Multilayer Perceptron Neural Networks for Estimating Missing Rainfall Data

2021 ◽  
Author(s):  
Roya Narimani ◽  
Jun Changhyun
Keyword(s):  
Neural Networks ◽  
Time Series ◽  
Missing Data ◽  
Multilayer Perceptron ◽  
Data Science ◽  
Missing Values ◽  
Rain Gauge ◽  
Rainfall Data ◽  
Rainfall Time Series ◽  
Mlp Neural Networks

<p>The quality and completeness of rainfall data have always played an important role in time series analysis and prediction for future water-related disasters. It requires to estimate missing data correctly for better results of rainfall prediction with high accuracy. In recent years, multilayer perceptron (MLP) neural networks have been applied to solve stochastic problems in data science. This study suggests a novel approach for estimating missing rainfall data with MLP neural networks. For this purpose, a mathematical model was created to analyze and predict the time series of daily rainfall data from 2003 to 2017 at six rain gauge stations in Seoul, Korea. Here, rainfall data with missing values during 20 days of time periods was considered for reconstruction of missing data at one specific rain gauge station from complete rainfall data records at five different stations. They were divided into training, validation, and testing datasets with a percentage of 70%, 15%, and 15%, respectively. This study investigates an effect of changes in data periods considered in MLP neural networks and it indicates that rainfall time series for a longer time period play a more effective role in rainfall data reconstruction.</p>

SLICES OF MULTIFRACTAL MEASURES AND APPLICATIONS TO RAINFALL DISTRIBUTIONS

Fractals ◽  
2000 ◽  
Vol 08 (04) ◽  
pp. 337-348 ◽  
Author(s):  
BIRGER LAMMERING
Keyword(s):  
Time Series ◽  
Field Data ◽  
Rainfall Data ◽  
Rainfall Time Series ◽  
Mathematical Ideas ◽  
Radar Images ◽  
The Usa ◽  
Self Similar ◽  
The Relationship ◽  
Plane Measure

We discuss the relationship between the multifractal functions of a plane measure and those of slices or sections of the measure with a line. Motivated by recent mathematical ideas about the relationship between measures and their slices, we formulate the "slice hypothesis," and consider the theoretical limitations of this hypothesis. We compute the multifractal functions of several standard self-similar and self-affine measures and their slices to examine the validity of the slice hypothesis. We are particularly interested in using the slice hypothesis to estimate multifractal properties of spatial rainfall fields by analyzing rainfall data representing slices of rainfall fields. We consider how rainfall time series at a fixed site and slices of composite radar images can be used for this purpose, testing this on field data from a radar composite in the USA and on appropriate time series.

scholarly journals Recurrent Neural Networks for Multivariate Time Series with Missing Values

2018 ◽  
Vol 8 (1) ◽  
Author(s):  
Zhengping Che ◽  
Sanjay Purushotham ◽  
Kyunghyun Cho ◽  
David Sontag ◽  
Yan Liu
Keyword(s):  
Neural Networks ◽  
Time Series ◽  
Recurrent Neural Networks ◽  
Missing Values ◽  
Multivariate Time Series

scholarly journals Time-Series Causality with Missing Data

2021 ◽  
Vol 6 (1) ◽  
pp. 1-4
Author(s):  
Bo Yuan Chang ◽  
Mohamed A. Naiel ◽  
Steven Wardell ◽  
Stan Kleinikkink ◽  
John S. Zelek
Keyword(s):  
Time Series ◽  
Missing Data ◽  
Missing Values ◽  
Time Series Data ◽  
Multivariate Time Series ◽  
Gaussian Process Regression ◽  
Series Data ◽  
Causal Relationships ◽  
Sampled Data ◽  
The Past

Over the past years, researchers have proposed various methods to discover causal relationships among time-series data as well as algorithms to fill in missing entries in time-series data. Little to no work has been done in combining the two strategies for the purpose of learning causal relationships using unevenly sampled multivariate time-series data. In this paper, we examine how the causal parameters learnt from unevenly sampled data (with missing entries) deviates from the parameters learnt using the evenly sampled data (without missing entries). However, to obtain the causal relationship from a given time-series requires evenly sampled data, which suggests filling the missing data values before obtaining the causal parameters. Therefore, the proposed method is based on applying a Gaussian Process Regression (GPR) model for missing data recovery, followed by several pairwise Granger causality equations in Vector Autoregssive form to fit the recovered data and obtain the causal parameters. Experimental results show that the causal parameters generated by using GPR data filling offers much lower RMSE than the dummy model (fill with last seen entry) under all missing values percentage, suggesting that GPR data filling can better preserve the causal relationships when compared with dummy data filling, thus should be considered when dealing with unevenly sampled time-series causality learning.

