Uncertainty In Rainfall Estimates Research Articles

Enhancing runoff simulation precision in the critical zone through spatiotemporal interpolation of areal rainfall with matrix decomposition

AbstractModelling hydrological process in the critical zone not only contributes to a better understanding of interactions across different Earth surface spheres but also holds significant practical implications for water resource management and disaster prevention. Rainfall‐runoff simulation in critical zones is particularly challenging due to the amalgamation of temporal and spatial complexity, rainfall variability, and data limitations. As a pivotal input variable of hydrological models, accurate estimation of areal rainfall is critical to successful runoff simulation. However, most estimation methods ignore temporal information, thereby increasing uncertainty in rainfall estimation and constraining the precision of rainfall‐runoff simulation. In this study, the matrix decomposition‐based estimation method (F‐SVD), which considers the spatial and temporal correlation of the rainfall process is employed to estimate areal rainfall. The superiority of the method in producing two‐dimensional rainfall information is evaluated through its application in runoff simulation with the Xin'anjiang model. The simulation results of selected flood events in the Jianxi basin in southeast China, spanning from 2009 to 2019, are compared with those of two widely used rainfall estimation methods, namely Arithmetical Mean (AM) and Thiessen Polygons (TP). The results show that (1) F‐SVD not only produces the highest Pearson correlation coefficient between rainfall and runoff series but also reduces the number of flood events with abnormal rainfall‐runoff relationships; (2) the Xin'anjiang model based on F‐SVD achieves the highest Nash‐Sutcliffe efficiency and lowest Relative Error, and performs best in simulating peak flow and its occurrence time as compared to AM and TP. This study contributes to a finer characterization of watershed rainfall distribution, enhancing the accuracy and sharpness of runoff simulation. It provides reliable data support for critical zone research and offers a scientific foundation for rationally allocating and managing water resources.

Optimizing Radar-Based Rainfall Estimation Using Machine Learning Models

Weather radar research has produced numerous radar-based rainfall estimators based on climate, rainfall intensity, a variety of ground-truthing instruments and sensors (e.g., rain gauges, disdrometers), and techniques. Although each research direction gives improvement, their collective application in an operational sense still yields uncertainty in rainfall estimation at times. This study aims to explore the concept of implementing Machine Learning (ML) models in optimizing the radar-based rainfall estimations at the bin level from a group of estimator. The Canadian King City C-Band radar was used with a GEONOR T-200B rain gauge (a total of 263 sample points) to establish a group of polarimetric-based rainfall estimators (R(Z), R(Z, ZDR), R(KDP)). The estimators were used to train three ML models, namely Decision Tree, Random Forest, and Gradient Boost, to choose the optimal rainfall estimators based on radar variables (Z, ZDR, KDP). Data from the Canadian Exeter C-Band radar and a Texas Electronics TE525 tipping bucket gauge at a different location were used to verify the ML models and compare their results to the most commonly used Z-R relations. The verification process shows promising results for the ML models, specifically the Gradient Boost model. These encouraging results need to be further explored with more sample points to further refine the ML models.

Joint treatment of point measurement, sampling and neighborhood uncertainty in space-time rainfall mapping

The importance of representing the spatial structure of rainfall accurately has been emphasized in various hydrological studies. It has also been widely acknowledged that there is a need to account for uncertainty in rainfall input. Common approaches focus on accounting for either point measurement or sampling uncertainty in rainfall estimation. We present a method that jointly considers three sources of uncertainty affecting the space-time mapping of rainfall: point measurement, sampling and neighborhood uncertainty. To our knowledge, neighborhood uncertainty has not been included in any prior rainfall uncertainty analysis. We generated an ensemble of 400 realizations of daily rainfall fields at a 2 km × 2 km spatial resolution for a catchment in Western Denmark (1055 km2). At the core of our method is the sequential Gaussian simulation (SGS) technique. Results indicate that our approach is able to reproduce key statistical features of the rainfall distribution. We examined the impact of different spatial (grid and catchment) and temporal supports (one day, one month, 5-year period) on the overall uncertainty. We also quantified the effect of each uncertainty source on rainfall field uncertainty. Finally, we compared our simulation results with those of a parallel expert elicitation study. We found that the expert elicitation uncertainty for average catchment rainfall in a 5-year period was considerably larger than quantified in our study (CV of 1.1% vs. 5%). An even larger discrepancy was found for the 5-year average of gauge rainfall, where expert elicitation resulted in a value that was an order of magnitude higher (CV of 0.2% vs. 2%). Possible reasons for this gap are discussed.

