Running head:
21kyr of Sediment Flux Simulations of the Po River
Title:
Predicting Discharge and Sediment Flux of the Po River, Italy since the Late Glacial
Maximum
Authors: Albert J. Kettner and James P.M. Syvitski
Environmental Computation and Imaging Group
Institute of Arctic and Alpine Research,
University of Colorado,
Boulder, CO 80309-0450
USA
Corresponding author: Albert J. Kettner
Fax.: +1-303-492-6388
Email: [email protected]
Keywords: Po River, paleoclimate, HydroTrend, drainage basin change, sediment
simulations, numerical model Abstract
HydroTrend numerically simulates the flux of water and sediment delivered to the coastal ocean on a daily time scale, based on drainage basin and climate characteristics.
The model predicts how a river may have behaved in the geological past, provided that appropriate assumptions are made regarding past climate and drainage basin properties.
HydroTrend is applied to the Po River in Italy, to simulate a high resolution discharge and sediment flux record since the Late Glacial Maximum (LGM). A short validation experiment of 12 years under present conditions shows a high correlation (r2 = 0.72) with
12yr daily measured discharge. Monthly variations in simulated Po River discharge and
sediment discharge for this same time period show even closer agreement.
The Po during the LGM was a period of much colder and dryer climate, and much
larger drainage basin area due to sea level change. The Younger Dryas is shown to be an
exceptional period as glacier ablation was dominant.
The Po River during the Late Pleistocene (21-10 Cal. kyr BP) had an average
suspended sediment flux of 32.5 Mt yr-1 with an average bedload of 0.91 Mt yr-1. This is
~70% more then during the Holocene average (10 – 0 Cal. kyr BP), when simulations indicate a suspended sediment flux of 18.8 Mt yr-1 and a bedload flux of 0.53 Mt yr-1. The
Würm Stadial shows the highest suspended sediment concentration. The Bølling and the
Younger Dryas periods were able to transport much coarser grain sizes, because of extreme river floods due to glacier ablation.
1 Introduction
It is of importance to understand and predict the impact of climate change and human influence on the accumulation of sediment on continental margins (Syvitski, 1999). In the framework of the EUROSTRATAFORM project the Adriatic Sea was selected as a study site where these dynamics are amplified, because of the semi-closed nature of the basin.
The Po River strongly influences the dynamics of the Adriatic Sea with its great input of fresh water (Cushman-Roisin, 2001) and sediment, both presently and during the Late
Glacial Maximum (Asioli, 2001). Unfortunately, sediment load measurements are limited and discharge measurements have only been collected during the last two centuries. It is certain that we can not extrapolate these records either forward or back in time, because a number of controlling factors have changed over time. Firstly, analysis of marine cores indicates that climate shifts like the regional warming since the Last Glacier Maximum
(LGM) (21 Cal. kyr) have profoundly impacted discharge and sediment flux of the river
(Asioli, 1999). Secondly, the large reservoirs originally formed from Alpine glaciers presently control one third of the total Po River discharge and have regulated the discharge and sediment load differentially over a long time period (Hinderer, 2001).
Thirdly, the size of the paleo-drainage basin changed significantly due to sea-level fluctuations in the shallow Adriatic Sea. We advocate that numerical modeling offers a possibility to predict the hydrograph and its sediment delivery to the coastal ocean by taking into account the shifts of climate and sea-level changes as well as the influence of reservoirs.
We simulate the Po River since the LGM employing three new subroutines for glacier dynamics, reservoir sediment trapping and fluvial sediment transport, with the climate-
2 driven hydrological model HydroTrend. The revised model is subsequently validated
against observations of the modern Po River. Next we discuss the assumptions used to
arrive at the input boundaries for climate, sea-level change and drainage basin
characteristics. Finally, we explain the results and investigate which processes are
dominant during certain characteristic time periods in driving the discharge and sediment
flux of the Po River.
Geological setting
Hydrology of the Po River
The Po River is 650 km long, and has over 141 distributaries. The 74,500 km2 river
catchment of the Po is bounded at the North by the Alps with peaks over 4,500 m, and at
the South-West by the Apennines mountain chain with peaks generally less than 2,000 m.
More then a third of the drainage area (30,800 km2) can be considered mountainous. The
Po River has two flood periods, June (freshets caused by snow melting) and November
(corresponding to precipitation maxima) and two low water periods, January and August
(Fig. 3) (Marchi et al., 1996; Cattaneo et al., 2003). The average discharge of the Po
River is 1.5 × 103 m3s-1 measured at Pontelagoscuro (near Ferrara) 90 km from the coast and just before the apex of the delta. Downstream of Pontelagoscuro, the Po forms a delta consisting of six major distributaries: Levante, Maistra, Pila, Tolle, Gnocca and Goro.
The main channel is the Pila, carrying 60% of the total discharge.
Although the river discharge fluctuations are dominated by rainfall, the hydropower management regime influences the discharge considerably. The largest reservoirs
Maggiore, Lugano, Como, Iseo and Garda Lakes (Fig. 1) were formed by Alpine glaciers
3 during the Pleistocene glaciation. A third of the total discharge of the Po is affected by these reservoirs (Camusso et al., 2001). The lakes are regulated for hydropower production and irrigation, and are located in the most highly populated and industrialized area of Italy, Insubria, the northern area of the Po catchment (Marchi et al., 1996).
