Analysis of Spatial Distribution of Precipitation Using Hydrological Modeling of Watersheds

doi:10.21203/rs.3.rs-4855854/v1

Download PDF

Research Article

Analysis of Spatial Distribution of Precipitation Using Hydrological Modeling of Watersheds

https://doi.org/10.21203/rs.3.rs-4855854/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

One of the significant challenges in hydrological modeling is the spatial variability of rainfall; even a slight change in rainfall patterns can lead to fundamental differences in runoff responses, affecting the accuracy of predictions. Moreover, the complexity of rainfall patterns, influenced by topography, increasingly leads to mispredictions. Therefore, this study investigates and analyzes the spatial distribution of rainfall and the role of the dispersion of raindrop data and its impact on rainfall-runoff modeling using raindrop data from two gridded precipitation products, PERSIANN-CDR and ERA5, at daily and monthly scales with the semi-distributed SWAT model. For this purpose, raindrop data from four stations—Talezang, Tang-e Panj Bakhtiari, Sepid Dasht Sezar, and the Dez Dam Rain gauge stations and Tele-Zang hydrometry station —from 2008 to 2019 were utilized.

The results of the simulations compared with the observed data indicate that due to the lack of a suitable raingage network and spatial distribution of rainfall, the SWAT rainfall-runoff simulation model was unable to establish a correspondence between the rainfall amounts and their corresponding discharge values. Consequently, an appropriate modeling with the observed data could not be achieved, as evidenced by the NSE coefficient values of 0.52 and -0.01 for daily calibration and validation periods, respectively, and 0.76 and 0.82 for monthly periods.

Additionally, it was observed that in the daily hydrological evaluation, the performance of the PERSIANN-CDR dataset was better, with NSE values of 0.29 and 0.59 for the calibration and validation periods, respectively. In contrast, the ERA5 dataset displayed superior capabilities in monthly hydrological evaluation for rainfall-runoff modeling, with calibration and validation coefficients of 0.77 and 0.82, respectively.

Hydrological evaluation

gridded data

ERA5

PERSIANN-CDR

SWAT

Spatial changes in precipitation have always been a topic of discussion and challenge in the construction of hydrological models (Wang et al. 2022). Hydrological models replicate the elements of the water balance to gain insights into different hydrological processes (Ogden 2021). Therefore, reliable outputs have always been desired by researchers. This requirement necessitates the proper representation of model inputs at suitable spatial and temporal scales (Mtibaa and Asano 2022), and the main source of uncertainty is prediction in many hydrological models (Chao et al. 2018). Therefore, accurate data and sufficient representation of precipitation variability are crucial for precise simulation outputs and effective decision-making (Mtibaa and Asano 2022). Striking a balance between the precision and spatial depiction of precipitation data in hydrological models continues to pose difficulties (Dembele et al. 2020). In practice, rainfall is traditionally measured using rain gauges (Sun et al., 2018), providing precise observations at a point scale with a diameter of 20 centimeters (Mtibaa and Asano 2022). These measurements are often considered as the actual amount of rainfall (Tapiador et al. 2017); however, when observations are transformed into a broader scale due to the scattered rain gauge network, errors and uncertainties are introduced (Deng et al. 2019). This uncertainty increases in environments with complex topography (such as mountainous watersheds) and specific water and weather conditions (such as convective rainfall, tropical storm rainfall (Li et al., 2019; Schroeer et al., 2018), as in such regions, rainfall shows significant spatial and temporal variability, making the assumption of a uniform rainfall distribution in hydrological models incorrect (Cho et al. 2009). Furthermore, rain gauge measurements may indicate significant errors in the presence of strong winds and severe rainfall conditions, leading to water losses and erratic behavior of the rain gauge system (Zhong et al. 2016; Cha and Lee 2021). These issues result in an incorrect spatial representation of precipitation that can impact simulation model outputs (Mtibaa and Asano 2022). Therefore, various organizations have taken steps towards collecting observational data and have provided models with regression equations and interpolation, offering models in different temporal and spatial scales under the name of gridded data. Improvements in gridded data, such as reanalyzed and satellite data, address the shortcomings of rain gauges by facilitating the spatial observation of precipitation. These technologies offer rainfall estimates with high spatial and temporal resolution on both global and regional levels (Sun et al. 2018), making them valuable alternatives for rainfall data in hydrological models. Various methods and algorithms used in gridded precipitation products have significantly improved the quality of precipitation estimates and significantly reduced the estimation errors (Chao et al., 2018; Goudenhoofdt and Delobbe, 2009; Yang et al., 2017). Research has assessed the effectiveness of gridded precipitation products through the Soil and Water Assessment Tool (SWAT) model (Arnold et al. 1998), recognized as one of the most prominent models available (Tan et al., 2021). The findings of these studies have demonstrated the suitability of these alternatives for simulating hydrological processes. Studies have shown that climatic gridded data has been extensively used in the SWAT model. Among the studies in this field, Dembele and colleagues (2020) found, after analyzing 17 gridded datasets, that the PERSIANN-CDR dataset performs best for streamflow in the predominantly semi-arid Valle de Bravo (VRB) river basin. This indicates that PERSIANN-CDR can be a suitable choice for hydrological modeling in this region (Dembele et al. 2020). Hafizi and Sorman (2022) evaluated 13 gridded precipitation datasets for hydrological modeling in a mountainous basin and found that the hydraulic performance of ERA5 is very good at daily and monthly scales, demonstrating its reliability for hydrological modeling (Hafizi and Sorman 2022). Ougahi and Mahmood (2022) evaluated satellite-based precipitation datasets with hydrological simulations in the Chennab river basin and reported that ERA5 shows very good hydrological performance at daily and monthly scales, supporting its reliability for hydrological modeling (Ougahi and Mahmood 2022). In Iran, Parisuoj and colleagues (2018) evaluated two rainfall sources, TRMM and PERSIANN, using the HEC-HMS rainfall-runoff model for simulating runoff in the Voshmgir dam basin. The results indicated higher simulation accuracy for TRMM rainfall compared to PERSIANN at the daily scale (Parisuoj et al. 2018). Similarly, Eini and colleagues (2018) used the SWAT model to simulate runoff at a monthly scale and evaluated the performance of two rainfall sources, CRU and NCEP-CFSR, in the Maharlou basin; they concluded that the Nash-Sutcliffe index averages around 0.91 for observational and computational data, with the CRU source performing better than the NCEP-CFSR source for surface runoff estimation (Eini et al. 2018).Rafiee and colleagues (2024) focused on modeling runoff using gridded and observational precipitation data in the Beshar basin in Iran; the results indicated the appropriateness of using gridded data for hydrological modeling (Rafiee et al. 2024).Few studies have highlighted the role of rain gauge station dispersion and the effectiveness of various types of precipitation data as alternatives for simulating runoff. Particularly in Iran, a country repeatedly facing severe floods and consequent damages. This study examines the role of rain gauge dispersion data and its impact on precipitation-runoff modeling to simulate precipitation-runoff with the semi-distributed SWAT model using observational data and to evaluate and compare the accuracy of two common gridded precipitation products, PERSIANN-CDR and ERA5, on daily and monthly scales to determine the superior dataset in terms of hydrological performance; the study also looks into the analysis of runoff coefficients in the study area to identify the strengths and weaknesses of precipitation-runoff modeling in the study area.

