MODELING SUSPENDED SEDIMENT TRANSPORT AND ASSESSING THE IMPACTS OF CLIMATE CHANGE IN A KARSTIC MEDITERRANEAN WATERSHED SOFIA NERANTZAKI TECHNICAL UNIVERSITY OF CRETE
Mediterranean features: Karstic Springs Greek Speleological Society: Chania 1110 m below surface (1600 -490 m) Liontari sinkhole
Data scarcity and flash floods Limitations: infrequency of intense rainfall, adverse field conditions
Koiliaris River Basin •The permanent flow of the basin is supplied by the karstic springs of Stilos (mean suspended sediment concentration: 4mg/L) • Keramianos: a temporary tributary responsible for the flash flood events at the hydrometric station H1 and for the majority of sediment mass transported at the basin exit Koiliaris watershed • Keramianos sub-basin: friable schist, invasive agricultural practices and overgrazing the soil is erodible and easily transported Dry Season H2 H1 H3 First Flood
Monitoring system • Water stage and turbidity are recorded every five minutes. Water stage is converted to flow with the use of a site specific rating curve. SedTrap – “CYBERSENSORS” • Flow proportionate sampling: interval between 20 and 80 m 3 /s is divided into 6 volumetrically equal classes of 10 m 3 /s each; the Suspendent Sediment Node “ Sediment Trap ”. pumping rate is augmented a class, every time the flow surpasses Consists of: + a class. • 1x Control Unit “ Libellium Waspmote ” , • 1x Radio Module (Zigbee), • 1x Peristaltic Pump, • 1x Stage Recorder • 1x Turbidity Sensor. Energy Independent.
Modified SWAT Model Two-part reservoir karst model •Augmented version of the SWAT model: simulation of the contribution of the extended karst to the spring discharge, accounting for the variability of the discharge recession due to two karst formations (Nikolaidis et al., 2013) Methodology •The precipitation in the karstic area is directed to deep groundwater after SWAT simulates surface hydrologic processes •The deep groundwater flow from SWAT in the karstic area that could be related to a specific spring is aggregated on a daily basis and becomes the input flow to a two part reservoir karst model Karst Model Hydrology Simulation •Input: Daily rainfall and temperature data 1973 -2014, 5 meteorological stations •Goodness of fit (daily records ): •NSE = 0.62 (> 0.5) • PBIAS = 22.3% (< 25%) • RSR = 0.62 (< 0.7) • Annual average outflow of Koiliaris River Basin was estimated to be 621 mm/yr (131 Mm 3 /yr)
Suspended Sediment Concentration (SSC) Simulation Equations d V C 1 1 a Q C Q C 1 in,1 in,1 1 1 d t d V C 2 2 1 a Q C a Q C Q C 1 in,1 in,1 2 1 1 2 2 d t •The input concentration Cin,1 is determined from the SWAT model, as a fraction of the sediment yield from the HRUs which contribute to the karstic flow Concentration of mixed flow Q * C Q * C karst karst surface surf flow C model Q Q karst surf flow •SSC simulation at (a) H1, (b) H3, (c) Stilos Springs • 30% of the sediment yield of the extended karst sub-basins is driven into the karst system through sinkholes •During a flood event, the Keramianos tributary transports 95% of the sediment mass
SSC Modeling Results and Erosion •( i) Keramianos (mean erosion rate):0.97 t/ha/yr. (ii) Wet year: 1.6 t/ha/yr. (iii) Dry year: 0.55 t/ha/yr • Keramianos mean sediment export is 3000 t/yr and contributes 67% of the mean annual sediment export of the Koiliaris watershed (4500 t/yr). On a wet year the sub-basin of Keramianos is responsible for the 63% of it and on a dry year Keramianos is responsible for the 70% of the total sediment export due to the high erodibility of its sub-basin.
Climate Scenarios •Set of IPCC “A1B” climate change scenarios: major decreases in surface flow (69.6%) and in the flow of the springs (76.5%) between the 2010-2049 and 2050-2090 time periods •Sediment export is also decreasing (54.5%)
Future Steps Input Uncertainty and Impact on Results Input Time Series - Quantification of Uncertainty Stochastic Generation for flow and sediment concentration output KARST SWAT Cumulative distribution function of a random variable R showing the probability of any MONTE CARLO SIMULATION observed value of R being less than or equal Deterministic simulation and the Monte-Carlo to a given value r. expected averages (and standard deviation) Values generation from known probability simulations for flow (the same stands for distributions for precipitation and temperature. sediment concentration) This section was implemented with the state scholarship IKY (SSF) funded by the Act "Strengthening the human research potential by implementing doctoral research" from funds of the OP "Human Resources Development, Education and Lifelong Learning" 2014-2020 co- financed by the European social Fund (ESF) and the Greek government
Thank you!
Recommend
More recommend
