Abstract
In the present study, missing daily rainfall data of nine rain gauge stations, each of the two different sub-basins, have been predicted by using different ANN models. The different ANN models have been developed by using variations in input data point numbers, the algorithm used, the activation function, the number of hidden layers, and neurons in the hidden layers used. The performance of different models has been analyzed using four statistical values: R2, NSE, PBIAS and AIC. As different ANN models perform differently at different stations, a rank-based methodology has been developed to determine the best ANN model. Since hydrological studies typically focus on the estimation of flood levels, the comparison has been made only of the upper 25 percent data (Q4), which is more important for flood estimation and basin management. Different conventional methods were also used, and their respective performance parameters were evaluated. It is found that ANN models performed much better as compared to conventional methods for both the entire range of data as well as the Q4 range of data. To further elaborate on the performance of the best ANN models, the performance values obtained for the three best ANN models, namely ANN079, ANN095, and ANN035, are presented. The results achieved by these selected best ANN models are satisfactory and prove the methodology presented.
Research Highlights
-
1.
Missing daily rainfall data of nine stations each from two different basins predicted by developing a large number of ANN models in this study.
-
2.
For entire range of data, the best performing model was developed using rprop– algorithm, logistic activation function, and, two hidden neuron layers using month as one input neuron with rainfall of nine nearby stations as other input neuron.
-
3.
The performance of the developed models was analyzed by four statistical values, i.e., R2, NSE, PBIAS and AIC.
-
4.
ANN model’s performance found far better than the conventional methods. For example, average NSE value of ANN model type ANN079 was 0.64, which reduced to only 0.13 for the best conventional IDWA 6 model.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Adams III T E and Pagano T C 2016 Flood forecasting: A global perspective; Amsterdam, Elsevier, https://doi.org/10.1016/B978-0-12-801884-2.09999-0.
Ahrens B 2006 Distance in spatial interpolation of daily rain gauge data; Hydrol. Earth Syst. Sci. 10 197–208, https://doi.org/10.5194/hess-10-197-2006.
Akaike H 1973 Information theory and an extension of the maximum likelihood principle: In: Second International Symposium on Information Theory (eds) Petrov B N and Csaki B F, Academiai Kiado, Budapest, pp. 267–281.
Bustami R, Bessaih N, Bong Charles H J and Sushila S 2007 Artificial neural network for precipitation and water level prediction of bed up river; Int. J. Comp. Sci. 34 228–233.
Calzado P R, Camacho J G, Bernal F P and Lopez M F P 2008 A novel approach to precipitation series completion in climatological datasets: Application to Andalusia; Int. J. Climatol. 28 1525–1534, https://doi.org/10.1002/joc.1657.
Chang C L, Lo S L and Yu S L 2006 The parameter optimization in the inverse distance method by genetic algorithm for estimating precipitation; Environ. Monit. Assess. 117 145–155, https://doi.org/10.1007/s10661-006-8498-0.
Cheng J, Kuang Q, Shen C, Liu J, Tan X and Liu W 2020 ResLap: Generating high-resolution climate prediction through image super-resolution; IEEE Access 8 39,623–39,634, https://doi.org/10.1109/ACCESS.2020.2974785.
Chiu P C, Selamat A, Krejcar O and Kuok-King K 2019 Missing rainfall data estimation using artificial neural network and nearest neighbor imputation; Front. Artif. Intell. Appl. 318 132–143, https://doi.org/10.3233/FAIA190044.
CWC 2014 Tapi basin watershed atlas; Cent. Water Commission, New Delhi.
Dadhich A, Goyal R and Dadhich P N 2021 Assessment and prediction of groundwater using geospatial and ANN modeling; Water Resour. Manag. 35 2879–2893, https://doi.org/10.1007/s11269-021-02874-8.
Djerbouai S 2022 Missing precipitation data estimation using long short-term memory deep neural networks; J. Ecol. Eng. 23 216–225, https://doi.org/10.12911/22998993/147322.
Garrett J H Jr 1994 Where and why artificial neural networks are applicable in Civil Engineering Editorial; J. Comput. Civil Eng. 8 129–130.
Giorgi F 2006 Climate change hot-spots; Geophys. Res. Lett. 33 L08707, https://doi.org/10.1029/2006GL025734.
Golkhatmi N S, Sanaeinejad S H, Ghahraman B and Pazhand H R 2012 Extended modified inverse distance method for interpolation rainfall; Int. J. Eng. Invent. 1 57–65.
Goncalves A M and Alpuim T 2006 Precipitation measurement and the analysis of hydrological resources in a river basin; Seventh Int. Symp. Spat. Accuracy Assess. Natural Resour. Environ. Sci., pp. 851–860.
Gupta H V, Sorooshian S and Yapo P O 1999 Status of automatic calibration for hydrologic models: Comparison with multilevel expert calibration; J. Hydrol. Eng. 4 135–143, https://doi.org/10.1061/(ASCE)1084-0699(1999)4:2(135).
Kumar B, Atey K and Singh B 2023 On the modern deep learning approaches for precipitation downscaling; Earth Sci. Info. 16 1459–1472, https://doi.org/10.1007/s12145-023-00970-4.