3.1 The study area

The Dez River watershed in southwestern Iran is located in a semi-arid mountainous region between geographical longitudes of 48 degrees 20 minutes 24 seconds and 50 degrees 18 minutes 59 seconds east and latitudes of 32 degrees 33 minutes 29 seconds north. It is considered a sub-basin of the Great Karun Basin and is categorized under the larger sub-basin of the Persian Gulf and the Sea of Oman (Figure 1). The majority of the precipitation in the fall and winter seasons falls as snow, and its melting continues from late winter to late spring, providing a significant portion of the annual surface runoff in this watershed. The total area of the watershed is approximately 17,057 square kilometers. The elevation of the basin ranges from a maximum of 4,075 meters to a minimum of 317 meters, with an average elevation of 1,676 and a maximum elevation frequency of 1,829 meters (Khoshchehreh et al. 2021). Figure 2 illustrates the hypsometric curves and altitude meters of the Dez River watershed. In this study, based on available and reliable statistics, four rain gauge stations, Telezang, Dez Dam, Tang-e Panj Bakhtiari, and Sepid Dasht Sazar, were selected for the period from 01/01/2008 to 31/12/2019.

The gridded data are cell-based data for all points within their spatial coverage and are used to estimate climatic parameters such as precipitation. The availability of these data has increased the accessibility of precipitation datasets at various spatial and temporal resolutions at a global or semi-global scale (Tapiador et al. 2012). Table 1 show the characteristics of grid data used in this research.

Table 1- Characteristics of grid data used in this research

Dataset Name	Full Name	Source	Temporal Cov.	Temporal Res.	Spatial Cov.	Spatial Res.	Download Link
PERSIANN-CDR	Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks- Climate Data Record	S, G	1983-NRT	3 hours	60◦N- 60◦S	0.25^o	https://chrsdata.eng.uci.edu
ERA5	Enhanced Global Dataset for The Land Component of The Fifth Generation of European ReAnalysis	R	1950-NRT	Hourly	Global	0.25^o	https://cds.climate.copernicus.eu

In the data source column, S, G, and R stand for satellite, rain gauge, and reanalysis, respectively.

3.2.1 ERA5 dataset

Processed reanalysis data, combined from model prediction and observational data, are used. The model predictions, also known as the first guess, are obtained based on the model input data and a defined mathematical relationship for the model (Dee et al. 2011; Balsamo et al. 2015). Since model predictions are always associated with uncertainties, these initial predictions are controlled by available observational data received by the forecasting center. The model is then optimized to improve its predictions and reduce errors. Therefore, the production of processed reanalysis data takes place in two stages. First, the model predicts a time step (6 hours ahead) based on the past hours’ data available in the prediction center archives; then, the results of this prediction are combined with the observational data of the same time step received from different sources to produce the processed reanalysis data. As each data source mentioned above has different spatial distributions on Earth’s surface, the model accuracy varies in different regions of the Earth.

ERA5 is the latest generation of reanalysis data (fifth generation) developed by the European Centre for Medium-Range Weather Forecasts (ECMWF); its key improvements over ERA-Interim include a better spatial grid (31 kilometers compared to 79 kilometers), higher temporal resolution (1 hour compared to 3 hours), more vertical levels (137 compared to 60), a new NWP model (IFS_Cycle_41r2), and increased data availability for data assimilation (Urraca et al., 2018). The dataset provides coverage from 1950 to near real-time. In this study, daily ERA5 precipitation data with 0.25-degree spatial resolution was used, and the data extraction was done using the ECMWF_Web_API.

3.2.2 PERSIANN-CDR dataset

The PERSIANN-CDR dataset is prepared jointly by the University of California and NOAA, covering the period from 1983 to the present (Ashouri et al. 2015; Duan et al. 2016) and its precipitation estimates are derived from IR and MW satellite data. The PERSIANN algorithm combines IR and PMW data to estimate rainfall amounts. Each of these data sources is derived from GOE and LOE satellite data. Additionally, the PERSIANN algorithm utilizes an artificial neural network algorithm to estimate precipitation. PERSIANN model output without the use of microwave data in areas without such data and without data correction is referred to as PERSIANN-B1. To reduce errors in PERSIANN-B1 precipitation data products while maintaining high spatial and temporal resolution, GPCP monthly data is used. The corrected PERSIANN data maintains a good correlation with GPCP data at a monthly scale. The final product produced is called PERSIANN-CDR. PERSIANN-CDR has been developed to provide a stable, long-term, high-resolution, global dataset for studying rainfall changes and trends, especially severe rainfall events due to climate and natural changes. In this study, daily PERSIANN-CDR precipitation data with a spatial resolution of 0.25 degrees was utilized.

3.3. Quantile Mapping for Bias Correction

Generally, raw predictions from global numerical models have biases in the mean and range of forecasts, which prevent direct evaluation of model results. Bias refers to systematic differences between predicted and observed data. Therefore, bias correction methods are often used in many studies for post-processing model predictions to remove biases and improve the performance of weather models (Tao et al. 2014). Quantile Mapping (QM) is one of the suitable methods for bias correction. QM corrects biases by aligning the observed and estimated values within each time scale using a cumulative distribution function (Mendoza Paz and Willems 2022). This algorithm is commonly used to correct systematic distributional biases in precipitation outputs from weather models (Cannon et al. 2015). This method involves replacing the predicted values with observed values in similar percentiles. (Voisin et al. 2010) To perform bias correction using the quantile mapping method, it is necessary to calculate and plot the cumulative distribution function or the probability density function of observational and gridded precipitation data. This allows for the determination of the bias at different quantiles; a schematic representation of the concept of the Quantile Mapping bias correction method is shown in Error! Reference source not found..

