic
recipient with bi-distilled water to measure the reference signal and then
within a second recipient with the runoff sample to measure the light loss
due to the sediments that were being kept in suspension by means of a
magnetic agitator. The measurements were carried out during a period of
1 min, during which the POF sensor performed 120 readings. Following visual
inspection for and possible elimination of anomalous readings, the average
values of both sets of readings were then used to compute the normalized
transmitted light loss.
Sediment concentrations and corresponding organic matter (OM)
contents of the micro-plot scale overland flow samples at the six study
sites, and best-fitting power functions between sediment concentration (x;
in g L-1) with normalized light loss (y). Euc.: eucalypt; med.: median; iqr: inter-quartile range; 3rd q: third quartile;
max.: maximum; sign: statistically significant differences, at
α= 0.05, are indicated by different roman numbers in the case of the
treatments tested at the S site and by different letters in the case of the
other sites.
SiteForestParentTreatmentNSediment OM Best-fittingAdjustedcodetypematerialconcentration content power functionR2g L-1% med.iqr3rd qmax.signmed.iqrSeuc.schistnone1120.410.530.727.48ii6124y=0.1735x0.59830.64PAM780.641.071.378.19iii5222y=0.1962x0.52470.72mulching570.140.200.272.19i6721y=0.1268x0.65770.69Beuc.schistnone330.380.660.853.89cd5412y=0.2965x0.49380.77Deuc.schistnone470.210.160.271.06bc5514y=0.2054x0.80900.67Eeuc.schistnone420.731.762.006.06d6429y=0.2510x0.53720.76Aeuc.granitenone190.130.160.220.91ab5818y=0.2971x0.79120.81Cpineschistnone280.080.120.151.48a5412y=0.2808x0.67940.76Results and discussionMicro-plot scaleWithin-site differences related to erosion mitigation
treatments
In line with the findings of Prats et al. (2014) regarding specific soil
losses, the sample sets of the three treatments differed significantly in
sediment concentrations (Table 2). The median sediment concentration of the
untreated samples was 35 % lower than that of the PAM samples but almost
3 times higher than that of the mulching samples. The median organic
matter contents of all three sample sets were high (52–67 %), suggesting
that charred material was a major component of the sediments exported under
all three treatments. These median values closely matched the average values
in Prats et al. (2014), attesting to the representativeness of the sample
sets included in this study. Furthermore, they agreed well with the figures
in Malvar et al. (2011, 2013) for sediments eroded during the first 2 years
following a fire.
All three sample sets revealed a relationship of increasing normalized light
loss with increasing sediment concentration (Fig. 2), as was expected based
on the findings with an earlier prototype of the turbidity sensor (Bilro et
al., 2010, 2011). The power function provided reasonably good fits of these
relationships in all three instances, with adjusted R2 values ranging from
0.64 in the case of the untreated samples to 0.72 in the case of the PAM
samples (Table 2). Bilro et al. (2011) found clearly better fits
(R2> 0.95) for clay as well as ash particles but the authors used
dilution series of artificial samples rather than runoff samples collected in
the field.
The curves fitted to the untreated and the PAM samples were very similar, at
least within the range of measured sediment concentrations (i.e.
< 8.5 g L-1). It is possible that the somewhat divergent curve of the
mulching samples was due to a smaller range of measured sediment concentrations
(< 2.5 g L-1), and because the relationships between sediment
concentration and normalized light loss seemed to reveal more spread at
higher concentrations.
Between-site differences related to land cover and parent
material
Conspicuous and, in various instances, significant differences existed
between the study sites in the sediment concentration of the micro-plot
runoff samples (Table 2). Median sediment concentrations appeared to be
influenced by both parent material and forest type, as median values were
significantly lower for the pine plantation on schist (0.08 g L-1) and
for the eucalypt plantation on granite (0.13 g L-1) than for the
eucalypt plantations on schist (≥ 0.21 g L-1). Significant
differences, however, also existed among the eucalypt plantations on schist,
with the median sediment concentration of the D site (0.21 g L-1)
being 3.5 times lower than that of the E site (0.73 g L-1).
Between-site differences did not seem to be related to fire severity, at
least as suggested by the field indicators used in this study (see Sect. 2).