Adriatic Sea
The Adriatic Sea forms the northernmost part of the Mediterranean. It is a relatively shallow almost rectangular basin bordered to the north by the Alps, to the west by the
Apennines and on the east by the Dinaric mountain chain. This temperate warm sea is more than 800 km long in a NW-SE direction and has an average width of about 200 km
(Fig. 1). The Adriatic Sea is often divided into three geographical regions, namely the
Northern, Middle and Southern Adriatic basins. The Northern Adriatic, defined as the area lying north of the 100m isobath, has a wide continental shelf, sloping gently south and is quite shallow. The Middle Adriatic comprises the three trenches of the Middle
Adriatic Pit, with a maximum depth of 270 m and is bounded by the 170 m deep
Palagruza Sill. The Southern Adriatic extends from the Palagruza Sill to the Strait of
Otranto; including the South Adriatic Pit, which is at its deepest point around 1200 m.
The Ionian Sea connects the Adriatic Sea and Mediterranean Sea (Fig. 1).
The Po River, as main source of the river water discharge into the northern Adriatic, forms mainly during the winter. The river plume hugs the western side of the basin as aided by the dominant cyclonic circulation (Cattaneo et al., 2003). Along with wave resuspension, the plume is responsible for the formation of a 35-m thick mud wedge that extends from the Po Delta to the Gargano subaqueous delta, 500km further south. During
4 summer the Po River plume generally spreads over the entire northern subbasin as a thin
surface layer, ~5m (Cushman-Roisin et al., 2001).
HydroTrend methodology
Model description
HydroTrend numerically simulates discharge and sediment loads at a river mouth at
a daily time scale (Syvitski, 2002; Morehead et al., 2003). The model is designed to make discharge predictions based on drainage basin characteristics and climate, even when
field measurements of river flow are not available (Syvitski et al., 1995). Provided that appropriate assumptions are made regarding past climate, the model can predict how a river behaves in prehistoric periods (Syvitski et al., 1999). Syvitski et al. (2003; 2004) show that sediment transport predictions are accurate to the same level of accuracy of most global field observations.
HydroTrend incorporates drainage basin properties (river networks, hypsometry, relief, reservoirs) based on high-resolution digital elevation models (for example
HYDRO1k DEM). A number of additional biophysical parameters are incorporated to calculate the steady-state hydrological balance (basin-wide temperature, precipitation, glacier equilibrium line altitude (ELA), evapo-transpiration, canopy, soil depth, hydraulic conductivity). We refer to a detailed description of HydroTrend in Syvitski et al. (2003) and in this section discuss only recently changed modules relevant for the Po River simulations.
The Po River drainage basin consists substantial Alpine glaciers, which affect the hydrological balance even more strongly at glacial times and during subsequent warming.
5 The model uses the ELA in combination with the hypsometry to determine glacier area.
ELA changes over time result in a glacier area changes. HydroTrend now employs an
exponential relationship for glacier area, Ag (km2) versus glacier volume, Vg (km3) (Bahr
et al., 1997) (equation 1) to simulate glacier ablation or growth, and tracks changes in
hydrological balance and sediment flux at the river outlet.
1.38 Vg = 31.1× ()Ag (1)
A stochastic model (Morehead et al., 2003) is used to calculate the daily suspended
sediment load fluxes:
C
⎛ Qs ⎞ ⎛ Q ⎞ ⎜ ⎟ =ψ ⎜ ⎟ (2) ⎝ Qs ⎠ ⎝ Q ⎠
Wherein Qs is the daily suspended sediment discharge (kgs-1), Q is the daily
3 -1 Q Q discharge (m s ), s is the long-term average of Qs, is the long-term average of Q, ψ is a log-normal random variable and c is a normal random variable.
The long-term average of Qs is defined as:
α 4 α5 k T Qs = α 3 A R e (3)
Wherein A is the drainage basin area (km2), R is the maximum relief (m), T is the basin-average temperature (°C), α3, α4, α5 and k are dimensionless coefficients which
depend on climatic zone based on the geographical location of the drainage basin (table
3) (Syvitski et al., 2003). Note: Equation 3 can be rewritten in terms of discharge as a
substitute for area. In the case of the Po River specifically, climate changes from Last
Glacial Maximum (LGM) until present do not exceed the threshold values to force
6 changes in α’s or k. The long-term average suspended sediment will only be affected by changes in A, R and T .
-1 The daily bedload Qb (kgs ) is simulated using a modified Bagnold (1966) equation:
⎛ ρ ⎞ ρgQ β se Q = ⎜ s ⎟ b when u ≥ u (4) b ⎜ ⎟ cr ⎝ ρ s − ρ ⎠ g tan f
-3 -3 Where ρs is sand density (kgm ), ρ is fluid density (kgm ), Q is the daily discharge
3 -1 (m s ), s is the slope of the riverbed (°), eb is the bedload efficiency, β is a dimensionless
bedload rating term, f is the limiting angle of repose of sediment grains lying on the river
-1 -1 bed (°), u is stream velocity (ms ), and ucr is the critical velocity (ms ) needed to initiate
bedload transport.
The Po River discharge is strongly affected by five large reservoirs, sediment fluxes
are likely influenced even more by sediment trapping. HydroTrend has been adjusted to
incorporate reservoir effects on sediment load. The model simulates Trapping Efficiency
(TE) depending on the reservoir volume either by the adjusted Brown equation
(Verstreaten et al., 2000), for reservoirs smaller or equal than 0.5km3, or the modified
Brune equation (Vörösmarty et al., 1997), for reservoirs larger than 0.5km3 (formula 5
and 6).