In this study, the quantile mapping method has been applied to each of the daily gridded precipitation datasets for the period from 2008 to 2019, and both datasets used in the research have been bias-corrected using this method.

3.4 SWAT model

There are many models for hydrological evaluation of watersheds that are employed by water resource managers, decision-makers, and researchers. Watershed models can be broadly categorized based on how they address the spatial components related to the hydrology of the watershed. Integrated models consider the entire watershed as a single unit without taking into account spatial variations in processes, inputs, boundary conditions, or hydrological characteristics. In contrast, distributed models account for spatial variations by solving governing equations for each pixel in the watershed grid.

The SWAT model is a watershed-scale model introduced by Arnold and colleagues in 1998. This model is designed to predict the effects of various land management practices on water, sediment, and agricultural chemical yields at the level of complex and large watersheds, considering soils, land use, and long-term management conditions. Instead of incorporating regression relationships to describe the relationships between input and output variables, the SWAT model requires information on weather, soil characteristics, topography, vegetation cover, and management and land-use practices at the watershed level.

Physical processes related to water movement, sediment transport, plant growth, nutrient cycling, etc., are directly simulated from input parameters. One advantage of this approach is that watersheds lacking measured data can still be simulated. Additionally, the relative effects of input data (changes in management practices, weather, vegetation cover, etc.) on water quality and other relevant variables can be quantified. The SWAT model uses accessible input parameters and is computationally efficient. It is capable of simulating large and complex watersheds with various management strategies without requiring excessive time. This model allows users to study long-term impacts. For this analysis, the model outputs must span several decades.

To simulate a watershed, it must be divided into several sub-watersheds. Utilizing sub-watersheds in simulations is particularly beneficial when different areas of the watershed have varying soils or land uses, as their heterogeneity and differences can influence the watershed's hydrology. Figure 4 illustrates the components and input/output data of the SWAT model."

In this study, the SWAT model was implemented for observational data and the ERA5 and PERSIANN-CDR models. Calibration and uncertainty analysis of the model results were conducted using the SUFI2 algorithm in the SWAT-CUP software. The period used for hydrological assessment was based on existing network data and observational data validated by the Ministry of Energy from 2008 to 2019, where the years 2008 and 2009 were used for model warming, the years 2010 to 2016 for calibration, and the years 2017 to 2019 for validation.

3.5 Meteorological Evaluation Indices

The daily data from network products in the cells containing the rain gauges were compared to the gauge measurements using three quantitative statistics: (1) the Pearson correlation coefficient (R), which assesses the extent of linearity; and (2) BIAS, which determines the average inclination of the integrated dataset. In this context, positive values indicate that the network product data exceed the gauge measurements. (3) Root mean squares error (RMSE) calculates the differences (Mtibaa and Asano 2022). Various evaluation and classification indices presented in Table 2 were used to compare and assess the networked dataset for determining the accuracy of this data in the study area. In this table, Pi represents the predicted value, Gi represents the observed value, P̅ is the average of the predicted values, and G̅ is the average of the observed values.

3.6 hydrological evaluation indicators

Coefficient of Determination (R²)

The coefficient of determination, or R², is a statistical index that measures the closeness of data to the fitted regression line. R² indicates what percentage of the variability in the dependent variable is explained by the independent variable. In other words, the coefficient of determination shows how much of the variability in the dependent variable is influenced by the corresponding independent variable, while the rest of the variability in the dependent variable is related to other factors. The coefficient of determination is always between 0 and 1. A value of 0 indicates that the model does not explain any of the variability of the response data around its mean, while a value of 1 indicates that the model explains all the variability of the response data around its mean. Theoretically, if R² equals 1, it indicates that all observed values will equal the fitted values, and all data points will be located on the fitted line. The coefficient of determination cannot determine whether the fitted model is biased, so residual plots need to be evaluated. R² does not indicate whether a regression model is appropriate or not. It is noteworthy that a good regression model might have a low R², and a regression model with a high R² may not fit the data adequately. R² is useful for determining how well a regression equation fits the data.

3.6.1 Nash-Sutcliffe Efficiency (NSE)

The Nash-Sutcliffe Efficiency criterion (Equation (3.1)) is one of the performance metrics used in hydrology (Nash and Sutcliffe, 1970). NSE focuses on variability in measured flow time series (Guse et al., 2017). This index gives more weight to low flows than to high flows (Pfannerstill et al. 2014).

In the above equation, P_i is the predicted value, G_i is the observed value, and G̅ is the mean of the observed data.

3.6.KGE

The KGE criterion (Hoshin Vijai et al. 1999; Gupta et al. 2009) is based on the decomposition of NSE into three separate components. In this equation, α represents the rate of change of the standard deviation of the modeled flow time series (σ_S) and the measured flow (σ_O), β indicates the rate of bias between the average modeled flow (μ_S) and the observed flow (μ_o), and r represents the correlation coefficient based on Pearson's correlation for analyzing the trends in changes between the modeled flow time series and the measured flow. The ideal value for all three KGE components is one. Each KGE component specifically elucidates a part of the hydrological behavior, which is why examining KGE components separately is important (Guse et al. 2017). The range of KGE values extends from negative infinity to one, with a value of one indicating excellent agreement between simulated and observed flow (Gupta et al. 2009).

4.1 Comparison of Gridded Precipitation Datasets

Figure 5 shows the scatter plot of gridded precipitation data versus rain gauge data. As can be observed, based on the R values, the ERA5 product exhibits better conformity across all stations compared to the PERSIANN-CDR dataset, and the highest correlation corresponds to the Sepid Dasht Sazr station.

Based on the RMSE values, which indicate the differences between observed values and gridded data, it is clearly concluded that at the Tele Zang and Dez Dam stations, the P-CDR product is more suitable, while at the Tang Panjeh and Sepid Dasht stations, the ERA5 product has demonstrated better performance. The orientation of the gridded products is also indicated by the BIAS values. Therefore, it can be observed that at all stations, the P-CDR dataset underestimates the data, while the ERA5 database overestimates it. This aspect is visually represented by the aggregation of data concerning the line y=x in the figure.

4.2 Analysis of Flood Runoff Coefficient in the Study Basin

Table 3 lists the name of the flood based on its occurrence date, base flow, peak event flow, flood duration in days, rainfall from all stations, the average rainfall from the stations within the study area, and the runoff coefficient based on the rainfall of each station and the average runoff coefficient from the stations. The base flow values presented in the table reflect the moisture conditions of the basin before the occurrence of the flood, and the flood duration indicates whether the rainfall was widespread or localized within the study area; in other words, the longer the flood duration, the more widespread the rainfall has been, indicating a greater percentage of the basin area contributing to runoff.

