Predicting Discharge and Sediment Flux of the Po River, Italy Since the Late Glacial
Total Page:16
File Type:pdf, Size:1020Kb
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).