0.05 TE =1− (5) ∆τ
Wherein ∆τ is the approximated residence time and is estimated by:
ni
∑Vi ∆τ = 1 (6)
Q j
7 3 Where Vi is operational volume of the reservoir I (km ), Q is the discharge at mouth
of each regulated subbasin j (m3s-1).
The model with documentation, example files and references is available on the web: http://instaar.colorado.edu/deltaforce/models/hydrotrend.html.
Model validation
Present-day climate statistics including monthly mean temperature and precipitation
and the associated standard deviations of the Po drainage basin are retrieved from daily
data records (1977 - 1991) of 20 climate stations distributed over the drainage basin. The
stations have been weighted by elevation to determine characteristic basin-wide values.
The yearly mean temperature, precipitation and their standard deviations, similarly
weighted by elevation, are obtained from monthly data records (1760 – 1995) of 13
climate stations from the Global Historical Climatology Network of National Oceanic
and Atmospheric Administration, (NOAA) (Vose et al., 1997)
(http://ingrid.ldeo.columbia.edu/SOURCES/.NOAA/.NCDC/.GHCN/).
HydroTrend proved an appropriate model to simulate the Pontelagoscuro gauging
station, which is located closest to the river mouth before the main channel splits into
distributary channels (Fig. 1). At this delta apex a high correlation is found between
modeled discharges and time series of 12 yr daily discharge data (1990-2001) and
monthly discharge (1918-2002) (Fig. 2 & 3). Modeled data correlates significantly with
the observed data (r2 of 0.72), as calculated from 100m3s-1 intervals.
Figure 3 indicates that the monthly mean discharges are underpredicted during the winter months (November till March) and slightly overpredicted during the summer
8 (April till September). This deviation relates to the well known 'reservoir effect' common in river basins containing major reservoir lakes (Bobrovitskaya, 1996). The river
discharge is artificially kept low in summer by storage of water in the reservoirs and extra water is released in winter to generate hydro-power, which results in a dampening of the seasonal discharge signal. Over a third of the discharge of the Po River discharge is affected by these hydro-power reservoirs (Camusso et al., 2001).
At the delta apex, HydroTrend predicts an average discharge of 1,542 m3 s-1 and peak
daily discharges of 10,800 m3 s-1 comparable to the measured floods of 1951 and October
2000 (respectively 10,300 and 9,600 m3 s-1, Table 1). Average suspended sediment load
of 17 × 106 t yr-1 with peak years of 39 × 106 t yr-1 are predicted. Bedload is predicted to contribute only ≈2.5% of the total sediment output of the Po river system (table 1).
Monthly variations in simulated Po River discharge and sediment discharge are similar to measurements (fig. 4). The one difference between simulated and observations is in the area of scatter. The simulations show less variation in the sediment discharge because it does not include hysteresis effects that provide much of the scatter in the rating plot of the observed data.
Human impact in the form of damming and sediment mining has caused a decrease of
38% in sediment supply during the second half of the 20th century for the Po (Surian et al., 2003). The model takes the trapping of sediment due to the reservoirs into account as discussed previously. The total volume of the mined sediment is estimated to be more than 2 Mm3yr-1 in the 1990’s (Marchetti, 2002). Sediment mining cannot be modeled,
because this is a 'policy-driven' amount and is impossible to include in a physical model.
9 However, subtracting the extracted sediment on a yearly basis from the simulated suspended sediment load improves the match with the measured field data (table 1).
Input and boundary conditions
The Mediterranean Sea and the Adriatic Sea level were about 120 meter below present msl during the LGM (Fairbanks, 1989; Lambeck et al., 2000). The evolution of the Po basin area since LGM has been reconstructed based on an approximation of the paleotopography constructed by merging the present-day Digital Elevation Model (DEM) and bathymetric data (fig. 2). GIS analysis provides the shifted coastline and subsequently the paleo-drainage basin. During LGM, the shoreline was about ≈250 km
SE of the present-day Po River outlet (Amorosi, et al., 1999), because of the relatively shallow gradient of the Northern Adriatic Sea. This implies that part of the Apennine rivers, for example the Metauro, Potenza and the Chienti, as well as present-day Northern
Italian and Croatian rivers were contributaries to the Po River which eventually drained into the modern foredeep basin of the Apennine chain. We increased the sea level in 12 steps of 10 m. Each step caused a drainage area decrease. The total area of the Po drainage basin increased during the LGM by a factor of 2.6 compared to its present area
(Fig. 2).
Monthly climate statistics at the LGM are estimated based on the Community Climate
Model1 (CCM1) (Kutzbach et al., 1998). These modeled climate statistics for LGM have been combined with the present climate statistics and interpolated over time using a normalized δ18O GRIP curve as a forcing factor (Dansgaard et al., 1993). Using the
CCM1 climate statistics is justified by the fact that Peyron et al. (1998) find cool steppe
10 vegetations to be coherent with climatic conditions predicted by CCM1. According to the
CCM1 model predictions the mean basin temperature is much lower (2.4°C vs.8.8°C) and precipitation decreases by 18% compared to present day values (fig. 5). This is confirmed by Fauquette et al. (1999) and Zonneveld (1996), who concluded that the LGM glacial period was characterized by very low temperatures and a less humid climate, based on pollen data.
Despite the large latitudinal difference between the location of the GRIP ice core and our study area (respectively 72° vs. 41°N), we are confident of the relevance the normalized δ18O GRIP curve, because it correlates strongly with local paleo-climate proxies like tree-rings (Friedrich et al., 2001) stable isotopes in fossil mammal remains
(Huertas et al., 1997), marine sedimentary cores (Asioli et al., 2001; Sbaffi et al., 2001), biostratigraphical pollen data (Asioli et al., 1999) and lake cores (Watts et al., 1996;
Ramrath et al., 1999; Allen et al., 2000).
