scholarly journals Investigation of satellite rainfall-driven rainfall–runoff model using deep learning approaches in two different catchments in India

2021 ◽  
Author(s):  
Pavan Kumar Yeditha ◽  
Maheswaran Rathinasamy ◽  
Sai Sumanth Neelamsetty ◽  
Biswa Bhattacharya ◽  
Ankit Agarwal
Keyword(s):  
Deep Learning ◽  
River Basin ◽  
Flood Forecasting ◽  
Precipitation Data ◽  
Rainfall Runoff ◽  
Data Set ◽  
Satellite Precipitation ◽  
Satellite Rainfall ◽  
Rainfall Runoff Model ◽  
Runoff Model

Abstract Rainfall–runoff models are valuable tools for flood forecasting, management of water resources, and drought warning. With the advancement in space technology, a plethora of satellite precipitation products (SPPs) are available publicly. However, the application of the satellite data for the data-driven rainfall–runoff model is emerging and requires careful investigation. In this work, two satellite rainfall data sets, namely Global Precipitation Measurement-Integrated Multi-Satellite Retrieval Product V6 (GPM-IMERG) and Climate Hazards Group Infrared Precipitation with Station (CHIRPS), are evaluated for the development of rainfall–runoff models and the prediction of 1-day ahead streamflow. The accuracy of the data from the SPPs is compared to the India Meteorological Department (IMD)-gridded precipitation data set. Detection metrics showed that for light rainfall (1–10 mm), the probability of detection (POD) value ranges between 0.67 and 0.75 and with an increasing rainfall range, i.e., medium and heavy rainfall (10–50 mm and >50 mm), the POD values ranged from 0.24 to 0.45. These results indicate that the satellite precipitation performs satisfactorily with reference to the IMD-gridded data set. Using the daily precipitation data of nearly two decades (2000–2018) over two river basins in India's Eastern part, artificial neural network, extreme learning machine (ELM), and long short-time memory (LSTM) models are developed for rainfall–runoff modelling. One-day ahead runoff prediction using the developed rainfall–runoff modelling confirmed that both the SPPs are sufficient to drive the rainfall–runoff models with a reasonable accuracy estimated using the Nash–Sutcliffe Efficiency coefficient, correlation coefficient, and the root-mean-squared error. In particular, the 1-day streamflow forecasts for the Vamsadhara river basin (VRB) using LSTM with GPM-IMERG inputs resulted in NSC values of 0.68 and 0.67, while ELM models for Mahanadhi river basin (MRB) with the same input resulted in NSC values of 0.86 and 0.87, respectively, during training and validation stages. At the same time, the LSTM model with CHIRPS inputs for the VRB resulted in NSC values of 0.68 and 0.65, and the ELM model with CHIRPS inputs for the MRB resulted in NSC values of 0.89 and 0.88, respectively, in training and validation stages. These results indicated that both the SPPs could reliably be used with LSTM and ELM models for rainfall–runoff modelling and streamflow prediction. This paper highlights that deep learning models, such as ELM and LSTM, with the GPM-IMERG products can lead to a new horizon to provide flood forecasting in flood-prone catchments.

scholarly journals A note on leveraging synergy in multiple meteorological datasets with deep learning for rainfall-runoff modeling

2020 ◽  
Author(s):  
Frederik Kratzert ◽  
Daniel Klotz ◽  
Sepp Hochreiter ◽  
Grey S. Nearing
Keyword(s):  
Deep Learning ◽  
Short Term Memory ◽  
Rainfall Runoff ◽  
Data Set ◽  
Simulation Accuracy ◽  
Meteorological Forcing ◽  
Runoff Modeling ◽  
Long Short Term Memory ◽  
Rainfall Runoff Model ◽  
Runoff Model

Abstract. A deep learning rainfall-runoff model can take multiple meteorological forcing products as inputs and learn to combine them in spatially and temporally dynamic ways. This is demonstrated using Long Short Term Memory networks (LSTMs) trained over basins in the continental US using the CAMELS data set. Using multiple precipitation products (NLDAS, Maurer, DayMet) in a single LSTM significantly improved simulation accuracy relative to using only individual precipitation products. A sensitivity analysis showed that the LSTM learned to utilize different precipitation products in different ways in different basins and for simulating different parts of the hydrograph in individual basins.

