scholarly journals Subdaily Rainfall Estimation through Daily Rainfall Downscaling Using Random Forests in Spain

Water ◽  
2019 ◽  
Vol 11 (1) ◽  
pp. 125 ◽  
Author(s):  
Javier Diez-Sierra ◽  
Manuel del Jesus
Keyword(s):  
Daily Rainfall ◽  
Rainfall Data ◽  
Machine Learning Techniques ◽  
Explanatory Variables ◽  
Rain Gauges ◽  
Daily Rainfall Data ◽  
Learning Techniques ◽  
Hourly Rainfall ◽  
Rainfall Statistics ◽  
Poisson Cluster

Subdaily rainfall data, though essential for applications in many fields, is not as readily available as daily rainfall data. In this work, regression approaches that use atmospheric data and daily rainfall statistics as predictors are evaluated to downscale daily-to-subdaily rainfall statistics on more than 700 hourly rain gauges in Spain. We propose a new approach based on machine learning techniques that improves the downscaling skill of previous methodologies. Results are grouped by climate types (following the Köppen–Geiger classification) to investigate possible missing explanatory variables in the analysis. The methodology is then used to improve the ability of Poisson cluster models to simulate hourly rainfall series that mimic the statistical behavior of the observed ones. This approach can be applied for the study of extreme events and for daily-to-subdaily precipitation disaggregation in any location of Spain where daily rainfall data are available.

scholarly journals Improved input to distributed hydrologic model in areas with sparse subdaily rainfall data using multivariate daily rainfall disaggregation

2018 ◽  
Vol 20 (4) ◽  
pp. 784-797 ◽  
Author(s):  
Marija Ivković ◽  
Andrijana Todorović ◽  
Jasna Plavšić
Keyword(s):  
Daily Rainfall ◽  
Flood Forecasting ◽  
Rain Gauge ◽  
Hydrologic Model ◽  
Rainfall Data ◽  
Rain Gauges ◽  
Observation Network ◽  
Hourly Rainfall ◽  
Extreme Flood ◽  
Flood Simulations

Abstract Flood forecasting relies on good quality of observed and forecasted rainfall. In Serbia, the recording rain gauge network is sparse and rainfall data mainly come from dense non-recording rain gauges. This is not beneficial for flood forecasting in smaller catchments and short-duration events, when hydrologic models operating on subdaily scale are applied. Moreover, differences in rainfall amounts from two types of gauges can be considerable, which is common in operational hydrological practice. This paper examines the possibility of including daily rainfall data from dense observation networks in flood forecasting based on subdaily data, using the extreme flood event in the Kolubara catchment in May 2014 as a case study. Daily rainfall from a dense observation network is disaggregated to hourly scale using the MuDRain multivariate disaggregation software. The disaggregation procedure results in well-reproduced rainfall dynamics and adjusts rainfall volume to the values from the non-recording gauges. The fully distributed wflow_hbv model, which is under development as a forecasting tool for the Kolubara catchment, is used for flood simulations with two alternative hourly rainfall data. The results show an improvement when the disaggregated rainfall from denser network is used, thus indicating the significance of better representation of rainfall temporal and spatial variability for flood forecasting.

scholarly journals Posterior Predictive of Bayesian Vector Autoregressive (BVAR) and Adjusting Transformation on the Spatio Temporal Disaggregation Method: Predict Hourly rainfall data at the outsampled Locations

2019 ◽  
pp. 357-369
Author(s):  
Suci Astutik ◽  
Umu Sa’adah ◽  
Supriatna Adhisuwignjo ◽  
Rauzan Sumara
Keyword(s):  
State Space ◽  
Daily Rainfall ◽  
Rainfall Data ◽  
Bayesian Var ◽  
Vector Autoregressive ◽  
Daily Rainfall Data ◽  
Hourly Rainfall ◽  
The Future ◽  
Temporal Disaggregation ◽  
Spatio Temporal

This research is a development from previous research that has studied the method of spatio temporal disaggregation with State space and adjusting procedures for predicting hourly rainfall based on daily rainfall (Astutik et al, 2013). However, this study is limited to predicting hourly rainfall in some sampled locations in the future. Astutik et al (2017, 2018) have modeled hourly and daily rainfall using posterior predictive bayesian VAR at the Sampean watershed of Bondowoso. This study aims to predict hourly rainfall data based on daily rainfall data in the future at the outsampled locations using posterior predictive bayesian VAR and adjusting procedures in the method of spatio temporal disaggregation.