Analysis of the Influence of Rainfall Spatial Uncertainty on Hydrological Simulations Using the Bootstrap Method

Rainfall stations of a certain number and spatial distribution supply sampling records of rainfall processes in a river basin. Uncertainty may be introduced when the station records are spatially interpolated for the purpose of hydrological simulations. This study adopts a bootstrap method to quantitatively estimate the uncertainty of areal rainfall estimates and its effects on hydrological simulations. The observed rainfall records are first analyzed using clustering and correlation methods and possible average basin rainfall amounts are calculated with a bootstrap method using various combinations of rainfall station subsets. Then, the uncertainty of simulated runoff, which is propagated through a hydrological model from the spatial uncertainty of rainfall estimates, is analyzed with the bootstrapped rainfall inputs. By comparing the uncertainties of rainfall and runoff, the responses of the hydrological simulation to the rainfall spatial uncertainty are discussed. Analyses are primarily performed for three rainfall events in the upstream of the Qingjian River basin, a sub-basin of the middle Yellow River; moreover, one rainfall event in the Longxi River basin is selected for the analysis of the areal representation of rainfall stations. Using the Digital Yellow River Integrated Model, the results show that the uncertainty of rainfall estimates derived from rainfall station network has a direct influence on model simulation, which can be conducive to better understand of rainfall spatial characteristic. The proposed method can be a guide to quantify an approximate range of simulated error caused by the spatial uncertainty of rainfall input and the quantified relationship between rainfall input and simulation performance can provide useful information about rainfall station network management in river basins.

Uncertainty analysis of radar rainfall estimates over two different climates in Iran

ABSTRACTAccuracy of rainfall quantification is one of the most important concerns in meteorological and hydrological modelling. Satellites and weather radars can provide meteorological information with higher temporal and spatial resolution than ground stations. Rain gauges measure rain rate directly; however, weather radars estimate rain rate by converting radar reflectivity aloft to rain rate at ground level. This conversion with a power law relation between radar reflectivity and rain rate could be altered from place to place or in various precipitation types. This variety may be the source of errors and uncertainty of radar rainfall estimates. One way to assess the uncertainty of radar rainfall is simulation of rainfall fields. In this article, after calibrating two radars located in the south-western and northern parts of Iran, uncertainty of rainfall estimates of these radars has been analysed using the Gaussian Copula model. Reliability of this model was examined for 10% of the rainfall events that were not included in the simulation process. Obtained results of the current research indicate that recalibration of radars can considerably reduce bias and root mean error. In addition, the Copula-based model can generate rainfall fields with similarly spatial structures to those of observed rainfall data.

Assessment of TRMM-based TMPA-3B42 and GSMaP precipitation products over India for the peak southwest monsoon season