The difference in median sediment concentration between the pine plantation
on schist and the eucalypt plantation on granite agreed well with the
difference in the sites' median specific sediment losses reported by Martins
et al. (2013; 0.08 vs. 0.16 g m-2 mm-1 of runoff), once again
testifying to the representativeness of the sample sets included in this
study.
Best-fitting power functions of the relationships of post-fire
sediment concentration with normalized light loss at the micro-plot scale for
one pine plantation on schist and five eucalypt (euc.) plantations on schist
or granite (see Table 2).
The untreated sample sets from all six study sites showed the expected
increases in normalized light loss with increasing sediment concentrations.
Furthermore, these increases agreed well with power functions, with the
adjusted R2 values of the fitted curves ranging from 0.64 to 0.81 (Fig. 3;
Table 2). The fits were somewhat worse for sites D and S than for the
remaining four sites (adjusted R2 values: 0.64–0.67 vs. 0.76–0.81) but this
difference was apparently unrelated to parent material, forest type, sediment
concentrations or their organic matter contents. However, the shape of the
fitted curves did seem related to sediment concentrations. The curves were
steeper for sites A, C and D than for sites B, E and S, and the former three
sites had clearly lower median, third quartile and maximum sediment
concentrations than the latter three sites (e.g. in the case of maximum
concentrations, 0.91–1.48 vs. 3.89–7.48 g L-1). This contrast could
be due to differences in the size of the exported sediment particles, since
the sensor's light attenuation was shown to decrease with increasing particle
size (Bilro et al., 2011) and since the lower sediment concentrations at
sites A, C and D could be explained by overland flow with a lower transport
capacity, preferentially exporting smaller particles. Nonetheless, the
contrast could also be an artefact from the lower ranges of sediment
concentrations measured at sites A, C and D, as these ranges only covered the
initial, steeper parts of the fitted curves.
Relationships of post-fire sediment concentration with normalized
light loss at the slope scale for two treatments at study site S (symbols),
and best-fitting power functions at the slope as well as micro-plot scale
(lines) (see Tables 2 and 3). S_CTRL_slope/micro: untreated;
S_MLCH_slope/micro: mulching with forest residues.
Sediment concentrations and corresponding organic matter (OM)
contents of the slope-scale overland flow samples at the seven study sites
and best-fitting power functions between sediment concentration (x;
in g L-1) with normalized light loss (y). Euc.: eucalypt; med.: median; iqr: inter-quartile range; 3rd q: third quartile;
max.: maximum; sign: statistically significant differences, at
α= 0.05, are indicated by different roman numbers in the case of the
treatments tested at the S site and by different letters in the case of the
other sites.
SiteWildfireForestParentTreatmentNSediment OM Best-fittingAdjustedcodetypematerialconcentration content power functionR2g L-1% med.iqr3rd qmax.signmed.iqrSburnteuc.schistnone890.631.071.358.99ii6416y=0.2272x0.60950.85mulching850.190.410.511.74i4723y=0.2163x0.75100.71Bburnteuc.schistnone450.630.751.098.14bcd5815y=0.2576x0.56700.89Dburnteuc.schistnone900.290.841.005.86bc5514y=0.1704x0.67070.87Eburnteuc.schistnone701.212.262.868.62cd5312y=0.2768x0.52620.83Aburnteuc.granitenone730.691.121.396.59bcd3812y=0.2356x0.59440.86Cburntpineschistnone570.110.320.356.60a5319y=0.2281x0.60200.72Funburnteuc.schistnone560.050.140.150.74a7822y=0.1314x0.58320.52Slope scaleWithin-site differences related to erosion mitigation
treatment
Like the micro-plot samples, the slope-scale samples revealed clear and
significant differences in sediment concentration between the untreated and
mulching samples (Table 3). The median sediment concentration of the
untreated samples was more than 3 times higher than that of the mulching
samples (0.63 vs. 0.19 g L-1). These differences agreed well with the
stronger runoff response of the untreated plot compared to the mulched plot during the first
year after a fire (Prats et al., 2015; 58 vs. 30 mm).
The slope-scale samples tended to have higher median, third quartile and
maximum sediment concentrations than the micro-plot samples of the same
treatment (Table 3). The only exception was the maximum sediment
concentration of the mulched samples, being 20 % lower in the case of the
slope-scale samples than for the micro-plot samples (1.74 vs.
