Ανάλυση ευαισθησίας παραμέτρων του υδρολογικού μοντέλου HEC-HMS και εφαρμογή στη λεκάνη του ρέματος Ραφήας

Abstract

Αντικείμενο της παρούσας μεταπτυχιακής εργασίας είναι η υδρολογική προσομοίωση μεμονωμένων γεγονότων βροχής με τη βοήθεια του υδρολογικού μοντέλου HEC – HMS, και η ανάλυση ευαισθησίας των υδρολογικών παραμέτρων αυτού. Η υδρολογική προσομοίωση έγινε με δύο διαφορετικές μεθόδους υπολογισμού της άμεσης απορροής, του Μοναδιαίου Υδρογραφήματος της Soil Conservation Service (SCS) και του Συνθετικού Μοναδιαίου Υδρογραφήματος Snyder. Η μελέτη πραγματοποίηθηκε στην περιοχή της λεκάνης απορροής του ρέματος Ραφήνας στο νομό Αττικής, με έκταση 123 km2. Η λεκάνη απορροής του ρέματος Ραφήνας παρουσιάζει έντονο ενδιαφέρον για πλημμυρική διερεύνηση λόγω της έντονης οικοδομικής δραστηριότητας στην ευρύτερη περιοχή τα τελευταία χρόνια, της ραγδαίας αύξησης του πληθυσμού στους κύριους οικισμούς, τη συχνότητα φαινομένων πυρκαγιάς, όσο και για τις ειδικές γεωμορφολογικές και γεωλογικές συνθήκες που επικρατούν στην περιοχή. Η συνεχής υδρομετεωρολογική παρακολούθηση της περιοχής από το Υδρολογικό Παρατηρητήριο Αθηνών (HOA) και το Εθνικό Αστεροσκοπείο Αθηνών (ΝΟΑ), παρείχαν τα απαραίτητα βροχομετρικά και σταθμημετρικά δεδομένα, σε βήμα 10 min, τα οποία αποτέλεσαν δεδομένα εισόδου στην προσομοίωση και επέτρεψαν στη βαθμονόμηση του υδρολογικού μοντέλου. Επιπλέον δεδομένα αντλήθηκαν από τη χρήση χωρικών πληροφοριών (ψηφιακό μοντέλο εδάφους, χάρτης κάλυψης γης και γεωλογικός χάρτης) και η επεξεργασία τους από το πρόγραμμα ArcGIS 9.3, όπου έδωσε τη δυνατότητα εισαγωγής και αξιοποίησης των γεωμορφολογικών παραμέτρων στη διαδικασία της προσομοίωσης. Στο κυρίως μέρος της, εργασίας η λεκάνη απορροής προσομοιώθηκε με πέντε (5) υπολεκάνες, στις οποίες υπήρχαν διαθέσιμα σταθμημετρικά δεδομένα στις εξόδους σε τρεις από αυτές (Ντράφι, Ραφήνα - ανάντη και Ραφήνα). Επιλέχθησαν τρία ξεχωριστά γεγονότα βροχής (Δεκέμβριος 2009, Φεβρουάριος 2011 και Φεβρουάριος 2013), όπου εφαρμόζοντας τις δύο μεθόδους υπολογισμού άμεσης απορροής SCS και Snyder, υπολογίστηκαν τα υδρογραφήματα εξόδου. Ακολούθησε η διαδικασία της βαθμονόμησης, έως του σημείου ικανοποιητικής προσέγγισης των κριτηρίων σύγκλισης ( R, PEV και PEPF) και η επαλήθευση του υδρολογικού μοντέλου με δύο (2) επεισόδια βροχόπτωσης (Φεβρουάριος και Δεκέμβριος 2012). Τέλος, για την αποτίμηση της βαρύτητας των υδρολογικών παραμέτρων στη λειτουργία του μοντέλου, επιχειρήθηκε μέσω της μεθόδου διακύμανσης μιας παραμέτρου τη φορά (OAT), η ανάλυση ευαισθησίας τεσσάρων (4) υδρολογικών παραμέτρων (CN, Ia, Tlag, παραμέτρου εξασθένησης x της μεθόδου Muskingum) και επίδρασή τους στην παροχή αιχμής και συνολικό όγκο απορροής για τα επεισόδια βροχής Φεβρουαρίου 2011 και Φεβρουαρίου 2013.