Reliable information of rainfall and its variability at finer spatio-temporal scales over the Indian monsoon region is crucial for various applications in hydrology, meteorology, water resource management, and evaluation of high-resolution numerical model outputs. In this study, the Tropical Rainfall Measuring Mission (TRMM)-based high-resolution multisatellite rainfall products namely, TRMM Multisatellite Precipitation Analysis (TMPA)-3B42 and Global Satellite Mapping of Precipitation (GSMaP) are evaluated against recently released gauge-based India Meteorological Department (IMD) gridded data set. Both the research (delayed mode) and near real-time rainfall products of TMPA-3B42 (V7 and RT) and GSMaP (MVK and NRT) are evaluated over India at a daily scale for the peak southwest monsoon months of July–August. Two study periods (2000–2010 and 2009–2013) are used to accommodate the analyses based on data availability. Comparison of TMPA-3B42 (V7 and RT) and GSMaP-MVK with IMD gauge-based data for 2000–2010 shows that these data sets are able to represent large-scale monsoon rainfall spatial features, but are having region-specific biases. The interannual variability of monsoon rainfall averaged over India is also reasonably captured. However in terms of biases, TMPA-3B42RT overestimates rainfall by about 21% and GSMaP-MVK underestimates by about 22% at all-India scale. In terms of continuous skill metrics, TMPA-3B42V7 is closer to the IMD gauge-based observations possibly due to gauge adjustments. Categorical verification against gauge-based observations shows that the satellite-based data sets have difficulties in rainfall detection primarily over the northern Jammu & Kashmir regions, northeast India, and southern peninsular India. At subregional scale within India, satellite rainfall estimates show rather lower skill score in rainfall identification and larger uncertainty in rainfall estimation over the northeast India. Comparison of near real-time products of TMPA and GSMaP for July–August months of 2009–2013 shows that TMPA-3B42RT overestimates and GSMaP-NRT underestimates rainfall over most parts of India. Even though TMPA-3B42 (V7 and RT) and GSMaP-MVK rainfall data sets show similar diurnal variations over the Indian monsoon region, TMPA-3B42RT (GSMaP-MVK) shows larger (smaller) magnitude in diurnal cycle as compared to TMPA-3B42V7. Results from different skill metrics suggest that both research and near real-time versions of TMPA products can be used with relatively higher confidence as compared to GSMaP over the Indian region during peak monsoon period.

Uncertainty in rainfall input data in a conceptual water balance model: effects on outputs and implications for predictability

<p>As hydrological models become more prevalent in water resources planning and management, increasing levels of detail and precision are needed. Currently, reliable models that simulate the hydrological behavior of a basin are indispensable; however, it is also necessary to know the limits of the predictability and reliability of the model outputs. The present study evaluates the influence of uncertainty in the main input variable of the model, rainfall, on the output uncertainty of a hydrological model. Using concepts of identifiability and sensitivity, the uncertainty in the model structure and parameters was estimated. Then, the output uncertainty caused by uncertainties in i) the rainfall amounts and ii) the periods of the rainfall was determined. The main conclusion is that uncertainty in rainfall estimation during rainy periods produces greater output uncertainty. However, in non-rainy periods, the output uncertainty is not very sensitive to the uncertainty in rainfall. Finally, uncertainties in rainfall during the basin filling and emptying periods (Apr. – Jun. and Sep. – Nov., respectively) alter the uncertainty in subsequent periods. Therefore, uncertainties in these periods could result in limited ranges of model predictability.</p><p> </p><p><strong>Resumen</strong></p><p>Los modelos hidrológicos se han vuelto cada vez más necesarios en la planificación y gestión de recursos hídricos, donde un aumento en los niveles de detalle y precisión es necesario. Actualmente disponer de modelos para simular el comportamiento hidrológico de una cuenca resulta indispensable, sin embargo, también es necesario conocer los límites de predictibilidad y de confiabilidad de las salidas de un modelo. En este estudio se evalúan la influencia de la incertidumbre en la principal entrada de un modelo, la precipitación, sobre la incertidumbre de las salidas de un modelo hidrológico. Utilizando conceptos de identificabilidad y sensibilidad se estima la incertidumbre de los parámetros y estructura de un modelo. Luego, la incertidumbre en las salidas causadas por incertidumbre en i) los montos de precipitación, ii) los períodos de precipitación fue calculada. Como conclusiones se obtuvo que la incertidumbre en la estimación de la precipitación en períodos de lluvia produce mayor incertidumbre sobre las salidas. En períodos no lluviosos, la incertidumbre de las salidas es poco sensible a incertidumbre sobre las precipitaciones. Finalmente, incertidumbres en periodos de llenado y vaciado (Abril-Junio y Septiembre-Noviembre respectivamente) afectan la incertidumbre en las salidas en los períodos subsecuentes. Por lo tanto incertidumbres en aquellos períodos pueden resultar en rangos limitados de predictibilidad de un modelo.</p>