Li J and Heap A D 2008 A review of spatial interpolation methods for environmental scientists; Geosci. Australia 23 137.
McCuen R H, Leahy R B and Johnson P A 1990 Problems with logarithmic transformations in regressions; J. Hydrol. Eng., ASCE 116 414–428.
McCuen R H, Knight Z and Cutter A G 2006 Evaluation of the Nash–Sutcliffe efficiency index; J. Hydrol. Eng. 11 597–602.
Mishra Sharma S and Mitra A 2022 ResDeepD: A residual super-resolution network for deep downscaling of daily precipitation over India; Environ. Data Sci. 1 1–10, https://doi.org/10.1017/eds.2022.23.
Munoth P and Goyal R 2019 Effects of DEM source, spatial resolution and drainage area threshold values on hydrological modeling; Water Resour. Manag. 33 3303–3319, https://doi.org/10.1007/s11269-019-02303-x.
Munoth P and Goyal R 2020a Impacts of land use land cover change on runoff and sediment yield of Upper Tapi River Sub-basin, India; Int. J. River Basin Manag. 18 177–189, https://doi.org/10.1080/15715124.2019.1613413.
Munoth P and Goyal R 2020b Hydromorphological analysis of Upper Tapi River Sub-basin, India, using QSWAT model; Earth Syst. Environ. 6 2111–2127, https://doi.org/10.1007/s40808-020-00821-x.
Narimani R and Changhyun J 2021 Multilayer perceptron neural networks for estimating missing rainfall data; EGU Gen. Assembly 19 EGU21–10542, https://doi.org/10.5194/egusphere-egu21-10542.
Nash J E and Sutcliffe J V 1970 River flow forecasting through conceptual models part I – A discussion of principles; J. Hydrol. 10 282–290, https://doi.org/10.1016/0022-1694(70)90255-6.
Nourani V, Baghanam A H and Gebremichael M 2012 Investigating the ability of artificial neural network (ANN) models to estimate missing rain-gauge data; J. Environ. In-form. 19 38–50.
Papailiou I, Spyropoulos F, Trichakis I and Karatzas G P 2022 Artificial neural networks and multiple linear regression for filling in missing daily rainfall data; Water 14 2892, https://doi.org/10.3390/w14182892.
Paraskevas T, Dimitios R and Benardos A 2014 Use of artificial neural network for spatial rainfall analysis; J. Earth Syst. Sci. 123 457–465.
Roman U C, Patel P L and Porey P D 2012 Prediction of missing rainfall data using conventional and artificial neural network techniques; ISH J. Hydraul. Eng. 18 224–231, https://doi.org/10.1080/09715010.2012.721660.
Sugiura N 1978 Further analysis of the data by Akaike’s information criterion and the finite corrections; Comm. Stats. – Theory Methods 7 13–26, https://doi.org/10.1080/03610927808827599.
Tapi 2020 INDIA WRIS WIKI; http://indiawris.gov.in/wiki/doku.php?id=tapi&rev=1579259893.
Teegavarapu R S V, Tufail M and Ormsbee L 2009 Optimal functional forms for estimation of missing precipitation data; J. Hydro. 374 106–115, https://doi.org/10.1016/j.jhydrol.2009.06.014.
Wilk J, Kniveton D, Andersson L, Layberry R, Todd M C, Hughese D, Ringrose F and Vanderpost C 2006 Estimating rainfall and water balance over the Okavango River Basin for hydrological applications; J. Hydrol. 331 18–29, https://doi.org/10.1016/j.jhydrol.2006.04.049.
Xia Y, Fabian P, Stohl A and Winterhalter M 1999 Forest climatology: Estimation of missing values for Bavaria, Germany; Agr. Forest Meteorol. 96 131–144.
Acknowledgements
The authors are thankful to the Central Water Commission, Gujarat (India) and the Water Resources Department, Government of Rajasthan, for providing the required data on daily rainfall. The authors are also thankful to the India Meteorological Department, Prof P L Patel, Dr P V Timbadiya from the Department of Civil Engineering, SVNIT Surat, for providing the rainfall data for the Upper and Middle Tapi basin. The authors also thank MNIT Jaipur for providing all the working facilities.
Author information
Authors and Affiliations
Contributions
All authors contributed to this manuscript. Gyani Ram Kumawat: Collected data, designed and performed analysis, and wrote the first draft of the manuscript. Priyamitra Munoth: Collected data, prepared maps and figures, and provided inputs in writing and editing the manuscript. Rohit Goyal: Conceived and designed the analysis; contributed to analysis tools and methodology; provided inputs in writing the manuscript and editing.
Corresponding author
Additional information
Communicated by Parthasarathi Mukhopadhyay
Rights and permissions
About this article
Cite this article
Kumawat, G.R., Munoth, P. & Goyal, R. Improving prediction of missing rainfall data by identifying best Artificial Neural Network model. J Earth Syst Sci 132, 192 (2023). https://doi.org/10.1007/s12040-023-02203-0
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12040-023-02203-0