The Alps were almost completely covered by the late-Würmian ice-sheet (Hinderer,
2001; Florineth et al., 1998), but the ice cap did not add significant area to the Po drainage basin as the glacier drainage divide roughly followed the basin divide (Baroni,
1996). We used the normalized δ18O curve to force the glacier equilibrium line altitude
(ELA) changes. ELA values are based on global latitude-specific averages both for the
LGM and present day (Fig. 5)
Hinderer, (2000) estimates the timing of deglaciation of the peri-alpine lake basins which form the present-day reservoirs of the Po basin The deglaciation and consequent activation of the reservoirs is a two-step process due to difference in elevation of the main reservoirs (~180 m). In addition, the Trapping Efficiency (TE) is a function of basin
11 area (equation 5 and 6), so that for the TE decreases for the whole basin over time
concurrent with the decreasing basin area due to rising sea level (fig. 5).
Detailed description of four characteristic periods.
Four time periods have been more closely studied because we presumed that these
periods may be distinguished in the stratigraphy of marine cores; The Würm LGM (21
Cal. kyr – 17.5 Cal. kyr), Bølling (15.6 Cal. kyr – 13.9 Cal. kyr), Younger Dryas (12.9
Cal. kyr – 11.1 Cal. kyr) and the Upper Subatlantic (0.8 Cal. kyr – present).
The maximum extension of the late-Würmian ice-sheet in the Alps was reached
during the Würm LGM, when glaciers were extending as far as the Alpine foreland (fig
1). Yearly average temperature for the Po drainage basin was 2.4°C and precipitation was
18% less than nowadays. The drainage basin was 1.9 × 105 km2, or 2.6 times larger than
today. The average ELA was 1920m and increased slowly with 0.10myr-1.
During the Bølling, the major valleys were ice-free as were the deeply-incised lake
basins in major intra-Alpine valleys and in the Alpine foreland. In our model, these lakes
start acting as reservoirs. TE is calculated for the whole river basin and was during the
Bølling ~8%. Furthermore, the ELA was rapidly increasing from 2015m with on average
1.0myr-1 from 15.6 Cal. kyr till 14.4 Cal. kyr, after that time it dropped again with
0.6myr-1 untill the end of the Bølling. The drainage basin area not significantly compared to the Würm LGM, due to the steep gradient of the outermost part of the Northern
Adriatic Sea, which stabilizes the coastal line position for a long period, despite sea level rise.
12 The Younger Dryas (YD) was the most significant rapid climate change event that
occurred during the last deglaciation of the North Atlantic region (Yu, 2001). This
transition marks the last major climate reorganization during the deglaciation. ELA
increased on average 0.7myr-1. Furthermore sea level rise decreased the drainage basin by
almost 50% compared to the Würm LGM (1.1 × 105 km2). Due to this decrease the
average TE for the drainage basin as a whole increased to 15%.
The Upper Subatlantic (S) period reflects the present climate conditions with less
interannual variability as compared to the Pleistocene. The drainage basin of the Po River
is similar as today. The glacier ELA is 1300m higher than during the Würm LGM at
3200m. Presently the Po River has a TE of 20%.
HydroTrend Long-term Simulation Results
Drainage basin area changes due to sea level fluctuation has a dominant impact on discharge and sediment load of the Po River. Figure 9 illustrates the relation between
total yearly discharge and drainage basin area (r2 = 0.80). The decreasing drainage area is
reflected in a decreasing discharge. The large spread around the mean values is due to the
stepwise nature of the drainage basin changes. Other forcing factors (e.g. climate) change
at higher time resolution than drainage area and impact the discharge prediction as well,
causing the variability over the y-axis (fig 9). The exception of the linear relationship
occurs at the end of the Younger Dryas, when the area is ≈10.7 × 104 km2. Glacier ablation has a more significant impact during that time than area change.
Climate studies in the Alpine region indicate that glaciers reached their maximum extent during the LGM and were approximately at their present position at the start of the
13 Holocene (Florineth et al., 1998; Hinderer, 2001). HydroTrend simulations reflect this trend (fig. 6). Based on Kotlyakov et al. (1997) we mapped the glacier extent into a GIS and found that fifteen percent (≈ 2.9 × 104km2) of the total Po River drainage basin during the LGM was covered by glaciers. A HydroTrend simulation for the LGM, based
on the ELA, predicts a similar glacier area as our GIS reconstruction (fig. 6). Climate changed rapidly during the Bølling and the end of the Younger Dryas. The model simulates several melting phases including the Bølling and the Younger Dryas which all have significant impact on the total discharge for the Po (some even up to 40% and 60% of the total annual discharge, fig. 7). The two highest simulated glacier meltwater pulses coincide with the measured two global major pulses of meltwater entering the world ocean (Bradley, 1999).
Figure 7a shows two years of daily discharge simulations. The first year represents a typical year where the glaciers are in a steady state or slightly growing. Glacier melt in the summer months has no significant impact on the total discharge. The second year represents fast glacier ablation, similar to the melting at the end of the Younger Dryas. In that year, 60% of the total annual discharge is derived from glacier ablation, and it is mostly released during the summer months. During those summer months 90% of the discharge is from glacier ablation. Figure 7b represents the daily average suspended sediment load for the same two years. As a result of the high discharges the Po River transports 45 times more sediment during the glacier ablation year, when 97% of the transport takes place during the summer months (May - Oct).
