Artificial intelligence and geo-statistical models for stream-flow forecasting in ungauged stations: state of the art

Valizadeh, Nariman; Mirzaei, Majid; Allawi, Mohammed Falah; Afan, Haitham Abdulmohsin; Mohd, Nuruol Syuhadaa; Hussain, Aini; El-Shafie, Ahmed

doi:10.1007/s11069-017-2740-7

Cited by 28 publications

(7 citation statements)

References 108 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Drought and flood events can adversely affect the normal operation of water supply, irrigation, and hydropower systems, and produce socio-economic and ecological damages. Therefore, discharge modeling and forecasting are crucial and have been largely studied in the literature employing different approaches based on physical processes and data-driven techniques [1][2][3][4][5]. Distributed and semi-distributed models are by far the most adopted and well-known rainfall-runoff models used for discharge forecasting.…”

Section: Introductionmentioning

confidence: 99%

“…On the other hand, new attempts of using machine learning models for discharge forecasting have been pursued in the last decade with encouraging results. Some authors [2,4,5] provide reviews of the application of machine learning-based models for this purpose. Several streamflow forecasting studies [23][24][25][26][27] use data-driven models employing radar-derived rainfall as input; this requires a preliminary step for transforming reflectivity (native ground radar variable) into rainfall rate [14].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Assessment of Native Radar Reflectivity and Radar Rainfall Estimates for Discharge Forecasting in Mountain Catchments with a Random Forest Model

et al. 2020

View full text Add to dashboard Cite

Discharge forecasting is a key component for early warning systems and extremely useful for decision makers. Forecasting models require accurate rainfall estimations of high spatial resolution and other geomorphological characteristics of the catchment, which are rarely available in remote mountain regions such as the Andean highlands. While radar data is available in some mountain areas, the absence of a well distributed rain gauge network makes it hard to obtain accurate rainfall maps. Thus, this study explored a Random Forest model and its ability to leverage native radar data (i.e., reflectivity) by providing a simplified but efficient discharge forecasting model for a representative mountain catchment in the southern Andes of Ecuador. This model was compared with another that used as input derived radar rainfall (i.e., rainfall depth), obtained after the transformation from reflectivity to rainfall rate by using a local Z-R relation and a rain gauge-based bias adjustment. In addition, the influence of a soil moisture proxy was evaluated. Radar and runoff data from April 2015 to June 2017 were used. Results showed that (i) model performance was similar by using either native or derived radar data as inputs (0.66 < NSE < 0.75; 0.72 < KGE < 0.78). Thus, exhaustive pre-processing for obtaining radar rainfall estimates can be avoided for discharge forecasting. (ii) Soil moisture representation as input of the model did not significantly improve model performance (i.e., NSE increased from 0.66 to 0.68). Finally, this native radar data-based model constitutes a promising alternative for discharge forecasting in remote mountain regions where ground monitoring is scarce and hardly available.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Assessment of Native Radar Reflectivity and Radar Rainfall Estimates for Discharge Forecasting in Mountain Catchments with a Random Forest Model

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Reservoir inflow patterns entail highly complex processes to be described using simple predictive models because of the nonlinearity, spatial distribution, and time varying characteristics of the data (Bai et al 2015;Valizadeh et al 2017). There are two primary methods for inflow forecasting that have been examined in previous studies: mechanistic or Bphysically-based^models a n d s y s t e m t h eo r e t i c ( o r da t a-d r i v en ) m o d e l s .…”

Section: Introduction 1backgroundmentioning

confidence: 99%

Reservoir inflow forecasting with a modified coactive neuro-fuzzy inference system: a case study for a semi-arid region