2.19 g L-1). This tendency in sediment concentrations was opposed to
that in overland flow, as Prats et al. (2015) reported roughly 15 times less
overland flow at the slope scale than at the micro-plot scale (409–956 vs. 30–58).
The fit of the power function was substantially better for the slope-scale
samples than for the micro-plot samples in the case of the untreated plot but
basically the same in the case of the mulched plot (Table 3; adjusted
R2 values: 0.85 vs. 0.64 and 0.71 vs. 0.69, respectively). In both cases,
light loss with increasing sediment concentration was larger for the
slope-scale samples than for the micro-plot samples (Fig. 4). Only in the
case of the mulching samples, however, was this due to a clearly higher
attenuation coefficient (0.75 vs. 0.66) and, as noted earlier, could be
explained by a greater prevalence of smaller particles in the slope-scale
samples
than in the micro-plot samples (see Bilro et al., 2011), reflecting a reduced
transport capacity of the overland flow. This explanation could also account
for the lower median organic matter concentration of the slope-scale samples,
with the larger charred particles being beyond the runoff's
detachment/transport capacity.
Best-fitting power functions of the relationships of sediment
concentration with normalized light loss at the slope scale for one
long-unburnt eucalypt (euc.) plantation on schist (F), five recently burnt
eucalypt (euc.) plantations on schist or granite (A, B, D, E, S) and one
recently burnt pine plantation on schist (C) (see Table 3).
Between-site differences related to fire, land cover and parent
material
The slope-scale samples tended to have higher median, third quartile and
maximum sediment concentrations than the micro-plot samples, as was also
noted in the previous section. At the same time, however, they revealed
similar contrasts between the six burnt study sites, except in the case of
the eucalypt plantation on granite (Table 3). The median sediment
concentration was significantly lower for the pine plantation
(0.11 g L-1) than for the burnt eucalypt plantations (on schist and
granite; ≥ 0.29 g L-1). The median sediment concentration for
the eucalypt plantation on granite lied within the range of values for the
other eucalypt plantations, unlike the case of the micro-plot samples.
This reflected a comparatively large increase in median sediment
concentration from the micro-plot to slope scale. This was in line with the
findings of Machado et al. (2015), who reported a marked increase in sediment
losses with spatial scale for the eucalypt plantation on granite (from 50 to
140 g m-2) as opposed to clear decreases for the pine plantation and the eucalypt plantation on schist at site B (from 85 and 200
to 3.5 and 6.1 g m-2, respectively).
The sediment concentrations for the unburnt eucalypt plantation were
significantly lower than those for the burnt eucalypt plantations. This
agreed with the slope-scale sediment losses reported by Machado et
al. (2015), being clearly lower for the unburnt than for the burnt eucalypt site on
schist (1.2 vs. 3.5 g m-2).
Sediment concentrations and corresponding organic matter (OM)
contents of the streamflow samples at the catchment outlet, and best-fitting
power function between sediment concentration (x; in g L-1) with
normalized light loss (y). Med.: median; iqr: inter-quartile range;
3rd q: third quartile; max.: maximum.
NSediment OM Best-fittingAdjustedconcentration content power functionR2g L-1% med.iqr3rd qmax.med.iqr1580.500.831.054.55228y=0.2809x0.70710.85
Better fits of the power function were obtained for the slope-scale samples
than for the micro-plot samples in the case of five of the six burnt study
sites, the pine site being the exception (Table 3). The pine plantation also
stood out for its low adjusted R2 value (0.72) compared to the other burnt
plantations (0.83–0.89). The R2 value was similarly low for the mulching
samples (0.71) and even considerably lower for the samples from the unburnt
eucalypt stand (0.52), suggesting an association between poor fits and
reduced sediment concentrations, unlike the case of the micro-plot
samples.
Relationships of sediment concentration with normalized light loss
at the catchment scale (symbols), and best-fitting power functions at the
catchment as well as slope scale for the eucalypt (euc.) and pine plantations
located within the catchment (see Tables 3 and 4).