Figure 10 shows the maximal peak discharges per year plotted against the average discharge for that year. Years with strong peak discharges tend to be in years with higher
14 than average discharge. One would expect that peak discharges during the Pleistocene are mostly occurring during glacier ablation years. Yet, analyses indicate that above normal snowmelt and rainfall events are the controlling factor. Discharge peaks higher than 2σ occur mostly in years with higher than average rainfall (97%) and snowfall (87%). Only
48% of the peaks occur during glacier ablation years. Thus the rapidly fluctuating short- term precipitation signal, based on the climate statistics, dominates the peak event generation in the model instead of the medium-term signal of the glacier melt.
Figure 8 indicates that periods with consistently high discharge due to glacial melt have more impact on the bedload than on suspended sediment load. We predict that these periods could be identified by layers with coarser grain sizes in the stratigraphic record.
For the entire Po River, we estimate an average Pleistocene (21-10 Cal. kyr BP) suspended sediment flux of 32.5 Mt yr-1 with an average bedload of 0.91 Mt yr-1. This is
1.7 times more than during the Holocene (10 – 0 Cal. kyr BP), when we simulate on average a suspended sediment flux of 18.8 Mt yr-1 and a bedload of 0.53 Mt yr-1. These results correlate with values presented by Cattaneo et al. (2003). They analyzed the sediment load entering the Adriatic Sea for four different catchment areas: Eastern Alpine rivers, Po River, Eastern Apennine rivers and rivers south of Gargano. These subsystems have sediment loads of respectively 3 Mtyr-1, 15 Mtyr-1, 32.2 Mtyr-1and 1.5 Mtyr-1. Areas surrounding the Adriatic that are not included (mostly in Croatia) are mainly carbonatic and their sediment supply is negligible. During the LGM we estimated that the Po drainage area (fig 1.) covers the Eastern Alpine rivers, the present Po River and captures roughly half of the Eastern Apennine rivers, as defined by Cattaneo et al. (2003). The
15 estimated total according to their analyzes is 34.1 Mtyr-1 which closely matches our
prediction (32.5 Mtyr-1).
The focus on the four different time steps confirms that the different conditions
(climate and basin area) result in distinctly different water and sediment pulses, which
makes us postulate that these time periods will be identifiable in the stratigraphy.
Although research is still in preliminary phase, we analyzed similarities in sedimentation
rate of a well dated gravity core IN-68-9 described by Zonneveld (1996) and our modeled
suspended sediment load data.
In addition, the grain size distributions may carry distinct signals as well. The Würm
LGM has the highest suspended sediment concentration of the four time periods (table 4).
But because the discharge is higher due to glacier ablation during the Bølling and the
Younger Dryas those periods will identify them selves because they have much coarser grain sizes due to a higher percentage of bedload inflow (Fig 11, A2/D2).
Conclusions
HydroTrend is a viable model for predicting the Po River flux of water and sediment under modern conditions. We validated the HydroTrend model predictions on a daily time scale for a 12year record and found a highly significant correlation with the discharge and sediment load data at the most downstream gauging station. In addition, a
medium time scale validation has been applied by comparing model output to the
monthly discharge fluctuations for the record that exist from 1760-1995. HydroTrend was
able to capture the monthly variations over that time scale. Also, the simulated daily
peaks are comparable to two measured peak floods over this last century, which is
16 important in view of the generation of distinct flood layers in the offshore stratigraphy.
HydroTrend is designed to be sensitive to hydro-climatological forcing, this implies that
some human influences are difficult to capture. The present-day yearly Po River fluxes
shows effects that are a direct result of hydro-power management and sediment mining.
Both effects are not relevant for long-term hindcasting of river fluxes, which is the objective of our simulations.
HydroTrend offers a tool to systematically combine a number of controlling factors;
temperature and precipitation changes, sea level rise, glacial ice cap melt, and evolution of reservoirs. Subsequently, it is possible to disentangle the dominance of the separate controls.
The long-term experiment shows that sea level fluctuation is the largest control on the fluxes of the Po since the LGM. Sea-level rise controls the 2.6-fold decrease in drainage area of the Po River, which proportionally affects the absolute discharge and sediment flux.
Temperature and precipitation are of importance only as they impact the melt or growth of the Alpine Icecap. The release of melt-water from the ablating Alpine Icecap adds substantial water and sediment to the coastal ocean system. We found that in strong glacier ablation year 90% of the total discharge could be attributed to glacial melting in the summer months. Sediment transport increased upto 45 times in such an extreme glacier ablation year. The temperature and precipitation are the main forcers of daily variations, and determine the occurrence of peak events.
The Younger Dryas event and its recovery, is the single largest excursion in terms of sediment delivery including the delivery of bedload. In that specific period, the system is
17 completely glacial-melt dominated. We expect that the extremely high discharges and
high bedload flux have produced a significant signal of coarse sediments in the offshore record.
The modern Po is a relatively quiet and dynamically boring river system compared to the
Late Glacial maximum.
The input variables as well as the daily and yearly HydroTrend simulations since the Last
Glacial Maximum are made available on the web
(ftp://instaar.colorado.edu/pub/forHydrousers/sediment_flux_Po_since_LGM).
