On the Downscaling of Meteorological Fields Using Recurrent Networks for Modelling the Water Balance in a Meso-Scale Catchment Area of Saxony, Germany

Abstract

In this study, recurrent networks to downscale meteorological fields of the ERA-40 re-analysis dataset with focus on the meso-scale water balance were investigated. Therefore two types of recurrent neural networks were used. The first approach is a coupling between a recurrent neural network and a distributed watershed model and the second a nonlinear autoregressive with exogenous inputs (NARX) network, which directly predicted the component of the water balance. The approaches were deployed for a meso-scale catchment area in the Free State of Saxony, Germany. The results show that the coupled approach did not perform as well as the NARX network. But the meteorological output of the coupled approach already reaches an adequate quality. However the coupled model generates as input for the watershed model insufficient daily precipitation sums and not enough wet days were predicted. Hence the long-term annual cycle of the water balance could not be preserved with acceptable quality in contrary to the NARX approach. The residual storage change term indicates physical restrictions of the plausibility of the neural networks, whereas the physically based correlations among the components of the water balance were preserved more accurately by the coupled approach.

Application of Synthetic Meteorological Time Series in BROOK90: A Case Study for the Tharandt Forest in Saxony, Germany

“…While LARS reaches a correlation of 0.95 ERM achieved fairly 0.89. However both results are outstanding compared to other investigations [37] where precipitation, because of its supposed randomness in time and space at this scale, is the most defficile element to simulate. The significant better performance of LARS can be explained by the usage of a semi empiric distribution for the modelling of precipitation, which obviously estimates the daily amount more precise.…”

“…ERM) could be extended from a single site to a raster-based multi site weather generator, which might be coupled with a cascade model for the downscaling from daily to 5 min time series [39]. However, this would require a change for the hydrological modelling to a raster based model like WaSim-ETH [37].…”

An artificial neural network-based snow cover predictive modeling in the higher Himalayas

“…The study conducted by Shen and Chang (2013) used NARX to forecast multistep-ahead inundation depth in an inundation area and gave the better result than feedforward time-delay and an online feedback configuration of NARX networks. Similarly, NARX gave a superior result on downscaling of meteorological fields meso-scale water balance modeling in comparison to the coupling between recurrent neural network and a distributed water shed model (Kronenberg et al 2013).…”