Effect of input variables on rainfall-runoff modeling using a deep learning method

2021 ◽  
Author(s):  
Kazuki yokoo ◽  
Kei ishida ◽  
Takeyoshi nagasato ◽  
Ali Ercan
Keyword(s):  
Deep Learning ◽  
Air Temperature ◽  
Meteorological Variables ◽  
Learning Method ◽  
Rainfall Runoff ◽  
Model Accuracy ◽  
Runoff Modeling ◽  
Input Variables ◽  
Rainfall Runoff Model ◽  
Runoff Model

<p>In recent years, deep learning has been applied to various issues in natural science, including hydrology. These application results show its high applicability. There are some studies that performed rainfall-runoff modeling by means of a deep learning method, LSTM (Long Short-Term Memory). LSTM is a kind of RNN (Recurrent Neural Networks) that is suitable for modeling time series data with long-term dependence. These studies showed the capability of LSTM for rainfall-runoff modeling. However, there are few studies that investigate the effects of input variables on the estimation accuracy. Therefore, this study, investigated the effects of the selection of input variables on the accuracy of a rainfall-runoff model by means of LSTM. As the study watershed, this study selected a snow-dominated watershed, the Ishikari River basin, which is in the Hokkaido region of Japan. The flow discharge was obtained at a gauging station near the outlet of the river as the target data. For the input data to the model, Meteorological variables were obtained from an atmospheric reanalysis dataset, ERA5, in addition to the gridded precipitation dataset. The selected meteorological variables were air temperature, evaporation, longwave radiation, shortwave radiation, and mean sea level pressure. Then, the rainfall-runoff model was trained with several combinations of the input variables. After the training, the model accuracy was compared among the combinations. The use of meteorological variables in addition to precipitation and air temperature as input improved the model accuracy. In some cases, however, the model accuracy was worsened by using more variables as input. The results indicate the importance to select adequate variables as input for rainfall-runoff modeling by LSTM.</p>

scholarly journals A Modified IHACRES Rainfall–Runoff Model for Predicting Hydrologic Response of a River Basin System with a Relevant Groundwater Component

Proceedings ◽  
2018 ◽  
Vol 7 (1) ◽  
pp. 24
Author(s):  
Iolanda Borzì ◽  
Brunella Bonaccorso ◽  
Aldo Fiori
Keyword(s):  
River Basin ◽  
Slow Component ◽  
Temperature Data ◽  
Continuous Series ◽  
Effective Rainfall ◽  
Rainfall Runoff ◽  
Aquifer Recharge ◽  
Daily Streamflow ◽  
Rainfall Runoff Model ◽  
Runoff Model

A flow regime can be broadly categorized as either perennial, intermittent, or ephemeral, depending on whether the streamflow is continuous all year round, or ceasing for weeks or months each year. Various conceptual models are needed to capture the behavior of these different flow regimes, which reflect differences in the stream–groundwater hydrologic connection. As the hydrologic connection becomes more transient and a catchment’s runoff response more nonlinear, such as for intermittent streams, the need for explicit representation of the groundwater increases. In the present study, we investigated the connection between the Northern Etna groundwater system and the Alcantara River basin in Sicily, which is intermittent in the upstream, and perennial since the midstream, due to groundwater resurgence. To this end, we apply a modified version of IHACRES rainfall–runoff model, whose input data are a continuous series of concurrent daily streamflow, rainfall and temperature data. The structure of the model includes three different modules: (1) a nonlinear loss module that transforms precipitation to effective rainfall by considering the influence of temperature; (2) a linear module based on the classical convolution between effective rainfall and the unit hydrograph which is able to simulate the quick component of the runoff; and (3) a second nonlinear module that simulates the slow component of the runoff and that feeds the groundwater storage. From the sum of the quick and slow components (except for groundwater losses, representing the aquifer recharge), the total streamflow is derived. This model structure is applied separately to sub-basins showing different hydrology and land use. The model is calibrated at Mojo cross-section, where daily streamflow data are available. Point rainfall and temperature data are spatially averaged with respect to the considered sub-basins. Model calibration and validation are carried out for the period 1984–1986 and 1987–1988 respectively.