The best-fitting curves for the slope-scale samples revealed a greater
similarity between the six burnt plantations than those for the micro-plot
samples (Fig. 5). Among the burnt plantations, only the D site stood out but
mainly because of a comparatively low base constant rather than a different
attenuation coefficient. For the same reason, the curve for the long-unburnt
plantation stood out even more from those of the burnt plantations. The
discrepancy of these two curves could well be an artefact from the
comparatively low sediment concentrations measured at the D and F sites, and
because possible differences in particle size due to reduced transport
capacity would point to steeper curves, as was the case of the curves fitted
to the micro-plot samples of sites A, C and D (see Sect. 4.1.2). Unlike in
the case of the latter three sites, the curves fitted to the slope-scale
samples of sites B and E agreed particularly well with those fitted to the
sites' micro-plot samples. This suggested that wider ranges of measured
sediment concentrations provided a more reliable basis for a consistent
relation between turbidity and sediment concentrations over spatial scales as
well as across study sites.
Catchment scale
The sediment concentrations of the streamflow samples were more similar to
those of slope-scale samples from the B site than from the A and C sites
(Table 4). This fitted in well with the fact that the B site represented the
dominant land cover–parent material combination within the catchment
(Table 4). Nonetheless, the maximum value of the streamflow samples was well
below the maximum values for all three slopes (4.55 vs. ≥ 6.59 g L-1). The median organic matter concentration of the
streamflow samples was also comparatively low (22 vs. ≥ 38 %). Even so,
it was substantially higher than the organic matter content of the sediments
deposited as bed load within the flume at the catchment outlet (Keizer et
al., 2015; 5 %).
The power function provided a good fit to the relationship of increasing
normalized light loss with increasing sediment concentration as revealed by
the streamflow samples, with an adjusted R2 of 0.85 (Table 4). The
fitted curve, however, differed considerably from the curves fitted to
slope-scale samples of the three slopes located within the catchment (Fig. 6). The
stronger attenuation coefficient for the streamflow samples (0.71 vs.
0.57–0.60) could be due to a prevalence of smaller particles in suspension,
especially because of the deposition of sediments in the flume at the
catchment outlet as well as in two upstream retention ponds (see Keizer et
al., 2015).
Conclusions
The principal conclusions of this study into the performance of a novel
plastic optical fibre (POF) turbidity sensor for measuring soil erosion
following a wildfire were the following:
the observed sediment concentrations were within the measurement range
of the POF sensor, attesting to the suitability of the sensor to be employed
during the initial phases of the so-called window of disturbance when erosion
losses tend to be highest and exported sediments tend to contain the highest
content of – charred – organic matter;
the relationships of sediment concentration with normalized light loss
varied markedly with spatial scale and, in particular, between micro-plot and
slope-scale samples, on the one hand, and, on the other, catchment-scale
samples, suggesting that scale-specific calibration curves are required to
guarantee optimal sensor performance;
the slope-scale relationships of sediment concentration with
normalized light loss varied clearly less between study sites than the
micro-plot-scale relationships, indicating that the need for site-specific
calibration curves is greater when sediment concentrations and, thus, erosion
rates are comparatively low;
the previous conclusion was also suggested by the comparison of the
sediment concentrations with and without an effective erosion mitigation
treatment;
the POF sensor would allow speeding up considerably the processing of
the runoff samples in the laboratory (and, perhaps, even in the field) and,
at the same time, would permit an efficient, stratified-sampling approach
towards the construction of scale- and/or site-specific calibration curves.
