2nd UF Water Institute Symposium Abstract

   
Submitter's Name Zuzanna Zajac
Session Name Poster Session: Hydrologic, Biogeochemical and Ecological Processes 2
Category Hydrologic, biogeochemical and ecological processes
Poster Number 246
 
Author(s) Zuzanna Zajac,  University of Florida
  Rafael Munoz-Carpena,  University of Florida
   
  Global Uncertainty and Sensitivity Analysis of Spatially Distributed Hydrological Model, Regional Simulation Model (RSM), to spatially distributed factors.
   
  This research addresses two aspects of uncertainty assessment in spatially distributed modeling: uncertainty analysis (UA), described as propagation of uncertainty from spatially distributed input factors on model outputs; and sensitivity analysis (SA) defined as assessment of relative importance of spatially distributed factors on the model output variance. An evaluation framework for spatially distributed models is proposed based on a combination of sequential Gaussian simulation (sGs) and the global, variance-based, SA method of Sobol to quantify model output uncertainty together with the corresponding sensitivity measures. The framework is independent of model assumptions; it explores the whole space of input factors and provides measures of factor’s importance (first-order effects) and their interactions (higher-order effects). A spatially distributed hydrological model (Regional Simulation Model, RSM), applied to a site in South Florida (Water Conservation Area-2A, WCA-2A), is used as a benchmark for the study. The model domain is spatially represented by triangular elements (average size of 1.1 km2). High resolution land elevation measurements obtained by the USGS' Airborne Height Finder survey are used in the study. The original survey data, together with smaller density subsets drawn from this data are used for generating equiprobable maps of effective land elevation factor values via sGs. These alternative realizations are sampled pseudo-randomly and used as inputs for model runs. In this way, uncertainty regarding a spatial representation of the elevation surface is transferred into uncertainty of model outputs. The results show that below a threshold of data density, uncertainty of model outputs increases with a decrease of density of elevation data. Similar pattern is observed for the relative importance of sensitivity indexes of the land elevation factor. Therefore, reduced data density of land elevation could be used without significantly compromising the certainty of RSM predictions and the subsequent decision making process for the specific WCA-2A conditions.