Allawi

Jaafar

Hamzah

et al. 2017

Theor Appl Climatol

Self Cite

View full text Add to dashboard Cite

Existing forecast models applied for reservoir inflow forecasting encounter several drawbacks, due to the difficulty of the underlying mathematical procedures being to cope with and to mimic the naturalization and stochasticity of the inflow data patterns. In this study, appropriate adjustments to the conventional coactive neuro-fuzzy inference system (CANFIS) method are proposed to improve the mathematical procedure, thus enabling a better detection of the high nonlinearity patterns found in the reservoir inflow training data. This modification includes the updating of the back propagation algorithm, leading to a consequent update of the membership rules and the induction of the centre-weighted set rather than the global weighted set used in feature extraction. The modification also aids in constructing an integrated model that is able to not only detect the nonlinearity in the training data but also the wide range of features within the training data records used to simulate the forecasting model. To demonstrate the model's efficacy, the proposed CANFIS method has been applied to forecast monthly inflow data at Aswan High Dam (AHD), located in southern Egypt. Comparative analyses of the forecasting skill of the modified CANFIS and the conventional ANFIS model are carried out with statistical score indicators to assess the reliability of the developed method. The statistical metrics support the better performance of the developed CANFIS model, which significantly outperforms the ANFIS model to attain a low relative error value (23%), mean absolute error (1.4 BCM month −1 ), root mean square error (1.14 BCM month −1 ), and a relative large coefficient of determination (0.94). The present study ascertains the better utility of the modified CANFIS model in respect to the traditional ANFIS model applied in reservoir inflow forecasting for a semi-arid region.

show abstract

“…Several fields of research and applications have exploited this direction but state of the art for weather prediction analysis still is under development [4]. Few examples such as rainfall predictions using limited data setting [3], estimation of hydrological variables to forecast the runoff at ungauged river basins [5], air quality index estimation and prediction [6], analyzing and predicting an individual's movements/locations [7], optical flow based interpolation for structure-preservation using tomography images for improving data quality [8], [9], precipitation now casting as a spatio-temporal sequence forecasting problem [10], etc., are based on ANN.…”

Section: B Image Processing Frameworkmentioning

confidence: 99%

Deep Learning Algorithm for Satellite Imaging Based Cyclone Detection

Shakya

Kumar

Goswami

2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Satellite images are primary data in weather prediction modeling. Deep learning-based approach, a viable candidate for automatic image processing, requires large sets of annotated data with diverse characteristics for training purposes. Accuracy of weather prediction improves with data having a relatively dense temporal resolution. We have employed interpolation and data augmentation techniques for enhancement of the temporal resolution and diversifications of characters in a given dataset. Algorithm requires classical approaches during preprocessing steps. Three optical flow methods using 14 different constraint optimization techniques and five error estimates are tested here. The artificially enriched data (optimal combination from the previous exercise) are used as a training set for a convolutional neural network to classify images in terms of storm or nonstorm. Several cyclone data (eight cyclone datasets of a different class) were used for training. A deep learning model is trained and tested with artificially densified and classified storm data for cyclone classification and locating the cyclone vortex giving minimum 90% and 84% accuracy, respectively. In the final step, we show that the linear regression method can be used for predicting the path. Index Terms-Miscellaneous applications, optical data. I. INTRODUCTION A. Remote Sensing (RS) R S APPLICATIONS, mainly, via satellite imagery has expanded from conventional meteorology, geological exploration, oceanography toward homeland security, urban planning, ecology and several other novel unconventional fields. Costeffective unmanned aerial vehicle (UAV) and weather balloon approaches have shared the burden but suffer from limitations such as 1) low elevation point for imaging, 2) stability issues under bad weather conditions, and 3) dependence on navigation satellites. Meteorological applications, especially weather forecasting for disaster readiness, yet, requires dedicated but costly satellite infrastructure. Routine scheduling of multispectral Manuscript

show abstract

Artificial intelligence and geo-statistical models for stream-flow forecasting in ungauged stations: state of the art

Cited by 28 publications

References 108 publications

Assessment of Native Radar Reflectivity and Radar Rainfall Estimates for Discharge Forecasting in Mountain Catchments with a Random Forest Model

Assessment of Native Radar Reflectivity and Radar Rainfall Estimates for Discharge Forecasting in Mountain Catchments with a Random Forest Model

Reservoir inflow forecasting with a modified coactive neuro-fuzzy inference system: a case study for a semi-arid region

Deep Learning Algorithm for Satellite Imaging Based Cyclone Detection

Contact Info

Product

Resources

About