Performance of Multi-Layer Feedforward Neural Networks to Predict Liver Transplantation Outcome

1996 ◽  
Vol 35 (01) ◽  
pp. 12-18 ◽  
Author(s):  
M. Subotin ◽  
W. Marsh ◽  
J. McMichael ◽  
J. J. Fung ◽  
I. Dvorchik
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Liver Transplantation ◽  
Missing Data ◽  
Network Performance ◽  
Missing Values ◽  
Feedforward Neural Networks ◽  
Linear Scaling ◽  
Data Set ◽  
New Approach

AbstractA novel multisolutional clustering and quantization (MCO) algorithm has been developed that provides a flexible way to preprocess data. It was tested whether it would impact the neural network’s performance favorably and whether the employment of the proposed algorithm would enable neural networks to handle missing data. This was assessed by comparing the performance of neural networks using a well-documented data set to predict outcome following liver transplantation. This new approach to data preprocessing leads to a statistically significant improvement in network performance when compared to simple linear scaling. The obtained results also showed that coding missing data as zeroes in combination with the MCO algorithm, leads to a significant improvement in neural network performance on a data set containing missing values in 59.4% of cases when compared to replacement of missing values with either series means or medians.

Validation of Climate Research Unit High Resolution Time-Series Rainfall Data over Three Source Region: Results of 52 Years

2013 ◽  
Vol 726-731 ◽  
pp. 3542-3546 ◽  
Author(s):  
Jonathan Arthur Quaye-Ballard ◽  
Ru An ◽  
Richard Ruan ◽  
Kwaku Amaning Adjei ◽  
Samuel Akorful-Andam
Keyword(s):  
Time Series ◽  
High Resolution ◽  
Source Region ◽  
Research Unit ◽  
Rain Gauge ◽  
Rainfall Data ◽  
Climate Research Unit ◽  
Resolution Time ◽  
Map Algebra ◽  
Climate Research

The purpose of this paper was to validate the rainfall data of Climate Research Unit high resolution Time-Series version 3.1 (CRU TS 3.1) with meteorological ground-based Rain Gauge (RG) measurements and determine the possibility of its integration with ground-measured rainfall. The research primarily advocates on the need for complementing ground-based datasets with CRU TS 3.1global datasets for sustainable studies in protecting the environment. The Source Region of the Yellow, Yangtse and Lancang Rivers (SRYYLR), China was taken as the study area. The data was validated by using the data from seventeen meteorological RG stations at SRYYLR. Statistical technique based on Linear Regression (LR), Cumulative Residual Series Analysis (CRSA) and Geo-Spatial techniques based on batch processing, cell statistics, map algebra, re-sampling, extraction by mask, geo-statistical interpolation and profiling along transects by interpolation of a line were used. The study revealed that although CRU TS 3.1 datasets are underestimated compared to the RG datasets, they can be efficiently and effectively be used for rainfall trend analysis with 90% level of confidence because of the analyses by different techniques revealed similar profile trends.

scholarly journals The sensitivity of catchment runoff models to rainfall data at different spatial scales

2000 ◽  
Vol 4 (4) ◽  
pp. 653-667 ◽  
Author(s):  
V. A. Bell ◽  
R. J. Moore
Keyword(s):  
Time Series ◽  
Spatial Scales ◽  
Rainfall Data ◽  
Scale Model ◽  
Distributed Model ◽  
Rainfall Time Series ◽  
Time Step ◽  
Stratiform Rain ◽  
Convective Rain ◽  
Catchment Runoff

Abstract. The sensitivity of catchment runoff models to rainfall is investigated at a variety of spatial scales using data from a dense raingauge network and weather radar. These data form part of the HYREX (HYdrological Radar EXperiment) dataset. They encompass records from 49 raingauges over the 135 km2 Brue catchment in south-west England together with 2 and 5 km grid-square radar data. Separate rainfall time-series for the radar and raingauge data are constructed on 2, 5 and 10 km grids, and as catchment average values, at a 15 minute time-step. The sensitivity of the catchment runoff models to these grid scales of input data is evaluated on selected convective and stratiform rainfall events. Each rainfall time-series is used to produce an ensemble of modelled hydrographs in order to investigate this sensitivity. The distributed model is shown to be sensitive to the locations of the raingauges within the catchment and hence to the spatial variability of rainfall over the catchment. Runoff sensitivity is strongest during convective rainfall when a broader spread of modelled hydrographs results, with twice the variability of that arising from stratiform rain. Sensitivity to rainfall data and model resolution is explored and, surprisingly, best performance is obtained using a lower resolution of rainfall data and model. Results from the distributed catchment model, the Simple Grid Model, are compared with those obtained from a lumped model, the PDM. Performance from the distributed model is found to be only marginally better during stratiform rain (R2 of 0.922 compared to 0.911) but significantly better during convective rain (R2 of 0.953 compared to 0.909). The improved performance from the distributed model can, in part, be accredited to the excellence of the dense raingauge network which would not be the norm for operational flood warning systems. In the final part of the paper, the effect of rainfall resolution on the performance of the 2 km distributed model is explored. The need to recalibrate the model for use with rainfall data of a given resolution, particularly for periods of convective rain, is highlighted. Again, best performance is obtained using lower resolution rainfall data. This is interpreted as evidence for the need to improve the distributed model structure to make better use of the higher resolution information on rainfall and topographic controls on runoff. Degrading the resolution of rainfall data, model or both to achieve the smoothing apparently needed is not seen as wholly appropriate. Keywords: rainfall, runoff, sensitivity, scale, model, flood

Sign in / Sign up

Export Citation Format

Share Document