科研之友
微信
新浪微博
Facebook
分享链接
摘要
Saltwater intrusion has been recognized as a major driver of ecological regime shifts in coastal wetlands, but few studies have incorporated long-term salinity modelling data into the understanding of vegetation dynamics. This study uniquely combined artificial neural network (ANN) modelling and time series of aerial photographs to elucidate saltwater intrusion and mangrove encroachment into the freshwater floodplain of the Loxahatchee River over a period of six decades. Three ANN models were developed to simulate river salinity on a daily time step using freshwater inflow, rainfall and tide as inputs. With 8years of measured data for training and testing, the ANN models demonstrated comparable or superior model performance for salinity simulation to their hydrodynamic counterparts. Modelling of historical salinity (1948-2011) using the ANNs indicated that the intensity of saltwater intrusion clearly correlated with watershed hydrology, which, in turn, was linked to historical watershed alterations and regional rainfall variability. Interpretation of the 1940, 1953, 1964, 1979, 1985 and 1995 aerial photographs revealed the progressive encroachment of mangroves and displacement of bald cypress and freshwater communities in the floodplain. The spatial extent of mangrove encroachment was related to the increasing mortality of bald cypress under the chronic effect of cumulative salinity exposure. The ecohydrological controls over vegetation changes in the Loxahatchee River support the ongoing ecosystem restoration programme, which aims to achieve incremental freshwater flow targets to protect and restore the floodplain ecosystem.
- 出版日期2015-3