scholarly journals Technical Note: Updating procedure for flood forecasting with conceptual HBV-type models

2006 ◽  
Vol 10 (6) ◽  
pp. 783-788 ◽  
Author(s):  
Th. Wöhling ◽  
F. Lennartz ◽  
M. Zappa
Keyword(s):  
Flood Forecasting ◽  
Technical Note ◽  
Rainfall Runoff ◽  
Runoff Generation ◽  
Mountainous Catchments ◽  
Rainfall Runoff Model ◽  
Updating Procedure ◽  
Global And Local ◽  
Hbv Model ◽  
Runoff Model

Abstract. Flood forecasting is of increasing importance as it comes to an increasing variability in global and local climates. But rainfall-runoff models are far from being perfect. In order to achieve a better prediction for emerging flood events, the model outputs have to be continuously updated. This contribution introduces a rather simple, yet effective updating procedure for the conceptual semi-distributed rainfall-runoff model PREVAH, whose runoff generation module relies on similar algorithms as the HBV-Model. The current conditions of the system, i.e. the contents of the upper soil reservoirs, are updated by the proposed method. The testing of the updating procedure on data from two mountainous catchments in Switzerland reveals a significant increase in prediction accuracy with regards to peak flow.

scholarly journals Importance of the information content in the study area when regionalising rainfall-runoff model parameters: the role of nested catchments and gauging station density

2020 ◽  
Author(s):  
Mattia Neri ◽  
Juraj Parajka ◽  
Elena Toth
Keyword(s):  
Information Content ◽  
Model Parameters ◽  
Rainfall Runoff ◽  
Data Set ◽  
Study Region ◽  
Averaging Methods ◽  
Rainfall Runoff Model ◽  
Runoff Model ◽  
Station Density ◽  
Averaging Techniques

Abstract. The set up of a rainfall-runoff model in a river section where no streamflow measurements are available for its calibration is one of the key research activity for the Prediction in Ungauged Basins (PUB): in order to do so it is possible to regionalise the model parameters based on the information available in gauged sections in the study region. The information content in the data set of gauged river stations plays an essential role in the assessment of the best regionalisation method: this study analyses how the performances of different model regionalisation approaches are influenced by the information richness of the available regional data set, and in particular by its gauging density and by the presence of nested catchments, that are expected to be hydrologically very similar. The research is carried out over a densely gauged dataset covering the Austrian country, applying two different rainfall-runoff models: a semi-distributed version of the HBV model (TUW model), and the Cemaneige-GR6J model. The regionalisation approaches include both methods which transfer the entire set of model parameters from donor catchments, thus maintaining correlation among parameters (output averaging techniques), and methods which derive each target parameter independently, as a function of the calibrated donors’ ones (parameter averaging techniques). The regionalisation techniques are first implemented using all the basins in the dataset as potential donors, showing that the output-averaging methods outperform the parameter-averaging kriging method, highlighting the importance of maintaining the correlation between the parameter values. The regionalisation is then repeated decreasing the information content of the data set, by excluding the nested basins, identified taking into account either the position of the closing section along the river or the percentage of shared drainage area. The parameter-averaging kriging is the method that is less impacted by the exclusion of the nested donors, whereas the methods transferring the entire parameter set from only one donor suffer the highest deterioration, since the single most similar or closest donor is often a nested one. On the other hand, the output-averaging methods degrade more gracefully, showing that exploiting the information resulting from more than one donor increases the robustness of the approach also in regions that do not have so many nested catchments as the Austrian one. Finally, the deterioration resulting from decreasing the station density on the regionalisation was analysed, showing that the output averaging methods using as similarity measure a set of catchment descriptors, rather than the geographical distance, are more capable to adapt to less dense datasets. The study confirms how the predictive accuracy of parameter regionalisation techniques strongly depends on the information content of the dataset of available donor catchments and indicates that the output-averaging approaches, using more than one donor basin but preserving the correlation structure of the parameter set, seem to be preferable for regionalisation purposes in both data-poor and data-rich regions.

Sign in / Sign up

Export Citation Format

Share Document