As seen in Table 3, the average runoff coefficient in the Dez basin exhibits varied performance during different events. By comparing the runoff coefficients, base flow, and peak flood flow, it can be concluded that in many events, the spatial distribution of rainfall was not considered. As a result, the shown runoff coefficient may not represent the actual rate of conversion of rainfall to runoff. In other words, the distribution of precipitation that occurred in the Dez basin is such that the distribution of rain gauge stations was unable to capture the spatial distribution of precipitation, resulting in discrepancies between the shown rainfall rates at the study area's stations and the actual rates that occurred. This issue can lead to errors in estimating the runoff coefficient. For instance, in the floods on 03/11/2010 and 14/11/2018, it is observed that despite the base flow levels within a close range indicating similar moisture conditions and an average rainfall of 35 mm, the runoff coefficients differ significantly at 0.023 and 0.462, respectively. In contrast, the flood on 03/11/2010 suggests a widespread rainfall event, with the peak flow lasting for 25 days, but the recorded average rainfall from rain gauges showed 35 mm, leading to a calculated runoff coefficient of 0.023 for this event.

Similarly, in the floods on 08/01/2016 and 28/02/2019, which had the same flood duration of 8 days with similar moisture conditions and comparable peak flow rates, the average rainfalls were 61.62 mm and 9 mm, resulting in runoff coefficients of 0.059 and 0.368, respectively. This highlights the variability in rainfall distribution for example, in the flood on 11/01/2008, considering the base flow and peak flow rates of 142 and 159 cubic meters per second, respectively, it indicates that the rainfall event was localized and occurred near the rain gauge stations. All of this rainfall infiltrated, resulting in a very minimal amount of runoff produced. Meanwhile, the existing base flow in the river indicates that the moisture conditions of the area suggest that there had been rainfall in the days prior, and thus the runoff generated from this rainfall should have been more than the measured amount. In the flood on 12/04/2016, it is observed that the equivalent height of runoff is approximately 117 mm, and given the 25-day duration of the flood, it indicates a widespread rainfall over the basin. The moisture conditions of the basin are not dry, leading to a peak flow increase of up to about 5800 cubic meters per second. This not only suggests heavy rainfall in the region but also indicates snowmelt within the study area and an increase in peak flow rates.

4.3 Model Configuration and Preparation

After collecting information and correcting the biases in the gridded data using the QM method, input files were prepared, and the model execution steps included creating sub-basins, HRUs, and finally executing the model in the simulation period. In this study, to configure the model and delineate the sub-basins and waterways, a DEM map with a cell resolution of 30 meters was used. Accordingly, a total of 95 sub-basins were identified for the watershed. The next step in modeling involves introducing the necessary informational layers to form the hydrological response units to the model. These layers include land use maps, soil maps along with soil characteristic tables, and slope maps. Ultimately, by integrating the three layers of soil, land use, and slope, the watershed was divided into 2029 hydrological response units, and the required information was introduced to the model.

4.4 Evaluation of Simulated Daily Flow

Figure 6 illustrates the time series of observed runoff, simulated runoff, and the uncertainty ranges of model parameters, alongside the fitting coefficients and model parameter uncertainties using daily observed precipitation and P-CDR and ERA5 precipitation products. The evaluation and analysis of the desirability of model result uncertainties were conducted using the d-factor and p-factor indices. As shown in Figure 6, which depicts the simulation uncertainty of the model at a daily time step during the calibration and validation periods for both observational and gridded data, it is observed that in simulating the observed precipitation data, the p-factor values for the calibration and validation periods are 19% and 16%, respectively. This means that in the mentioned periods, 19% and 16% of the observed data fall within the 95% confidence band simulated by the model. Table 4 shows the performance of the SWAT model in estimating runoff from gridded precipitation models at a daily scale. Based on the R² values, it is determined what percentage of the observed data is explained by the modeled data; therefore, according to this parameter in the calibration period, 74% of the daily observed precipitation data was estimated by the SWAT model. However, based on this parameter's value in the validation period, it can be concluded that this model is unreliable for future periods.

According to the NSE values in the table, despite the fact that daily precipitation-runoff modeling with the SWAT model shows a high NSE coefficient, the low value of this coefficient in the validation period indicates that the modeling is not appropriate, and the model is essentially overfitted. The NSE coefficient gives more weight to maximum runoff values; thus, to provide a more comprehensive assessment and consider flow parameters, the KGE criterion was also used. The KGE values for simulating observational data in the calibration period (0.37) and in the validation period (-0.2) indicate that, in addition to the model's overfitting, the model performed better in estimating discharge during the calibration period than in estimating overall flow parameters.

Regarding the reasons for the poor performance of the model in simulating flow, it can be stated that, as mentioned above, a lack of an adequate rain gauge network and the spatial distribution of precipitation, along with potential issues in data collection and the measurement of rainfall and discharge, have prevented the model from accurately matching precipitation values with their corresponding discharge values. This situation has led the model to overestimate the simulated discharge values, resulting in unacceptable outcomes during the validation period, thereby making the precipitation-runoff modeling of observational data unsuitable.

Table 4- Comparing the performance of rain gridded data in the estimation of runoff by SWAT model on a daily scale

Precipitation Dataset	Period	R2	NSE	KGE	P-factor	d-factor
Gauge	Cal.	0.74	0.52	0.37	0.19	0.48
Gauge	Valid.	0.16	-0.01	-0.2	0.12	0.14
ERA5	Cal.	0.24	0.24	0.27	0.75	0.96
ERA5	Valid.	0.54	0.52	0.53	0.71	0.51
P-CDR	Cal.	0.29	0.29	0.35	0.89	0.89
P-CDR	Valid.	0.65	0.59	0.52	0.78	0.5

It is worth noting that the rainfall data from P-CDR and ERA5 products exhibit better performance compared to observational values during both the calibration and validation periods, as indicated in Figure 6 and Table 4.

Additionally, the lower evaluation indices in the simulation of observational precipitation-runoff data can be further understood when examining Figure 7, which shows daily precipitation and corresponding runoff for the years 2008-2009, 2014-2015 and 2016-2017 which is equivalent to the solar years 1387, 1393 and 1395. As clearly evident, the watershed's response to the available data is significantly different; for example, during the flood in March 2008, the measured rainfall was approximately 25 mm, resulting in a flood discharge of about 320 cubic meters per second, while in December of the same year, 47 mm of rainfall resulted in a discharge of 270 cubic meters per second.