Optimal design of rain gauge network in the Middle Yarra River catchment, Australia

AbstractRainfall data are a fundamental input for effective planning, designing and operating of water resources projects. A well‐designed rain gauge network is capable of providing accurate estimates of necessary areal average and/or point rainfall estimates at any desired ungauged location in a catchment. Increasing network density with additional rain gauge stations has been the main underlying criterion in the past to reduce error and uncertainty in rainfall estimates. However, installing and operation of additional stations in a network involves large cost and manpower. Hence, the objective of this study is to design an optimal rain gauge network in the Middle Yarra River catchment in Victoria, Australia. The optimal positioning of additional stations as well as optimally relocating of existing redundant stations using the kriging‐based geostatistical approach was undertaken in this study. Reduction of kriging error was considered as an indicator for optimal spatial positioning of the stations. Daily rainfall records of 1997 (an El Niño year) and 2010 (a La Niña year) were used for the analysis. Ordinary kriging was applied for rainfall data interpolation to estimate the kriging error for the network. The results indicate that significant reduction in the kriging error can be achieved by the optimal spatial positioning of the additional as well as redundant stations. Thus, the obtained optimal rain gauge network is expected to be appropriate for providing high quality rainfall estimates over the catchment. The concept proposed in this study for optimal rain gauge network design through combined use of additional and redundant stations together is equally applicable to any other catchment. © 2014 The Authors. Hydrological Processes published by John Wiley & Sons Ltd.

Impact of uncertainty in rainfall estimation on the identification of rainfall thresholds for debris flow occurrence

Estimation of rainfall intensity–duration thresholds, used for the identification of debris flow/landslide triggering rainfall events, has been traditionally based on raingauge observations. The main drawback of using information from gauges is that the measurement stations are usually located far away from the debris flow initiation areas. In complex terrain where debris flows take place, the spatial variability of rainfall can be very high and this translates in large uncertainty of raingauge-based estimates of debris flow triggering rainfall. This work focuses on the assessment of the impact of rainfall estimation uncertainty on identification and use of rainfall thresholds for debris flow occurrence. The Upper Adige River basin, Northern Italy, is the area of study. A detailed database of more than 400 identified debris flow initiation points during the period 2000–2010 and a raingauge network of 100 stations comprise the database used for this work. The methodology examines the intensity–duration thresholds derived from a set of raingauges that are assumed to be located at debris flow initiation points (DFRs) and an equivalent set of raingauges assumed to have the role of the closest (to debris flow) available measurement (MRs). A set of reference rainfall thresholds is used to identify the rainfall events at DFRs that “triggered” debris flows (i.e. exceed the threshold). For these same events, the corresponding rainfall thresholds are derived from MR observations. Comparison between the rainfall thresholds derived from DFRs and MRs revealed that uncertainty in rainfall estimation has a major impact on estimated intensity–duration thresholds. Specifically, the results showed that thresholds estimated from MR observations are consistently underestimated. Evaluation of the estimated thresholds for warning procedures showed that while detection is high, the main issue is the high false alarm ratio, which limits the overall accuracy of the procedure. Overall performance on debris flow prediction was shown to be good for low rainfall thresholds and poor for high rainfall thresholds examined. Lastly, it was found that filtering out rainfall events with duration <12h may improve bias in estimated thresholds and performance for high rainfall thresholds.