Temporal disaggregation of daily rainfall data in SuDS design: a case study in Tuscany

2021 ◽  
Author(s):  
Matteo Pampaloni ◽  
Alvaro Sordo Ward ◽  
Paola Bianucci ◽  
Ivan Gabriel Martin ◽  
Luis Garrote ◽  
...  
Keyword(s):  
Daily Rainfall ◽  
Rainfall Data ◽  
Design Parameters ◽  
Rainfall Regime ◽  
Rainfall Time Series ◽  
Storm Events ◽  
Time Step ◽  
Daily Rainfall Data ◽  
Hourly Rainfall ◽  
Temporal Disaggregation

<p>Sustainable urban Drainage Systems (SuDS), by themselves or combined with grey traditional infrastructures, help to diminish the runoff volume and peak flow, as well as to improve the water quality. Hydrological design of SuDS is usually based on rainfall volumetric percentiles as the number of rainfall events, N<sub>x</sub>, or the accumulated volume of the rainfall series, V<sub>x</sub>, to be managed. Sub-index x refers to common qualities used in SuDS design, like 80, 85, 90 and 95%. Usually, only daily rainfall data are available. Nevertheless, due to the characteristics of the urban watershed involved in the SuDS implementation, the quantification of design parameters for these facilities needs sub-hourly rainfall time series. To overcome this issue, a temporal disaggregation methodology was proposed based on the use of a stochastic rainfall generator model (RainSim V3). We analysed the case of Florence University rain gauge (Tuscany, Italy), by collecting 20 years (in the period from 1998 to 2018) of observed data at 15 minutes time step. First, we verified the ability of RainSim model to reproduce observed rainfall patterns at 15 minutes time-step. The parameters of the stochastic model were estimated using observed data with 24 hours time-step. We generated 100 series of 20 years each with a time step of 15 minutes. We accounted two variables to implement the storm events extraction: a) the Minimum Inter-event Time (MIT) between storm events; 2) the storm volume threshold. We obtained a better characterization of the rainfall regime by applying the temporal disaggregation methodology than using daily-observed data. Second, we compared the SuDS design parameters N<sub>x</sub> and V<sub>x</sub>, obtained by using the stochastically generated rainfall, the observed daily and 15 minutes data. Moreover, the effect of different MITs and different thresholds on N<sub>x</sub> and V<sub>x </sub>were evaluated. In all the cases, results show that N<sub>x</sub> and V<sub>x</sub> obtained with the median of the simulated series were closer to the actual observed parameters based on 15 minutes time step than the ones calculated with the observed daily data. Therefore, the proposed temporal disaggregation method arises as an efficient technique to overcome the lack of sub-hourly rainfall data, necessary to adequately design SuDS.</p>

scholarly journals The Daily and Hourly Rainfall Data Modeling Using Vector Autoregressive (VAR) with Maximum Likelihood Estimator (MLE) and Bayesian Method (Case Study in Sampean Watershed of Bondowoso, Indonesia)

2018 ◽  
Author(s):  
Suci Astutik ◽  
Umu Sa'adah ◽  
Supriatna Adhisuwignjo ◽  
Rauzan Sumara
Keyword(s):  
Maximum Likelihood ◽  
Maximum Likelihood Estimator ◽  
Daily Rainfall ◽  
Rainfall Data ◽  
Var Model ◽  
Likelihood Estimator ◽  
Vector Autoregressive ◽  
Daily Rainfall Data ◽  
Run Off ◽  
Hourly Rainfall

The hourly and daily rainfall data which is spatially distributed are required as an input for run-off rain model. Furthermore, the run-off rain model is used to detect early flooding. The daily and hourly rainfall data have characteristics that most of data are zero. Therefore we need a model which can capture the phenomenon. A time series model involving location, which is a model that can be developed to approach the daily and hourly rainfall data, we can call Vector Autoregressive (VAR) model. The VAR model allows us for modeling rainfall data in several areas. However, in certain conditions the VAR model often occurs over-parameterization and reduces degrees of freedom. The aim of this study is to compare the VAR model with Maximum Likelihood Estimator (MLE) and Bayesian to hourly and daily rainfall data in SampeanWatershed of Bondowoso. The results showed that the hourly and daily rainfall data are fitted to VAR process of orde 5 and 1 respectively. Based on the AIC and SBC values indicate that the Bayesian is better than the MLE method. The Bayesian is able to predict parameters by producing a smaller variance covariance matrix than the MLE.

