|
Mediterranean, Tropical and Monsoon Regions |
1 Institute of Earth Sciences and Department of Geography, Hebrew University, Jerusalem, 91904 Israel
2 Department of Geology, Central Washington University, Ellensburg, WA 98926, USA
3 Arizona Laboratory for Palaeohydrological and Hydroclimatological Analysis, Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA
Data about the magnitude and time of occurrence of palaeofloods from the lower Colorado River basin enable us to test two long-standing hypotheses which have affected many studies and applications in the field of flood hydrology. The two hypotheses are (a) the existence of an upper boundary to flood magnitudes and whether there is a possibility of determining it from the existing data, and (b) the random occurrence versus clustering of the large floods through time.
Earlier observations on regional flood envelope curves indicated the existence of an upper limit for flood magnitudes, but these studies limited their conclusions because of the short length of the systematic gauged data. This limitation is overcome here because palaeoflood data cover a much longer period of observation. Palaeoflood studies provide information about the largest individual floods experienced in many rivers in a specific region occurring over the last millennia. In the southwestern US, this information demonstrates that, even when the length of observational data increases to centuries and millennia, there is no change in the stabilized, regional envelope curves constructed from gauged and historical flood records. This pattern supports the hypothesis of an upper limit to flood magnitudes and points to a method for testing this hypothesis in other regions. There are surprising similarities between the envelope curve of the palaeoflood data and the envelope curve for the gauged and historical data in the lower Colorado River basin. These similarities indicate that in regions of the world where flood data is sparse envelop curves based on palaeoflood studies can provide basic data for engineering design purposes and other hydrological applications.
The random occurrence of large floods in time is tested by constructing chronologies for the largest palaeofloods in several basins in the lower Colorado River basin. These chronologies indicate a clustering of the large floods in specific time periods. The similarity between the various time periods characterized by high- and low-flooding and other palaeoclimatic indicators from the southwestern United States seems best explained by a climatic control on flood frequency over the last 5000 years.