Over the past ten years, Europe has suffered from more than 175 major floods, causing deaths, the displacement of people and large economic losses. Projections reveal an increase to the intensity and frequency of floods. Most of the observed upward trend in flood damage can be attributed to socio-economic factors, such as increases in population, wealth and urbanisation in flood-prone areas, and to land use changes, such as deforestation and loss of wetlands and natural floodplain storage. Floods are natural phenomena which cannot be prevented but through the right measures we can reduce their likelihood and limit their impacts. In addition to economic and social damage, floods can have severe environmental consequences, for example when industrial installations are inundated or wetland areas destroyed. The coming decades are likely to see a higher flood risk and greater economic damage. Having regard of the above, the European Commission, set into force the Directive 2007/60/EC on the assessment and management of flood risks. Its aim is to reduce and manage the risks that floods pose to human health, the environment, cultural heritage and economic activity. The Directive shall be carried out in coordination with the Water Framework Directive 2000/60/EC. The scope of this master thesis is to probe the importance of specific simulated hydrological parameters which represent a part of the components of the water cycle. This was attained through a sensitivity analysis based on the OFAT method. To assess the hydrological response of the watershed, a HEC – HMS, semi-distributed hydrological model has been applied to simulate the precipitation runoff precesses. The Rafina watershed has been selected as the case study of the current master thesis and it is located in the eastern part of the Attica River Basin District (GR06), according to the National WFD nomenclature. Administratively belongs to the Prefecture of Attica with a population of about 86.000 people and covers an area of 123 km2. The increasing interest of the society for residency, the new infrastructure works (International Airport, a dense grid of new highways),the repeated forest fires in the upstream areas and the great industrial expansion through the last ten years, make this area suitable for this research. A demonstration project aiming to the development of an integrated Decision Support System (DSS) for both flash floods and forest fires risk assessment and management is implemented in the area. The final model will be online available to key stakeholders and relevant authorities (local and national) for the combined, effective and efficient floods and fires risk management. The project, FLIRE, as a part of LIFE+ programme 2007-2013 will contribute to the implementation of European environmental policy and the development of innovative policy ideas, technologies, methods and instruments. It will also help monitor pressures on our environment. The necessary input of meteorologic and observed flow data have been acquired by the Hydrological Observatory of Athens (HOA), developed and operated by the Laboratory of Hydrology and Water Resources of the School of Civil Engineering of the National Technical University of Athens (NTUA). It is the result of further development of the hydrometeorological network METEONET and provides access to reliable hydrological and meteorological datasets of the area since 2005. It consists of 13 meteorological and 4 flow measuring stations in the greater Athens area, recording a variety of environmental parameters of hydrometeorological interest, presented and distributed online in a friendly user interface. These parameters include rainfall, temperature, water levels, relative humidity, evaporation, solar radiation, sunshine duration, wind direction etc. Additional rainfall data have been acquired from National Observatory of Athens (NOA). For the correct and accurate use of input data throughout the hydrological simulation, there must have been accomplished a series of necessary geospatial calculations with the HEC – GeoHMS, an ArcGIS extension toolbox, so that the geomorphological output data, will demonstrate the physical characteristics of the watershed. As a result, and according to the available observed flow measurement data, Rafina watershed has been divided into five (5) smaller subbasins. The HEC - HMS hydrological simulation that applied was chosen to be an event – based one. For this reason, three (3) different rainfall events were chosen ( December 2009, February 2011 and February 2013) to simulate the watershed hydrological response. The simulation process included some assumptions based on cumulative experience from previous studies in the area. As the procedure dectates, a basin and a meteorological model have been imported in the HEC – HMS algorithm, adopting Thiessen polygons spatial aggregation, alongside with time control specifications and time series data. The representation of the watershed hydrologic processes, was based in the formulation of computation of rainfall losses, transform method, baseflow method and channel flow routing. The choice of the methods was determined by the shortage of reliable data of the area. As a result, the SCS Curve Number method applied for the computation of rainfall losses, and Muskingum method for channel flow routing calculations. The transform method that was applied in this thesis was both SCS and Snyder Unit synthetic hydrographs. The simulation was run for three (3) rainfall events (9-11/12/2009 , 3-4/2/2011 and 22/2/2013) and the output of the simulated hydrograph is shown below.

Calibration is the process of the modeling, in which the user tunes the parameters of the simulation model, in such a way that the new output simulation data fit better with the observed ones. In this thesis tha calibration took place in two observed flow stations (Drafi and Rafina 1) due to the lack of observed flow data at Spata and Rafina 2 stations for the three rainfall events. The best results were accomplished by reducing CN curve number and Ia initial abstractions of the model as shown below.

Calibration criteria have been taken into consideration, adopting the R (Nash – Sutcliffe) efficiency coefficient, PEPF (percent error in peak flow) and PEV(percent error in volume) measures, which acted as indicators of acceptable simulation fitness.

Model validation is a process of testing model ability to simulate observed data other than those used for the calibration, with acceptable accuracy. During this process, calibrated model parameters are not subject to change, their values are kept constant. The quantitative measure of the match is again the degree of variation between computed and observed hydrographs. In the validation process adopted in this thesis, the input validation data was an average value of all hydrologic parameters applied to two (2) different rainfall events (6-7/2/2012 and 29-31/12/2012). Results are shown below only for the location of Drafi station, while on the other stations, results require further investigation.

Sensitivity analysis is the study of how the uncertainty in the output of a mathematical model or system (numerical or otherwise) can be apportioned to different sources of uncertainty in its inputs. (Saltelli, A. et al, 2008). In this thesis the hydrological parameters have been altered in equal intervals for each category specifically and separately. The simulation outputs that have been controlled over the sensitivity analysis are peak flow discharge and total outflow volume of the flow measurement station monitored. The validation results are shown below. Conlcuding, there was revealed the importance of the losses mechanisms in hydrologic models and their impact in calibration procedures. Their impacts are influential in both transform methods that examined. Data availability also raised as a crucial parameter in the whole modeling procedure providing the crucial required observed flow data for the calibration of the model. Further development of this thesis, could be an update in the structure of the model, including more components of the hydrologic cycle such as evapotransipration, detention, infiltration interception etc. , that didn’t simulated in this one. Investigation of the contribution of the groundwater aquifer in the hydrological behavior of the Rafina watershed and incoroporation would make the model more realistic.