Acknowledgements
We very much thank Dr. Tom Drake for supporting this work through the ONR’s
EuroSTRATAFORM project. Dr. Richard Signell is thanked for processing and
providing the high resolution bathymetry data of the Adriatic Sea. We thank Dr. Mark
Dyurgerov for discussing with us the new developed glacier subroutine in the
HydroTrend model. Dr. A. Correggiari is thanked for help in obtaining Po River
observations used in this and other studies. The authors are indebted to Dr. Irina Overeem
for her thoughtful review of an early version of this manuscript.
18 References
Allen, J.R.M. and Huntley, B., 2000. Weichselian palynological records from southern
Europe: correlation and chronology. Quaternary International 73/74, 111-125.
Amorosi, A., Colalongo, M.L., Pasini, G. and Preti, D., 1999. Sedimentary response to
Late Quaternary sea-level changes in the Romagna coastal plain (northern Italy).
Sedimentology 46, 99-121.
Asioli, A., Trincardi, F., Lowe, J.J. and Oldfield, F., 1999. Short-term climate changes
during the Last Glacial-Holocene transition: comparison between Mediterranean
records and the GRIP event stratigraphy. Journal of Quaternary science 14, 373-381.
Asioli, A., Trincardi, F., Lowe, J.J., Ariztegui, D., Langone, L. and Oldfield, F., 2001.
Sub-millennial scale climatic oscillations in the central Adriatic during the
Lateglacial: Palaeoceanographic implications. Quaternary Science Reviews 20, 1201-
1221.
Bahr, D.B., Meier, M.F., Peckham, S.D., 1997. The physical basis of glacier volume-area
scaling. Journal of geophysical research, vol. 102, no. B9, 20,355-20,362.
Baroni, C.,1996. The Alpine “Iceman” and Holocene Climatic Change. Quaternary
Research 46, 78-83.
Bradley, R.S., 1999. Paleoclimatology: Reconstructing Climates of the Quaternary. In:
Dmowska, R., Holton, J.R. (Eds.), International Geophysics series, Volume 64.
Bobrovitskaya, N.N., Zubkova, C. and R.H. Meade, 1996. Discharges and yields of
suspended sediment in the Ob' and Yenisy Rivers of Siberia. In: Walling, D.E. and
B.W. Webb (eds) Erosion and Sediment Yield: Global and Regional Perspectives.
Intern. Association of Hydrological Sciences Publ. no. 236, 1996.
19 Camusso, M., Balestrini, R. and Binelli, A., 2001. Use of zebra mussel (Dreissena
polymorpha) to assess trace metal contamination in the largest Italian sub alpine
lakes. Chemosphere 44, 263-270.
Cattaneo, A., Correggiari, A., Langone, L. and Trincardi, F., 2003. The late-Holocene
Gargano subaqueous delta, Adriatic shelf: Sediment pathways and supply
fluctuations. Marine Geology 193, 61-91.
Cushman-Roisin, B., Gacic, M., Poulain, P. and Artegiani, A., 2001. Physical
Oceanography of the Adriatic Sea; Past, Present and Future. Cushman-Roisin et al.,
eds., Kluwer Academic Publishers, 304 pp.
Dansgaard, W., Johnsen, S., Clausen, H.B., Dahl-Jensen, D., Gundestrup, N.S., Hammer,
C.U., Hvidberg, C.S., Steffensen, J.P., Sveinbjornsdottir, A.E., Jouzel, J. and Bond,
G., 1993. Evidence for general instability of past climate from a 250-kyr ice core
record. Nature 364, 218-220.
Fairbanks, R.G., 1989. A 17,000 year glacio-eustatic sea level record: Influence of glacial
melting rates on the Younger Dryas event and deep-ocean circulation. Nature 342,
637-642.
Fauquette, S., Guiot, J., Menut, J., de Beaulieu, J-L., Reille, M. and Guenet, P., 1999.
Vegetation and climate since the last interglacial in the Vienne area (France). Global
and Planetary Change 20, 1-17.
Florineth, D. and Schlüchter, C., 1998. Reconstructing the Last Glacial Maximum (LGM)
ice surface geometry and flowlines in the Central Swiss Alps. Eclogae geologicae
Helvetiae 91, 391-407.
20 Friedrich, M., Kromer, B., Kaiser, K.F., Spurk, M., Hughen, K.A. and Johnsen, S.J.,
2001. High-resolution climate signals in the Bølling-Allerød Interstadial (Greenland
Interstadial 1) as reflected in European tree-ring chronologies compared to marine
varves and ice-core records. Quaternary Science Reviews 20 1223–1232.
Hinderer, M., 2001. Late Quaternary denudation of the Alps, valley and lake fillings and
modern river loads. Geodinamica Acta 14, 231-263.
Huertas, A.D., Iacumin, P. and Longinelli, A., 1997. A stable isotope study of fossil
mammal remains from the Paglicci cave, southern Italy, 13 to 33 ka BP:
palaeoclimatological considerations. Chemical Geology 141, 211-233.
Kotlyakov, V.M., Chernova, L.P., D’yakovo, A.M., Glebova, L.N., Knonvalova, G.I.,
Osipova, G.B., Rototaeva, O.V., Timofeeva, N.A.and Varnakova, G.M., 1997. World
Atlas of Snow and Ice Resources.
Kutzbach, J., Gallimore, R., Harrison, S., Behling, P., Selin, R., and F. Laarif, 1998.
Climate and biome simulations for the past 21,000 years Quaternary Science Reviews
17, 473-506.
Lambeck, K. and Bard, E., 2000. Sea-level change along the French Mediterranean coast
for the past 30,000 years. Earth and Planetary Science Letters 175, 203-222.