This discrepancy highlights the inadequacy of the rain gauge stations or the problems associated with data recording.

Furthermore, despite improved spatial distribution of precipitation data, the low performance of gridded data can also be attributed to the role of snowmelt within the study area. For example, in the years 2008 and 2013, between May and November, there was no recorded precipitation, but base flow gradually decreased on non-rainy days at the beginning of the dry season, this indicates and demonstrates snowmelt in the study area. Therefore, it can be concluded that despite the significant improvement in the results of precipitation-runoff modeling with grid data, the low performance of this data can be attributed to the role of snowmelt within the study area. It is also noteworthy that some precipitation-runoff models, including the SWAT model, which do not accurately account for snowmelt in their computational core, are not recommended for modeling. This finding aligns with the research by Jaiswal et al. (2020).

4.5 Uncertainty in flow modeling

The uncertainty analysis was conducted for the three simulations using the SUFI2 method. The P-Factor and d-Factor values, which were calculated to assess prediction uncertainty, are presented in Table 4. The P-factor, which quantifies the proportion of observed streamflow within the 95% Prediction Uncertainty (95PPU) range, showed that the P-CDR model exhibited the least prediction uncertainty, achieving a P-factor of 0.89 during calibration and 0.78 during validation. Following this, the Era5 model recorded P-factors of 0.75 for calibration and 0.71 for validation. In contrast, the Gauge model failed to achieve an acceptable level of prediction uncertainty, as its P-factor was below

4.6 Evaluation of Simulated Monthly Flow

Figure 8 shows the time series of observed runoff, simulated runoff, and the uncertainty range of the model parameters along with the fit coefficients and uncertainty coefficients of the model parameters with input from observed monthly precipitation and the P-CDR and ERA5 precipitation products. As mentioned, the model has been overfitted in estimating runoff with input from rainfall measurement data., meaning that the calibration period indicators are suitable, but they are inadequate during the validation period. What appears to be important is that based on the NSE and KGE indices, the ERA5 model has demonstrated good performance in both calibration and validation periods. The P-CDR product also shows reasonable results in estimating monthly simulations, ranking second. The model's effective performance in estimating monthly runoff with input from gridded rainfall data supports the hypothesis that there may be errors in data collection and recording.

5.1 Evaluation of ERA5 and P-CDR precipitation datasets

Previous research (Zhu et al. 2018; Chen et al. 2020; Islam et al. 2020) indicates that passive microwave and infrared sensors are more effective at detecting high-intensity precipitation events while often overlooking low-intensity ones. This is attributed to their higher signal-to-noise ratio and the more suitable convective conditions for microwave measurements (Tan et al. 2015; Wu et al. 2018). (Chen et al. 2013) and Satgé et al., (2020) identified three reasons for the superior performance of satellite and reanalysis-based precipitation products at elevated intensity levels: (1) Rainfall during low-intensity events that is accurately detected by infrared and microwave sensors may evaporate before reaching gauging stations, (2) the gauges may lack the sensitivity to register low-intensity precipitation, and (3) there could be deficiencies in the algorithms used. Moreover, the gauge observations, which serve as ground truth for performance assessment, are prone to various errors, including those caused by wind and wetting-evaporation losses (Peng et al., 2020). Consequently, the impact of uncertainties in gauge observations on evaluation performance should not be overlooked.

The daily estimates from ERA5 were typically higher than expected, leading to an excess of water input in the SWAT model. In the area of special regions, ERA5 precipitation data in mainland shows specific biases (Jiang et al. 2021a). The studies by Bandhauer et al., (2022); Challis et al., (2021); Y. Jiang et al., (2021) and Lavers et al., (2022), collectively highlight the tendency of ERA5 to overestimate precipitation in various regions, raising concerns about the accuracy of the dataset for climate monitoring and other applications. However, the specific mechanisms underlying these discrepancies remain to be fully understood. This knowledge gap presents an opportunity for future research to investigate the factors contributing to the overestimation of precipitation in ERA5.

Overall, the literature suggests that ERA5 still shows biases and overestimations in specific regions and climates despite improvements in precipitation representation. This emphasizes the importance of further investigation into the root causes of these biases and the development of potential adjustments to enhance the accuracy of ERA5 precipitation data. Future research avenues may include comparisons with field measurements and implementing bias correction techniques to improve the reliability of ERA5 data for diverse uses. Moreover, localized studies in areas where inconsistencies exist between ERA5 data and actual observations can offer valuable insights into the factors leading to precipitation overestimations in the ERA5 dataset.

Daily P-CDR estimates were typically lower than actual values, leading to decreased water input into the SWAT model, leading to a low detection of precipitation errors mainly due to sampling errors in passive microwave imagery (Kubota et al. 2009). Additionally, the constraints of infrared and microwave sensors in infrared and microwave sensors in recording mountainous precipitation events are related to local bias and convection (Hayashi et al. 2021; Nepal et al. 2021). These estimated uncertainties indicate the limitations of calibrating precipitation estimates that rely exclusively on 24-hour accumulated rainfall values according to the GPCP rain gauge-based product, they recommended updating the bias correction algorithm (Mega et al. 2019) and the mountain precipitation correction scheme (Yamamoto and Shige 2015), or utilizing additional local rain gauges (Deng et al. 2019) to enhance product quality and accurately detect heavy rainfall. The findings of Wang and Zhao, (2022) show that the P-CDR dataset underestimates precipitation in this region. This indicates that the P-CDR dataset may exhibit a systematic bias in underestimation, which can have implications for water resource management and other related applications in the study area.

Evaluation of Flood Runoff Coefficient

The reported runoff coefficient may not accurately represent the actual rate at which rainfall converts to runoff. This issue arises because the placement of rain gauge stations often fails to capture the spatial distribution of precipitation effectively (Lees et al. 2021). Calculating the runoff coefficient is crucial for understanding this transformation However, if the distribution of rain gauge stations is insufficient, it may not provide an accurate depiction of how precipitation varies across different areas, complicating the assessment of rainfall (Sharma et al. 2020). Research has shown that these inaccuracies in measuring precipitation can lead to a reported runoff coefficient that does not truly reflect the conversion process.

5.2 The impact of choosing different precipitation datasets on simulating streamflow

The three precipitation datasets assessed in this study exhibited varying metrics, detection accuracy, and spatial representations of rainfall (ERA5, P-CDR, and Gauge). These variations led to differing rainfall inputs—such as the fraction of dry days and daily rainfall volume across the catchment—into the SWAT model, resulting in different streamflow simulation outputs. As indicated by the statistical measures presented in Section 3.6, it was evident that the gridded precipitation datasets (P-CDR and ERA5) surpassed the gauge data in simulating daily streamflow at the DEZ Dam catchment. This superior performance is likely due to enhanced spatial distribution of the precipitation data, which contradicts the conclusions drawn by Deng et al., (2019); Mtibaa and Asano, (2022), who stated that the superior performance is related to the gauge-based model.

