Global Sensitivity Analysis (GSA) with Hyper-Spectral Image (HSI) in the EFDC Water Quality Modelling

Abstract

Recently, with the rapid advances in computing power and water quality data collection, the development of sophisticated and complex water quality models has been accelerated, so that the number of parameters to be considered in the models also substantially increased. However, it is also challenging to calibrate all parameters of the model, thereby it is necessary to reduce the number of parameters to be an effective calibration. Sensitivity analysis (SA) would be an effective way to identify the most influential parameters for reducing errors between observations and model results. The main purpose of SA is to identify the most sensitive parameters on two different objectives (i.e., the output of model or errors between the model output and observation) by quantifying their influences on the two different objectives. However, there is a significant scaling-issue when output of water quality models directly compares with observations from waterbodies; a grid size of the water quality model ranges from 10m to 100m and observations usually are obtained from a tiny spot (i.e., point) from waterbodies.. Therefore, it is necessary to suggest a new approach to resolve the limitation of previous study above. In this study, we introduce a new approach on the SA of water quality model using airborne hyperspectral images (HSI) on waterbodies.