MODELING SPATIAL AND TEMPORAL SUCCESSION IN THE ATCHAFALAYA/TERREBONNE MARSH/ESTUARINE COMPLEX IN SOUTH LOUISIANA

Abstract A spatial simulation model was constructed to help understand the historical changes in the Atchafalaya/Terrebonne marsh/estuarine complex in south Louisiana and to project impacts of proposed human modifications. The model consists of almost 3,000 interconnected cells each representing 1 km 2. Each cell in the model contains a dynamic, nonlinear, simulation model. Variables include water volume and flow, sediment and salt concentrations, organic standing crop, and productivity. The balance between sediment deposition and erosion as influenced by these variables in this rapidly subsiding area is particularly critical to habitat succession and the productivity of the area. Primary input data for the model were detailed, digitized habitat maps prepared by the U.S. Fish and Wildlife Service for 1956 and 1978. In addition, long time series of field measurements are available for some variables. The data base assembled for the model includes annual changes in environmental forcing functions, such as river discharge, and human modifications, such as canals and levees. At present, we are in the preliminary stages of running and calibrating the model. Our current results mimic the spatial seasonal patterns of sediment, salinity and water flow reasonably well. Starting with the 1956 initial conditions the model correctly predicts the habitat type in 1978 of 68% of the almost 3000 cells. Changes in salinity zones were accurately predicted but the model predicts higher land loss rates in the fresh zone than those actually observed and does not do well at predicting the growth of the Atchafalaya delta. The former problem may have to do with the prevalence of floating marsh in the fresh zone, which the model does not adequately consider at present. This paper discusses: (1) the structure of the model; (2) the spatial data base necessary to run the model; (3) some preliminary results concerning the temporal and spatial distributions of some of the state variables; and (4) future directions for the model. In general, this approach seems to be applicable to modeling spatial ecosystem dynamics, but the size and computational complexity of the model makes calibration and verification difficult.