Marchetti, M., 2002. Environmental changes in the central Po Plain (northern Italy) due
to fluvial modifications and anthropogenic activities. Geomorphology 44, 361-373.
Marchi, E., Roth, G. and Siccardi, F., 1996. The Po: Centuries of River Training. Phys.
Chem. Earth, vol. 20, 475-478.
21 Morehead, M.D., Syvitski, J.P.M., Hutton, E.W.H. and Peckham, S.D. 2003. Modeling
the temporal variability in the flux of sediment from ungauged river basins. Global
and Planetary Change 39, 95-110.
Nelson, W.N., 1970. Hydrography, sediment dispersal, and recent historical development
of the Po river delta, Italy. In: Morgan, J.P. (eds) Deltaic Sedimentation; Modern and
Ancient Society of Economic Paleontologists and Mineralogists. Special Publication
No. 15, November 1970. Tulsa, Oklahoma, USA.
Peyron, A., Guiot, J., Cheddadi, R., Tarasov, P., Reille, M., Beaulieu, J.L. de, Bottema, S.
and V. Andrieu, 1998. Clomatic Reconstruction in Europe for 18,000 YR B.P. from
Pollen Data, Quaternary research 49, 183-196.
Ramrath, A., Zolitschka, B., Wulf, S. and Negendank, J.F.W., 1999. Late Pleistocene
climatic variations as recorded in two Italian maar lakes (Lago di Mezzano, Lago
Grande di Monticchio). Quaternary Science Reviews 18, 977-992.
Sbaffi, L., Wezel, F.C., Kallel, N., Paterne, M., Cacho, I., Ziveri, P. and Shackleton, N.,
2001. Response of the pelagic environment to palaeoclimatic changes in the central
Mediterranean Sea during the Late Quaternary. Marine Geology 178, 39-62.
Soldati, M., Corsini, A., Pasuto, A., 2003. Landslides and climate change in the Italian
Dolomites since the Late glacial. Catena 55, 141-161.
Surian, N. and Rinaldi, M., 2003. Morphological response to river engineering and
management in alluvial channels in Italy. Geomorphology 50, 307-326.
Syvitski, J.P.M., and Alcott, J.M. 1995. RIVER3: Simulation of water and sediment
river discharge from climate and drainage basin variables. Computers and Geoscience
21(1): 89-151.
22 Syvitski, J.P.M. and Kettner, A.J., accepted 2004. Predicting the Flux of Sediment to the
Coastal Zone: Application to the Lanyang watershed, northern Taiwan. Journal of
Coastal Research.
Syvitski, J.P.M. and Morehead, M.D., 1999. Estimating river-sediment discharge to the
ocean: application to the Eel margin, northern California. Marine Geology 154, 13-
28.
Syvitski, J.P.M., Peckham, S.D., Hilberman, R.D. and Mulder, T., 2003. Predicting the
terrestrial flux of sediment to the global ocean: A planetary perspective. Marine
Geology 162, 5-24.
Vose, R.S., Schmoyer, R.L., Steurer, P.M., Peterson, T.C., Heim, R., Karl, T.R. and J.
Eischeid, 1992. The Global Historical climatology Network: long-term monthly
temperature, precipitation, sea level pressure, and station pressure data.
ORNL/CDIAC-53, NDP-041. Carbon Dioxide Information Analysis Center, Oak
Ridge National Laboratory, Oak Ridge, Tennessee.
Vörösmarty, C.J., Meybeck, M., Fekete, B. and Sharma, K., 1997. The potential impact
of neo-Castorization on sediment transport by the global network of rivers. Human
Impact on Erosion and Sedimentation. IAHS Publication no. 245, 1997.
Watts, W.A., Allen, J.R.M. and Huntley, B., 1996. Vegetation history and palaeoclimate
of the last glacial period at Lago Grande di Monticchio, southern Italy. Quaternary
Science Reviews 15, 133-153.
Yu, Z., and Wright Jr., H.E., 2001. Response of interior North America to abrupt climate
oscillations in the North Atlantic region during the last deglaciation. Earth-Science
Reviews 52, 333-369.
23 Zonneveld. K.A.F., 1996. Palaeoclimatic reconstruction of the last deglaciation (18-8 ka
B.P.) in the Adriatic Sea region; a land-sea correlation based on palynological
evidence. Palaeogeography, Palaeoclimatology, Palaeoecology 122, 89-106.
24 Figure captions
Figure 1.
Overview of study area.
Figure 2.
Po drainage basin during the LGM (red) and today (yellow). The blue line represents the maximum extension of the late-Würmian ice-sheet during the LGM.
Figure 3.
12yr daily discharge events. Grey columns represent the measured discharge events from the Ufficio Idrografico del Magistrato per il Po, 1990-2001, black columns are the modeled discharge events.
Figure 4.
Monthly discharge distribution of 84yrs of measured data (solid grey line) versus modeled (solid black line). The dotted lines are representing the variation (plus or minus the standard deviation). Measured field is obtained from:
a) 1918-1979 http://espejo.unesco.org.uy/part%604/6_europa/italy/6iy`po_at_pontelagoscuro.htm.
b) 1980-1987 ANNALI IDROGRAFICI
c) 1989 – 2002 Ufficio Idrografico del Magistrato per il Po (Parma)
Figure 5.
25 17yrs of measured monthly suspended sediment load (∆) and river discharge from
ANNALI IDROGRAFICI. Superimposed on the plot (+) are a similar number of simulated values from the HydroTrend model.
Figure 6.