Given the very satisfactory performance of the sensor in this study, further
work will include redesigning the sensor and, in particular, its head to make
it more robust and more easy to handle, testing the new sensor for continuous
monitoring of streamflow turbidity under field conditions, and optimizing
data processing algorithms,
Acknowledgements
The present study was carried out in the framework of the projects TRANFIBRA
(project no. 23148) and FIRECNUTS (PTDC/AGRCFL/104559/2008), funded by FEDER,
through the Agência de Inovação S.A., in the framework of the
QREN SI I&DT program and funded by FCT/MCTES (PIDDAC), with co-funding by
FEDER through COMPETE (Programa Operacional Factores de Competitividade;
POFC), respectively. Additional financial support was provided by the EU-FP7
project RECARE (grant agreement no. 603498). We further
gratefully acknowledge the help of various colleagues of the earth surface
processes team with field data and sample collection and/or with laboratory
analysis of the sediment samples. Finally, we would like to acknowledge the
comments and suggestions by the handling editor as well as by the three
anonymous reviewers, which helped to improve this manuscript
considerably. Edited by: G. Guzmán
ReferencesAPHA (American Public Health Association): Total suspended solids dried at
105 degrees Celsius method 2540D, in: Standard Methods for the Examination of
Water and Waste Water, 20th Edn., Washington, DC, USA, 1998.Ben-Hur, M.: Using synthetic polymers as soil conditioners to control runoff
and soil loss in arid regions – a review, Aust. J. Soil Res., 44, 191–204,
2006.Bilro, L., Prats, S. A., Pinto, J. L., Keizer, J. J., and Nogueira, N.: Design and
performance assessment of a POF based sensor for measuring water turbidity,
Meas. Sci. Technol., 21, 107001, 10.1088/0957-0233/21/10/107001, 2010.Bilro, L., Prats, S., Pinto, J. L., Keizer, J. J., and Nogueira, R. N.: Turbidity
sensor for determination of concentration, ash presence and particle diameter
of sediment suspensions, Proc. Spie., 7753, 775356, 10.1117/12.885112,
2011.Campbell, C., Laycak, D., Hoppes, W., Tran, N. T., and Shib, F.: High
concentration suspended sediment measurements using a continuous fiber optic
in-stream transmissometer, J. Hydrol., 311, 244–253, 2005.Cardoso, J. C., Bessa, M. T., and Marado, M. B.: Carta dos solos de Portugal
(1:1,000,000), Serviço de Reconhecimento e de Ordenamento Agrário,
Secretaria de Estado da Agricultura, Lisbon, Portugal, 1971.Cardoso, J. C., Bessa, M. T., and Marado, M. B.: Carta dos solos de Portugal
(1:1,000,000), Agronomia Lusitana, 33, 461–602, 1973.Cerdà, A.: Changes in overland flow and infiltration after a rangeland
fire in a Mediterranean shrubland, Hydrol. Process., 12, 1031–1042, 1998.Downing, J.: Twenty-five years with OBS sensors: The good, the bad, and the
ugly, Cont. Shelf. Res., 26, 2299–2318, 2006.DRA-Centro (Direcção Regional do Ambiente do Centro): Plano de bacia
hidrográfica do Rio Vouga, 1∘ fase, Análise e diagnóstico
da situação de referência, Análise biofísica, Anexos,
Lisboa, Portugal, 1998.DUDF (Direcção de Unidade de Defesa da Floresta): Relatório
Anual de Áreas Ardidas e Ocorrências 2010, Autoridade Florestal
Nacional, Lisboa, 2011.Fernández, C., Veja, J. A., Fontúrbel, M. T., Pérez-Gorostiaga,
P., Jiménez, E., and Madrigal, J.: Effects of wildfire, salvage logging
and slash treatments on soil degradation, Land. Degrad. Dev., 18, 591–607,
2007.Ferreira, A. de Brum: Planaltos e montanhas do norte da Beira – estudo de
geomorfologia, Centro de Estudos Geográficos, Lisbon, Portugal, 1978.Inbar, M., Tamir, M., and Wittenberg, L.: Runoff and erosion processes after a
forest fire in Mount Carmel, a Mediterranean area, Geomorphology, 24, 17–33,
1998.Keeley, J. E.: Fire intensity, fire severity and burn severity: A brief review
and suggested usage, Int. J. Wildland Fire, 18, 116–126, 2009.Keizer, J. J., Martins, M. A. S., Prats, S. A., Faria, S. R., González-Pelayo,
O., Machado, A. I., Rial-Rivas, M. E., Santos, L. F., Serpa, D., and Varela,
M. E. T.: Within-in flume sediment deposition in a forested catchment
following wildfire and post-fire bench terracing, north-central Portugal,
Cuadernos de Investigación Geográfica, 41, 149–164, 2015.Lane, P. N. J., Sheridan, G. J., and Noske, P. J.: Changes in sediment loads and
discharge from small mountain catchments following wildfire in south eastern
Australia, J. Hydrol., 331, 495–510, 2006.Lavabre, J. and Martin, C.: Impact d'un incendie de forêt sur l'hydrologie
et l'érosion hydrique d'un petit bassin versant méditerranéan,
Human Impact on Erosion and Sedimentation, Proceedings of Rabat Symposium S6,
April 1997, IAHS Publication 245, 39–47, 1997.Machado, A. I., Serpa, D. R., Ferreira, R. S. V., Rodríguez-Blanco, M. L.,
Pinto, R., Nunes, M. I., Cerqueira, M. M., and Keizer, J. J.: Cation export
by overland flow in a recently burnt forest area in north-central Portugal,
Sci. Total Environ., 524–525, 201–212, 2015.Malvar, M. C., Prats, S. A., Nunes, J. P., and Keizer, J. J.: Post-fire overland flow
generation and inter-rill erosion under simulated rainfall in two eucalypt
stands in north-central Portugal, Environ. Res., 111, 222–236, 2011.Malvar, M. C., Martins, M. A., Nunes, J. P., Robichaud, P. R., and Keizer, J.
