Input variable selection for hydrological predictions in ungauged catchments: with or without clustering

Date

2018-04-08

Author

Doğulu, Nilay
Batmaz, İnci
Kentel Erdoğan, Elçin

Metadata

Show full item record

Item Usage Stats

148
views

0
downloads

A key step in data-driven environmental modelling, including for hydrological purposes, is input variable selection (IVS) to ensure that the least number of variables with minimum redundancy are used to characterize the inherent relationship between inputs and outputs. Hydrological predictions in ungauged catchments is one such area where the information on influential predictors of runoff signatures guides in understanding dominant controls of meaningful information transfer from gauged to ungauged locations (i.e. regionalization). This understanding is valuable especially for the analysis of hydrological similarity among ungauged catchments, e.g. to identify reference (donor) catchment(s). Large-sample hydrology can help to gain useful insights on how these significant predictors change over different runoff signatures representing particular hydrological conditions as well as within different groups of similar catchments. This study explores the added value of clustering for input variable selection in the case of catchments across continental USA using the CAMELS dataset (Addor et al., 2017). We employ the method of k-means clustering on the input space consisting of topography, soil, geology, vegetation and climate attributes as a way to deal with heterogeneity and complexity in hydrological processes, and thus, aiming to identify similar groups of catchments. The input variables (for predicting selected hydrological attributes) are determined by three IVS filter algorithms (partial mutual information, partial correlation input selection, and iterative input selection), then evaluated and compared for among different clusters and when no clustering method is applied. We present and discuss the results for three hydrological attributes – 95% flow percentile (low flows), mean daily discharge (medium flows), and 5% flow percentile (high flows) – in order to account for variability in hydrological conditions, and to investigate the dominant catchment and/or climate characteristics controlling low/medium/high flow predictability at ungauged locations. The findings of our study have implications for reference (donor) catchment selection and hydrological model independent regionalization (aka hydrostatistical or data-driven methods), and are particularly relevant to understanding hydrological similarity and catchment classification in the absence of local runoff observations. Addor, N., Newman, A. J., Mizukami, N., & Clark, M. P. (2017) The CAMELS data set: catchment attributes and meteorology for large-sample studies. Hydrology and Earth System Sciences, 21(10), 5293-5313.

URI

https://hdl.handle.net/11511/79919
https://www.researchgate.net/publication/324068043_Input_variable_selection_for_hydrological_predictions_in_ungauged_catchments_with_or_without_clustering

Conference Name

EGU General Assembly, 8 - 13 April 2018

Collections

Department of Statistics, Conference / Seminar

Suggestions

OpenMETU
Core

Algorithm Overview and Design for Mixed Effects Models Koca, Burcu; Gökalp Yavuz, Fulya (2021-06-06) Linear Mixed Model (LMM) is an extended regression method that is used for longitudinal data which has repeated measures within the individual. It is natural to expect high correlation between these repeats over a period of time for the same individual. Since classical approaches may fail to cover these correlations, LMM handles this significant concern by introducing random effect terms in the model. Besides its flexible structure in terms of modeling, LMM has several application areas such as clinical tri...
Linear parameter varying control for autonomous systems: methods and application examples Çalış, Fatih; Schmidt, Klaus Verner; Department of Electrical and Electronics Engineering (2022-8-24) Linear parameter varying (LPV) systems are nonlinear systems which can be modelled as linear systems whose parameters change as a function of different "scheduling parameters". In other words, the dynamics of the LPV systems change during the operation hence they require a parameter dependent controller. Although classical gain-scheduling approaches satisfy some performance criteria for constant dynamics, they don't guarantee stability while the scheduling parameter is changing. On the other hand, H∞-norm b...
Parallel Minimum Norm Solution of Sparse Block Diagonal Column Overlapped Underdetermined Systems Torun, F. Sukru; Manguoğlu, Murat; Aykanat, Cevdet (2017-03-01) Underdetermined systems of equations in which the minimum norm solution needs to be computed arise in many applications, such as geophysics, signal processing, and biomedical engineering. In this article, we introduce a new parallel algorithm for obtaining the minimum 2-norm solution of an underdetermined system of equations. The proposed algorithm is based on the Balance scheme, which was originally developed for the parallel solution of banded linear systems. The proposed scheme assumes a generalized band...
Residual based Adaptive Unscented Kalman filter for satellite attitude estimation Söken, Halil Ersin (2012-12-01) Determining the process noise covariance matrix in Kalman filtering applications is a difficult task especially for estimation problems of the high-dimensional states where states like biases or system parameters are included. This study introduces a simplistic residual based adaptation method for the Unscented Kalman Filter (UKF), which is used for small satellite attitude estimation. For a satellite with gyros and magnetometers onboard, the proposed adaptive UKF algorithm estimates the attitude as well as...
NUMERICAL ANALYSIS AND TESTING OF A FULLY DISCRETE, DECOUPLED PENALTY-PROJECTION ALGORITHM FOR MHD IN ELSASSER VARIABLE AKBAŞ, MİNE; Kaya Merdan, Songül; MOHEBUJJAMAN, Muhammed; rebholz, leo (2016-01-01) We consider a fully discrete, efficient algorithm for magnetohydrodynamic (MHD) flow that is based on the Elsasser variable formulation and a timestepping scheme that decouples the MHD system but still provides unconditional stability with respect to the timestep. We prove stability and optimal convergence of the scheme, and also connect the scheme to one based on handling each decoupled system with a penalty-projection method. Numerical experiments are given which verify all predicted convergence rates of ...

Citation Formats

N. Doğulu, İ. Batmaz, and E. Kentel Erdoğan, “Input variable selection for hydrological predictions in ungauged catchments: with or without clustering,” presented at the EGU General Assembly, 8 - 13 April 2018, Vienna, Austria, 2018, Accessed: 00, 2021. [Online]. Available: https://hdl.handle.net/11511/79919.