HydroTrend input data.
(Chronozones from Soldati et al., 2003)
Figure 7.
HydroTrend output table
Figure 8.
The influence of glacier ablation on total discharge (A) and suspended sediment load (B).
A) The solid line represents the total average daily discharge (m3s-1). The dotted line
represents the average daily discharge from glacier ablation.
B) Daily average suspended sediment load increases significantly during glacier ablation.
Figure 9.
Grey line represents the yearly ratio of bedload over suspended sediment over time. The
black line indicates the 50years running average of this ratio over time. Bedload shows a
more direct response to glacier melt than suspended load.
Figure 10.
26 Relation between yearly discharge and changing drainage basin of the Po River.
Figure 11.
The red dots are illustrating maximum peak discharges for the Holocene (10–0 cal. kyr
B.P.). Blue plus symbols are illustrating maximum peak discharges for the Pleistocene
(21–10 cal. kyr B.P.). The green circle is an example of what the impact of those events
is. 9.58 Mt suspended sediment load was carried to the Adriatic Sea during one day. That
is 12.5% of the yearly sediment going to the ocean.
Figure 12.
100 yr daily Sediment concentration (Cs) versus discharge (Q) (1) and the Suspended
sediment load (Qs) versus Bedload (Qb) (2). For four different time steps (A / D).
Table 1.
Model input parameters for present day simulation of the Po River.
Table 2.
Characteristics of the Po River, measured values against 12yr daily simulations.
a) Nelson, 1970
b) Cattaneo et al., 2003
c) Friend et al., 2002
Table 3.
27 Regression coefficients on α4 α5 k T Qs = α 3 A R e
Based on Syvitski en al., 2003.
Table 4.
Simulated characteristics of four time periods each based on 100yr daily simulations.
28 Figure 1.
Figure 2.
Figure 3.
2000 1912 1749 1643
1600 1532
1200
800 635 534 Number of events of Number 400 236 237 69 69 39 34 30 15 8 3 2 2 3 0 2 3 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 3 4 5 6 7 8 9 0 1 ------1 1 0 - - 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 1 2 3 4 6 7 8 0 0 9 0 Discharge ranges (m3/s) 1
Figure 4.
3500
3000
2500
2000 /s) 3
1500 Q (m
1000
500
0
l r r y y h i y e y t r r r r r l s e e e e c a n u b a a r p u u b b b u u a A M J g r J m to m m n u e a b M e c e A t v c J e O F p o e e Time (Months) N D S
Figure 5.
10000
1000
100 Suspended sediment discharge, Qs (kg/s) Qs discharge, sediment Suspended
10 100 1000 10000 Discharge, Q (m3/s)
Figure 6.
Figure 7. Figure 8.
Figure 9.
Figure 10.
200 180 )
-1 160 yr 3 140 m 9 120 10 * 100 80 60 40 Discharge ( 20 0 60 80 100 120 140 160 180 200 3 2 Area (* 10 km )
Figure 11.
Figure 12. Table 1.
Parameter Annual temperature (°C) 19.15 ± 0.41 Annual precipitation (m) 0.93 ± 0.13 Monthly parameters temperature (°C) precipitation (mm) January 3.25 ± 0.7 45.09 ± 33.8 February 5.0 ± 1.8 40.9 ± 29.2 March 9.8 ± 1.9 69.2 ± 34.9 April 13.2 ± 0.8 84.9 ± 56.6 May 18.8 ± 0.7 98.8 ± 53.6 June 22.4 ± 1.1 71.3 ± 24.2 July 25.9 ± 1.2 49.4 ± 31.5 August 25.1 ± 1.1 67.1 ± 32.8 September 19.6 ± 1.8 52.8 ± 41.5 October 14.4 ± 0.7 95.3 ± 55.1 November 8.0 ± 1.4 51.7 ± 49.6 December 3.5 ± 0.8 46.7 ± 28.6
Total drainage basin above the reservoirs (km2) 16335 Mean reservoirs volume (km3) 24.6 Glacier ELA line (m) 3270 River length (km) 650 Drainage area (km2) 74,259 Maximum elevation (m) 4,753
Table 2.
Measured at apex Modeled (Pontela-goscuro) Long-term average discharge (m3 s-1) a 1,500 1,480 Last century floods (m3 s-1) 10,300 10,800 Average suspended sediment load (Mt yr-1) b 15 17 Peak suspended sediment load (Mt yr-1) c 35 39 Average bedload (t yr-1) --- 4.0 x 105 Sediment yield (t km-2 yr-1) b 201 207 Table 3. Global sector α3 α4 α5 k
Temperate N (T > 0°C, lat > 30°N) 6.1 × 10-5 0.55 1.12 0.07
Table 4. Upper Younger Bølling Würm Subatlantic Dryas LGM Area (km2) 7.4 * 104 10.7 * 104 19.4 * 104 19.4 * 104 Average discharge (m3 s-1) 1557 4807 4684 3227 Peak discharge (m3 s-1) 9019 1.476 * 104 2.021 * 104 1.275 * 104 Average glacier ablation discharge (% of 0 52.9 30.4 5.5 total discharge) Average suspended sediment load (kg s-1) 588.3 1749 1472 1020 Peak suspended sediment load (kg s-1) 1.296 * 105 50655 1.322 * 105 1.82 * 105 Average bedload (kg s-1) 16.19 44.28 43.67 31.35 Peak bedload (kg s-1) 80.1 124.81 165.6 109.4 Average sediment yield (t km-2 yr-1) 250 512 238 156