The gridded precipitation products that excelled in validating precipitation also showed strong performance in simulating streamflow. Among these, the P-CDR model outperformed the others, including the gauge-simulated model. This strong performance is attributed to its generally high data quality and its effectiveness in capturing the spatial variability of precipitation within the catchment area. In contrast, the gauge-simulated model exhibited slightly weaker performance due to the sparse and uneven distribution of rainfall gauges, which hinders the accurate representation of the spatial and temporal variability of precipitation, consistent with the findings of Moges et al., (2022).

The study underscores the importance of addressing data network inadequacies, refining modeling approaches, and utilizing advanced datasets and technologies for future hydrological modeling efforts. These implications aim to enhance the predictive accuracy and applicability of hydrological models in varying geographical contexts, ultimately contributing to better water resource management and flood risk mitigation; Accordingly, this study evaluates the hydrology of two gridded databases, PERSIANN-CDR and ERA5, in the Dez Dam watershed. Daily rainfall data from the Dez Dam rain gauge, Talezang, Tang-e Panj Bakhtiari, Sepid Dasht Sezar, and the Talezang hydrometric station from 2008 to 2019 were used, and rainfall and runoff modeling were conducted using the semi-distributed SWAT simulation model. The objective of this study was to assess the impact of spatial distribution of rainfall and rain gauges in the watershed and evaluate the capabilities of gridded databases in hydrological modeling.

The results indicate that due to the lack of an adequate rain gauge network and the spatial distribution of precipitation, along with the potential for errors in data collection and the measurement of rainfall and discharge, the model is unable to correlate rainfall amounts with their corresponding discharge values. This issue has led the model to be overfitted to provide appropriate responses for the parameters of simulated discharge values, resulting in unacceptable outcomes during the validation period, and consequently rendering the observed precipitation-runoff modeling unsuitable.

Therefore, based on the R², NSE, and KGE parameter values during the calibration and validation periods, the gridded daily rainfall data simulated by the SWAT model using the P-CDR and ERA5 products, due to their higher spatial coverage, perform better than observed values in the daily time range. In daily hydrological evaluation, the PERSIANN-CDR dataset exhibits better performance, while the ERA5 dataset shows superior capabilities in monthly hydrological evaluation for rainfall-runoff modeling.

In conclusion, the analysis demonstrated that improving the spatial distribution of precipitation datasets is crucial for accurately estimating flood runoff coefficients and enhancing hydrological modeling outcomes. Future studies should focus on integrating advanced data assimilation techniques and innovative modeling approaches that account for the spatial heterogeneity of precipitation and the complex interactions within the watershed.

Funding The authors declare that no funding was received for this study.