J.: Assessing the role of pre-fire ground preparation operations and soil
water repellency in post-fire runoff and inter-rill erosion by repeated
rainfall simulation experiments in Portuguese eucalypt plantations, Catena,
108, 69–83, 2013.Martins, M. A. S., Machado, A. I., Serpa, D., Prats, S. A., Faria, S. R.,
Varela, M. E. T., Gonzalez-Pelayo, O., and Keizer, J. J.: Runoff and
inter-rill erosion in a Maritime Pine and a eucalypt plantation following
wildfire and terracing in north-central Portugal, J. Hydrol. Hydromech., 61,
261–269, 2013.Mayor, A. G., Bautista, S., Llovet, J., and Bellot, J.: Post-fire hydrological and
erosional responses of a Mediterranean landscape: seven years of
catchment-scale dynamics, Catena., 71, 68–75, 2007.Moody, J. A., Shakesby, R. A., Robichaud, P. R., Cannon, S. H., and Martin, D.
A.: Current research issues related to post-wildfire runoff and erosion
processes, Earth-Sci. Rev., 122, 10–37, 2013.Omar, A. F. B. and MatJafri, M. Z. B.: Turbidimeter design and analysis: a review on
optical fiber sensors for the measurement of water turbidity, Sensing, 9,
8311–8335, 2009.Postolache, O. A., Silva Girao, P. M. B., Dias Pereira, J. M., and Ramos, H. M.: Multibeam optical system and neural processing for turbidity
measurement, IEEE Sens. J., 7, 677–684, 2007.Prats, S. A., Martins, M. A. S., Malvar, M. C., Ben-Hur, M., and Keizer, J.
J.: Polyacrylamide application versus forest residue mulching for reducing
post-fire runoff and soil erosion, Sci. Total Environ., 468–469, 464–474,
2014.Prats, S. A., Wagenbrenner, J. W., Martins, M. A. S., Malvar, M. C., and Keizer, J.
J.: Hydrological implications of post-fire mulching across two spatial
scales, Land Degrad. Dev., 10.1002/ldr.2422, online first, 2015.Robichaud, P. R., Elliot, W. J., Pierson, F. B., Hall, D. E., and Moffet, C.
A.: Predicting postfire erosion and mitigation effectiveness with a web-based
probabilistic erosion model, Catena, 71, 229–241, 2007.Ruhl, C. A., Schoellhamer, D. H., Stumpf, R. P., and Lindsay, C. L.: Combined use of
remote sensing and continuous monitoring to analyse the variability of
suspended-sediment concentrations in San Francisco Bay, California, Estuar.
Coast. Shelf S., 53, 801–812, 2001.Shakesby, R.: Post-wildfire soil erosion in the Mediterranean: review and
future 730 research directions, Earth-Sci. Rev., 105, 71–100, 2011.Shakesby, R. A. and Doerr, S. H.: Wildfire as a hydrological and geomorphological
agent, Earth-Sci. Rev., 74, 269–307, 2006.
Thomas, A. D., Walsh, R. P. D., and Shakesby, R. A.: Nutrient losses in eroded
sediment after fire in eucalyptus and pine forests in the wet Mediterranean
environment of northern Portugal, Catena, 36, 283–302, 1999.WRB (World reference base for soil resources): World Soil Resources Reports,
103, FAO, Rome, 2006.Yeo, T. L., Sun, T., and Grattan, K. T. V.: Fibre-optic sensor technologies for
humidity and moisture measurement, Sensor Actuat. A-Phys., 144, 280–295,
2008.Ziemann, O., Krauser, J., Zamzow, P. E., and Daum, W.: POF Handbook – Optical
Short Range Transmission Systems, 2nd Edn., Springer-Verlag, Berlin
Heidelberg, 2008.