GLUE based marine X-band weather radar data calibration and uncertainty estimation

The Generalized Likelihood Uncertainty Estimation methodology (GLUE) is investigated for radar rainfall calibration and uncertainty assessment. The method is used to calibrate radar data collected by a Local Area Weather Radar (LAWR).In contrast to other LAWR data calibrations, the method combines calibration with uncertainty estimation. Instead of searching for a single set of calibration parameters, the method uses the observations to construct distributions of the calibration parameters. These parameter sets provide valuable knowledge of parameter sensitivity and the uncertainty.Two approaches are analyzed; the static calibration approach, where the LAWR is calibrated once for a long period and the dynamic approach, where the estimate is continuously adjusted based on ground observations.The analysis illustrates that the static calibration performs insufficiently, whereas the dynamic adjustment improves the performance significantly.It is found that even if the dynamic adjustment method is used the uncertainty of rainfall estimates can still be significant.

A scenario-incorporating analysis of the propagation of uncertainty to flash flood simulations

Summary This study proposes a methodology to analyze the propagation of uncertainty arising from radar rainfall and model parameter estimation to the flow simulations obtained from flash flood modeling. For this purpose an ensemble approach that describes the error from radar rainfall estimation and a novel approach that considers the seasonality and variability of catchment conditions in the estimation of model parameters have been implemented. The methodology analyzes the separate effect of each source of uncertainty and their interaction on runoff simulations. The methodology has been applied in the Besos river basin, Catalonia (Spain) with a drainage area of 1020 km 2 . The results show that the effect of parameter estimation uncertainty produces a smaller spread in the model realizations than the effect caused by the uncertainty in rainfall estimation.

Copula‐based uncertainty modelling: application to multisensor precipitation estimates

AbstractThe multisensor precipitation estimates (MPE) data, available in hourly temporal and 4 km × 4 km spatial resolution, are produced by the National Weather Service and mosaicked as a national product known as Stage IV. The MPE products have a significant advantage over rain gauge measurements due to their ability to capture spatial variability of rainfall. However, the advantages are limited by complications related to the indirect nature of remotely sensed precipitation estimates. Previous studies confirm that efforts are required to determine the accuracy of MPE and their associated uncertainties for future use in hydrological and climate studies. So far, various approaches and extensive research have been undertaken to develop an uncertainty model. In this paper, an ensemble generator is presented for MPE products that can be used to evaluate the uncertainty of rainfall estimates. Two different elliptical copula families, namely, Gaussian and t‐copula are used for simulations. The results indicate that using t‐copula may have significant advantages over the well‐known Gaussian copula particularly with respect to extremes. Overall, the model in which t‐copula was used for simulation successfully generated rainfall ensembles with similar characteristics to those of the ground reference measurements. Copyright © 2010 John Wiley &amp; Sons, Ltd.

Comparison of Two Methods for Estimating the Sampling-Related Uncertainty of Satellite Rainfall Averages Based on a Large Radar Dataset

The uncertainty of rainfall estimated from averages of discrete samples collected by a satellite is assessed using a multiyear radar dataset covering a large portion of the United States. The sampling-related uncertainty of rainfall estimates is evaluated for all combinations of 100-, 200-, and 500-km space domains; 1-, 5-, and 30-day rainfall accumulations; and regular sampling time intervals of 1, 3, 6, 8, and 12 h. These extensive analyses are combined to characterize the sampling uncertainty as a function of space and time domain, sampling frequency, and rainfall characteristics by means of a simple scaling law. Moreover, it is shown that both parametric and nonparametric statistical techniques of estimating the sampling uncertainty produce comparable results. Sampling uncertainty estimates, however, do depend on the choice of technique for obtaining them. They can also vary considerably from case to case, reflecting the great variability of natural rainfall, and should therefore be expressed in probabilistic terms. Rainfall calibration errors are shown to affect comparison of results obtained by studies based on data from different climate regions and/or observation platforms.