Abdolmanafi A, Saghafian B, Aminyavari S (2021) Evaluation of global ensemble prediction models for forecasting medium to heavy precipitations. Meteorology and Atmospheric Physics 133:15–26. https://doi.org/10.1007/S00703-020-00731-8/METRICS
Arnold JG, Srinivasan R, Muttiah RS, Williams JR (1998) LARGE AREA HYDROLOGIC MODELING AND ASSESSMENT PART I: MODEL DEVELOPMENT1. JAWRA Journal of the American Water Resources Association 34:73–89. https://doi.org/10.1111/J.1752-1688.1998.TB05961.X
Ashouri H, Hsu KL, Sorooshian S, et al (2015) PERSIANN-CDR: Daily precipitation climate data record from multisatellite observations for hydrological and climate studies. Bull Am Meteorol Soc 96:69–83. https://doi.org/10.1175/BAMS-D-13-00068.1
Balsamo G, Albergel C, Beljaars A, et al (2015) ERA-Interim/Land: a global land surface reanalysis data set. Hydrol Earth Syst Sci 19:389–407. https://doi.org/10.5194/hess-19-389-2015
Bandhauer M, Isotta F, Lakatos M, et al (2022) Evaluation of daily precipitation analyses in E-OBS (v19.0e) and ERA5 by comparison to regional high-resolution datasets in European regions. International Journal of Climatology 42:727–747. https://doi.org/10.1002/JOC.7269
Cannon AJ, Sobie SR, Murdock TQ (2015) Bias correction of GCM precipitation by quantile mapping: How well do methods preserve changes in quantiles and extremes? J Clim. https://doi.org/10.1175/JCLI-D-14-00754.1
Cha SM, Lee SW (2021) Advanced hydrological streamflow simulation in a watershed using adjusted radar-rainfall estimates as meteorological input data. J Environ Manage 277:111393. https://doi.org/10.1016/J.JENVMAN.2020.111393
Challis JK, Popick H, Prajapati S, et al (2021) Occurrences of Tire Rubber-Derived Contaminants in Cold-Climate Urban Runoff. Environ Sci Technol Lett 8:. https://doi.org/10.1021/acs.estlett.1c00682
Chao L, Zhang K, Li Z, et al (2018) Geographically weighted regression based methods for merging satellite and gauge precipitation. J Hydrol (Amst) 558:275–289. https://doi.org/10.1016/J.JHYDROL.2018.01.042
Chen H, Yong B, Shen Y, et al (2020) Comparison analysis of six purely satellite-derived global precipitation estimates. J Hydrol (Amst) 581:. https://doi.org/10.1016/j.jhydrol.2019.124376
Chen S, Hong Y, Cao Q, et al (2013) Similarity and difference of the two successive V6 and V7 TRMM multisatellite precipitation analysis performance over China. Journal of Geophysical Research Atmospheres 118:. https://doi.org/10.1002/2013JD019964
Cho J, Bosch D, Lowrance R, et al (2009) Effect of Spatial Distribution of Rainfall on Temporal and Spatial Uncertainty of SWAT Output. Trans ASABE 52:1545–1556. https://doi.org/10.13031/2013.29143
Dee DP, Uppala SM, Simmons AJ, et al (2011) The ERA-Interim reanalysis: configuration and performance of the data assimilation system. Quarterly Journal of the Royal Meteorological Society 137:553–597. https://doi.org/10.1002/qj.828
Dembele M, Schaefli B, Van De Giesen N (2020) Suitability of 17 gridded rainfall and temperature datasets for large-scale hydrological modelling in West Africa. Hydrol Earth Syst Sci 24:. https://doi.org/10.5194/hess-24-5379-2020
Deng P, Zhang M, Bing J, et al (2019) Evaluation of the GSMaP_Gauge products using rain gauge observations and SWAT model in the Upper Hanjiang River Basin. Atmos Res 219:153–165. https://doi.org/10.1016/J.ATMOSRES.2018.12.032
Duan Z, Liu J, Tuo Y, et al (2016) Evaluation of eight high spatial resolution gridded precipitation products in Adige Basin (Italy) at multiple temporal and spatial scales. Science of The Total Environment 573:1536–1553. https://doi.org/10.1016/j.scitotenv.2016.08.213
Eini MR, Javadi S, Delavar M (2018) Evaluating the performance of CRU and NCEP CFSR global reanalysis climate datasets, in hydrological simulation by SWAT model, Case Study: Maharlu basin. Iran-Water Resources Research 14:32–44
Gorjizade A, Akhoond-Ali AM, Shahbazi A, Moridi A (2020) Evaluation and Comparison of High Spatial Resolution Gridded Precipitation by TRMM, ERA5, and PERSIANN-CCS Datasets on the Upstream of the Maroon Basin, Iran. AUT Journal of Civil Engineering 4:529–542. https://doi.org/10.22060/AJCE.2020.17043.5610
Goudenhoofdt E, Delobbe L (2009) Evaluation of radar-gauge merging methods for quantitative precipitation estimates. Hydrol Earth Syst Sci 13:195–203. https://doi.org/10.5194/HESS-13-195-2009
Gupta H V., Kling H, Yilmaz KK, Martinez GF (2009) Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. J Hydrol (Amst). https://doi.org/10.1016/j.jhydrol.2009.08.003
Guse B, Pfannerstill M, Gafurov A, et al (2017) Identifying the connective strength between model parameters and performance criteria. Hydrol Earth Syst Sci. https://doi.org/10.5194/hess-21-5663-2017
Hafizi H, Sorman AA (2022) Assessment of 13 Gridded Precipitation Datasets for Hydrological Modeling in a Mountainous Basin. Atmosphere (Basel) 13:. https://doi.org/10.3390/atmos13010143
Hayashi Y, Tebakari T, Hashimoto A (2021) A comparison between global satellite mapping of precipitation data and high-resolution radar data – a case study of localized torrential rainfall over japan. Journal of Disaster Research 16:. https://doi.org/10.20965/JDR.2021.P0786
Hoshin Vijai G, Sorooshian S, Ogou PY (1999) Status of Automatic Calibration for Hydrologic Models: Comparison with Multilevel Expert Calibration. J Hydrol Eng
Islam MA, Yu B, Cartwright N (2020) Assessment and comparison of five satellite precipitation products in Australia. J Hydrol (Amst) 590:. https://doi.org/10.1016/j.jhydrol.2020.125474
Jiang Q, Li W, Fan Z, et al (2021a) Evaluation of the ERA5 reanalysis precipitation dataset over Chinese Mainland. J Hydrol (Amst) 595:125660. https://doi.org/10.1016/J.JHYDROL.2020.125660
Jiang Y, Yang K, Shao C, et al (2021b) A downscaling approach for constructing high-resolution precipitation dataset over the Tibetan Plateau from ERA5 reanalysis. Atmos Res 256:. https://doi.org/10.1016/j.atmosres.2021.105574
Khoshchehreh M, Ghomeshi M, Shahbazi A, et al (2021) Spatial and temporal evaluation of global rainfall products in a data-scarce region: The Dez Basin, Iran. Journal of Water and Land Development No 48:148–161. https://doi.org/10.24425/jwld.2021.136158
Kubota T, Ushio T, Shige S, et al (2009) Verification of High-Resolution Satellite-Based Rainfall Estimates around Japan Using a Gauge-Calibrated Ground-Radar Dataset. Journal of the Meteorological Society of Japan Ser II 87A:203–222. https://doi.org/10.2151/JMSJ.87A.203
Lavers DA, Simmons A, Vamborg F, Rodwell MJ (2022) An evaluation of ERA5 precipitation for climate monitoring. Quarterly Journal of the Royal Meteorological Society 148:. https://doi.org/10.1002/qj.4351
Lees T, Buechel M, Anderson B, et al (2021) Benchmarking data-driven rainfall-runoff models in Great Britain: A comparison of long short-term memory (LSTM)-based models with four lumped conceptual models. Hydrol Earth Syst Sci 25:. https://doi.org/10.5194/hess-25-5517-2021
Li X, Long D, Han Z, et al (2019) Evapotranspiration Estimation for Tibetan Plateau Headwaters Using Conjoint Terrestrial and Atmospheric Water Balances and Multisource Remote Sensing. Water Resour Res 55:8608–8630. https://doi.org/10.1029/2019WR025196
Mega T, Ushio T, Matsuda T, et al (2019) Gauge-Adjusted Global Satellite Mapping of Precipitation. IEEE Transactions on Geoscience and Remote Sensing 57:. https://doi.org/10.1109/TGRS.2018.2870199
Mendoza Paz S, Willems P (2022) Uncovering the strengths and weaknesses of an ensemble of quantile mapping methods for downscaling precipitation change in Southern Africa. J Hydrol Reg Stud 41:101104. https://doi.org/10.1016/J.EJRH.2022.101104
Moges DM, Kmoch A, Uuemaa E (2022) Application of satellite and reanalysis precipitation products for hydrological modeling in the data-scarce Porijõgi catchment, Estonia. J Hydrol Reg Stud 41:. https://doi.org/10.1016/j.ejrh.2022.101070
Mtibaa S, Asano S (2022) Hydrological evaluation of radar and satellite gauge-merged precipitation datasets using the SWAT model: Case of the Terauchi catchment in Japan. J Hydrol Reg Stud 42:101134. https://doi.org/10.1016/J.EJRH.2022.101134
Nepal B, Shrestha D, Sharma S, et al (2021) Assessment of GPM-Era satellite products’ (IMERG and GSMaP) ability to detect precipitation extremes over mountainous country nepal. Atmosphere (Basel) 12:. https://doi.org/10.3390/atmos12020254
Ogden FL (2021) Geohydrology: Hydrological Modeling. Encyclopedia of Geology: Volume 1-6, Second Edition 6:457–476. https://doi.org/10.1016/B978-0-08-102908-4.00115-6
Ougahi JH, Mahmood SA (2022) Evaluation of satellite-based and reanalysis precipitation datasets by hydrologic simulation in the Chenab river basin. Journal of Water and Climate Change 13:. https://doi.org/10.2166/wcc.2022.410
Parisuoj P, Goharnejad H, Moazami S (2018) Rainfall-Runoff Hydrologic Simulation Using Adjusted Satellite Rainfall Algorithms, a Case Study: Voshmgir Dam Basin. Golestan. Iran-Water Resources Research 14:140–159. https://doi.org/10.22055/JHS.2017.13318
Pfannerstill M, Guse B, Fohrer N (2014) Smart low flow signature metrics for an improved overall performance evaluation of hydrological models. J Hydrol (Amst). https://doi.org/10.1016/j.jhydrol.2013.12.044
Rafiee MR, Rad S, Mahbod M, et al (2024) Modeling run-off flow hydrographs using remote sensing data: an application to the Bashar basin, Iran. Journal of Water and Climate Change 15:1490–1506. https://doi.org/10.2166/WCC.2024.378
Satgé F, Defrance D, Sultan B, et al (2020) Evaluation of 23 gridded precipitation datasets across West Africa. J Hydrol (Amst) 581:. https://doi.org/10.1016/j.jhydrol.2019.124412
Schroeer K, Kirchengast G, Sungmin O (2018) Strong Dependence of Extreme Convective Precipitation Intensities on Gauge Network Density. Geophys Res Lett 45:8253–8263. https://doi.org/10.1029/2018GL077994
Sharma S, Chen Y, Zhou X, et al (2020) Evaluation of GPM-Era satellite precipitation products on the southern slopes of the central Himalayas against rain gauge data. Remote Sens (Basel) 12:. https://doi.org/10.3390/rs12111836
Sun Q, Miao C, Duan Q, et al (2018) A Review of Global Precipitation Data Sets: Data Sources, Estimation, and Intercomparisons. Reviews of Geophysics 56:79–107. https://doi.org/10.1002/2017RG000574
Tan ML, Ibrahim AL, Duan Z, et al (2015) Evaluation of six high-resolution satellite and ground-based precipitation products over Malaysia. Remote Sens (Basel) 7:1504–1528. https://doi.org/10.3390/rs70201504
Tao Y, Duan Q, Ye A, et al (2014) An evaluation of post-processed TIGGE multimodel ensemble precipitation forecast in the Huai river basin. J Hydrol (Amst). https://doi.org/10.1016/j.jhydrol.2014.04.040
Tapiador FJ, Navarro A, Levizzani V, et al (2017) Global precipitation measurements for validating climate models. Atmos Res 197:1–20. https://doi.org/10.1016/J.ATMOSRES.2017.06.021
Tapiador FJ, Turk FJ, Petersen W, et al (2012) Global precipitation measurement: Methods, datasets and applications. Atmos Res 104–105:70–97
Voisin N, Schaake JC, Lettenmaier DP (2010) Calibration and Downscaling Methods for Quantitative Ensemble Precipitation Forecasts. Weather Forecast. https://doi.org/10.1175/2010WAF2222367.1
Wang J;, Zhang J;, Xiong N;, et al (2022) Spatial and Temporal Variation, Simulation and Prediction of Land Use in Ecological Conservation Area of Western Beijing. Remote Sensing 2022, Vol 14, Page 1452 14:1452. https://doi.org/10.3390/RS14061452
Wang Y, Zhao N (2022) Evaluation of Eight High-Resolution Gridded Precipitation Products in the Heihe River Basin, Northwest China. Remote Sens (Basel) 14:. https://doi.org/10.3390/rs14061458
Wu L, Xu Y, Wang S (2018) Comparison of TMPA-3B42RT Legacy Product and the Equivalent IMERG Products over Mainland China. Remote Sensing 2018, Vol 10, Page 1778 10:1778. https://doi.org/10.3390/RS10111778
Yamamoto MK, Shige S (2015) Implementation of an orographic/nonorographic rainfall classification scheme in the GSMaP algorithm for microwave radiometers. Atmos Res 163:36–47. https://doi.org/10.1016/J.ATMOSRES.2014.07.024
Yang Z, Hsu K, Sorooshian S, et al (2017) Merging high-resolution satellite-based precipitation fields and point-scale rain gauge measurements—A case study in Chile. Journal of Geophysical Research: Atmospheres 122:5267–5284. https://doi.org/10.1002/2016JD026177
Zhong L, Zhang Z, Chen L, et al (2016) Application of the Doppler weather radar in real-time quality control of hourly gauge precipitation in eastern China. Atmos Res 172–173:109–118. https://doi.org/10.1016/J.ATMOSRES.2015.12.016
Zhu H, Li Y, Huang Y, et al (2018) Evaluation and hydrological application of satellite-based precipitation datasets in driving hydrological models over the Huifa river basin in Northeast China. Atmos Res. https://doi.org/10.1016/j.atmosres.2018.02.022