The effect of weather, drainage efficiency and duration of spring cultivations on barley yields in England

1972 ◽  
Vol 7 (2) ◽  
pp. 79-83 ◽  
Author(s):  
L P Smith
Keyword(s):  
Yield Loss ◽  
Daily Rainfall ◽  
Rainfall Data ◽  
Work Schedule ◽  
Average Yield ◽  
Daily Rainfall Data ◽  
Arable Farming ◽  
Work Distribution

Daily rainfall data for twenty years in arable farming areas are analysed with respect to four standards of drainage and for three lengths of schedule of spring work. Distribution and frequency in time of available work days are interpreted in terms of lateness of sowing and of barley yield. Formulae are established to calculate average yield loss in terms of drainage standard and work schedule, enabling estimates to be made of the effect of planned improvements.

scholarly journals Bayesian Time Series Modelling of the Italian Daily Rainfall Data using Mixed Distribution

2019 ◽  
Vol 8 (4) ◽  
pp. 2279-2288
Keyword(s):  
Time Series ◽  
Daily Rainfall ◽  
Rainfall Data ◽  
Seasonal Effects ◽  
Model Parameters ◽  
Daily Rainfall Data ◽  
Mixed Distribution ◽  
Time Series Modelling ◽  
Bayesian Time Series ◽  
The Ideal

A combination of continuous and discrete elements is referred to as a mixed distribution. For example, daily rainfall data consist of zero and positive values. We aim to develop a Bayesian time series model that captures the evolution of the daily rainfall data in Italy, focussing on directly linking the amount and occurrence of rainfall. Two gamma (G1 and G2) distributions with different parameterisations and lognormal distribution were investigated to identify the ideal distribution representing the amount process. Truncated Fourier series was used to incorporate the seasonal effects which captures the variability in daily rainfall amounts throughout the year. A first-order Markov chain was used to model rainfall occurrence conditional on the presence or absence of rainfall on the previous day. We also built a hierarchical prior structure to represent our subjective beliefs and capture the initial uncertainties of the unknown model parameters for both amount and occurrence processes. The daily rainfall data from Urbino rain gauge station in Italy were then used to demonstrate the applicability of our proposed methods. Residual analysis and posterior predictive checking method were utilised to assess the adequacy of model fit. In conclusion, we clearly found that our proposed method satisfactorily and accurately fits the Italian daily rainfall data. The gamma distribution was found to be the ideal probability density function to represent the amount of daily rainfall.

scholarly journals Structural break or long memory: an empirical survey on daily rainfall data sets across Malaysia

2013 ◽  
Vol 17 (4) ◽  
pp. 1311-1318 ◽  
Author(s):  
F. Yusof ◽  
I. L. Kane ◽  
Z. Yusop
Keyword(s):  
Long Memory ◽  
Structural Breaks ◽  
Structural Break ◽  
Daily Rainfall ◽  
Least Square ◽  
Rainfall Data ◽  
Data Sets ◽  
Original Series ◽  
Empirical Survey ◽  
Daily Rainfall Data

Abstract. A short memory process that encounters occasional structural breaks in mean can show a slower rate of decay in the autocorrelation function and other properties of fractional integrated I (d) processes. In this paper we employed a procedure for estimating the fractional differencing parameter in semiparametric contexts proposed by Geweke and Porter-Hudak (1983) to analyse nine daily rainfall data sets across Malaysia. The results indicate that all the data sets exhibit long memory. Furthermore, an empirical fluctuation process using the ordinary least square (OLS)-based cumulative sum (CUSUM) test for the break date was applied. Break dates were detected in all data sets. The data sets were partitioned according to their respective break date, and a further test for long memory was applied for all subseries. Results show that all subseries follows the same pattern as the original series. The estimate of the fractional parameters d1 and d2 on the subseries obtained by splitting the original series at the break date confirms that there is a long memory in the data generating process (DGP). Therefore this evidence shows a true long memory not due to structural break.

Sign in / Sign up

Export Citation Format

Share Document