*Result*: Improved Prediction of Managed Water Flow into Everglades National Park Using Empirical Dynamic Modeling.
*Further Information*
*Alteration of natural surface flow paths across South Florida has been detrimental to the environmental health and sustainability of the Everglades and surrounding ecosystems. As part of the Comprehensive Everglades Restoration Plan (CERP), predicting flows into Everglades National Park (ENP) is a central concern of effective management strategies. Management efforts have established weekly target flows into Everglades National Park through optimization of numerically intensive hydrological models. These target flows are focused specifically on flows across US Highway 41, also known as the Tamiami Trial. To aide in timely management assessments in response to current or predicted hydrologic conditions, the Tamiami Trail Flow Formula (TTFF) was developed previously to predict weekly target flows based on linear regression of six theorized flow drivers. It is known that these drivers exhibit nonlinear dynamics, suggesting that there is room for improvement in relation to the strictly linear TTFF. We used empirical dynamic modeling (EDM), a nonparametric modeling paradigm for forecasting and analyzing nonlinear time series, to show that prediction accuracy is improved when nonlinearity is accounted for. This method relies on weighted linear regressions that depend on specific environmental conditions or system states, i.e., in which the regression gives greater weight to input variables that have values that are more similar to the current state. Surprisingly, we found that only two of the six standard TTFF variables are required in the nonlinear weekly forecast model (upstream and downstream water levels), and thus the EDM model not only improves accuracy but also greatly simplifies hydrologic forecasting. [ABSTRACT FROM AUTHOR]
Copyright of Journal of Water Resources Planning & Management is the property of American Society of Civil Engineers and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)*