No competing interests reported.

Download PDF

Editorial decision: Revision requested
05 Sep, 2024
Reviews received at journal
03 Sep, 2024
Reviews received at journal
30 Aug, 2024
Reviewers agreed at journal
24 Aug, 2024
Reviewers agreed at journal
23 Aug, 2024
Reviewers agreed at journal
19 Aug, 2024
Reviewers agreed at journal
16 Aug, 2024
Reviewers invited by journal
16 Aug, 2024
Editor assigned by journal
16 Aug, 2024
Submission checks completed at journal
13 Aug, 2024
First submitted to journal
04 Aug, 2024

You are reading this latest preprint version

Analysis of Spatial Distribution of Precipitation Using Hydrological Modeling of Watersheds

Status:

Version 1

Abstract

Figures

Introduction

Materials and methods

3.1 The study area

3.2.1 ERA5 dataset

3.2.2 PERSIANN-CDR dataset

3.3. Quantile Mapping for Bias Correction

3.4 SWAT model

3.5 Meteorological Evaluation Indices

3.6 hydrological evaluation indicators

3.6.1 Nash-Sutcliffe Efficiency (NSE)

3.6.KGE

Results

4.1 Comparison of Gridded Precipitation Datasets

4.2 Analysis of Flood Runoff Coefficient in the Study Basin

4.3 Model Configuration and Preparation

4.4 Evaluation of Simulated Daily Flow

4.5 Uncertainty in flow modeling

4.6 Evaluation of Simulated Monthly Flow

Discussion

5.1 Evaluation of ERA5 and P-CDR precipitation datasets

5.2 The impact of choosing different precipitation datasets on simulating streamflow

Conclusion

Declarations

References

Additional Declarations

Status:

Version 1