Rainfall Estimation Errors Research Articles

IMERG-E satellite rainfall estimation error decomposition for early warning use in several parts of Indonesia

An early warning can be used with rainfall input from satellite estimation to prevent and reduce losses due to flood risk. IMERG-E satellite rainfall estimation has fast time latency and can be used for this purpose. We tried to separate systematic and random components of the error of IMERG-E rainfall estimation. An additive error model used for this decomposition assumes that the linear function can fit the relationship between satellite and references measurements. The references measurements are hourly hellman rain gauges in several parts of Indonesia. The result shows that more stations have dominant systematic components in JJA than in DJF seasons. In 1-hr temporal, the random components are more dominant (>50%) in almost all study areas. Systematic components are dominant at Padang Pariaman and Jembrana stations (>50%) for all periods. Coarse spatial aggregation (0.2° and 0.3°) increased systematic components and lowered temporal (3-hr and 6-hr) decrease systematic components throughout the study area. Dominant random components in almost the entire study area show IMERG-E challenging to correct, and caution is needed when used as input for early warning in most parts of Indonesia. Bias reduction IMERG-E in Indonesia needs to be developed considering different seasons and locations.

Sub-hourly resolution quality control of rain-gauge data significantly improves regional sub-daily return level estimates.

This research demonstrates how the use of high-resolution rain-gauge data for quality control (QC) significantly changes extreme rainfall estimates, with implications in scientific, meteorological and engineering applications. Current open QC algorithms only consider data at hourly or daily accumulations. Here we present the first open QC algorithm utilising sub-hourly rain-gauge data from official networks at a national, multi-decade scale. We use data from 1,301 rain-gauges in Great Britain (GB) to develop a threshold-based methodology for sub-hourly QC that can be used to complement existing, freely available hourly QC methods by developing an algorithm for sub-hourly QC that uses monthly thresholds for 1 hr, 15 min and 1 min rainfall totals. We then evaluated the effect of combining these QC procedures on rainfall distributions using graphical and statistical methods, with an emphasis on extreme value analysis. We demonstrate that the additional information in sub-hourly rainfall allows our new QC to remove spuriously large values undetected by existing methods which generate errors in extreme rainfall estimates. This results in statistically significant differences between extreme rainfall estimates for 15 min and 1 hr accumulations, with smaller differences found for 6 and 24 hr totals. We also find that extremes in the distributions of 15 min and 1 hr rainfall accumulations tend to grow more rapidly with return period than for longer accumulation periods. We observe similarities between the shape parameter populations for 15 min and 1 hr rainfall accumulations, suggesting that hourly records may be used to improve shape parameter estimates for extreme sub-hourly rainfall in GB. Sub-hourly QC moderates unrealistically large return level estimates for short-duration rainfall, with beneficial impacts on data required for the design of urban drainage infrastructure and the validation of high-resolution climate models.

Research on Rainfall Estimation Based on Improved Kalman Filter Algorithm

In order to solve the rainfall estimation error caused by various noise factors such as clutter, super refraction, and raindrops during the detection process of Doppler weather radar. This paper proposes to improve the rainfall estimation model of radar combined with rain gauge which calibrated by common Kalman filter. After data preprocessing, the radar data should be classified according to the precipitation intensity. And then, they are respectively substituted into the improved filter for calibration. The state noise variance and the measurement noise variance can be adaptively calculated and updated according to the input observation data during this process. Then the optimal parameter value of each type of precipitation intensity can be obtained. The state noise variance and the measurement noise variance could be assigned optimal values when filtering the remaining data. This rainfall estimation based on semi-adaptive Kalman filter calibration not only improves the accuracy of rainfall estimation, but also greatly reduces the amount of calculation. It avoids errors caused by repeated calculations, and improves the efficiency of the rainfall estimation at the same time.

Spatial and Time Warping for Gauge Adjustment of Rainfall Estimates

Many satellite-based estimates use gauge information for bias correction. In general, bias-correction methods are focused on the intensity error and do not explicitly correct possible position or timing errors. However, position and timing errors in rainfall estimates can also lead to errors in the rainfall occurrence or the intensity. This is especially true for localized rainfall events such as the convective rainstorms occurring during the rainy season in sub-Saharan Africa. We investigated the use of warping to correct such errors. The goal was to gauge-adjust satellite-based estimates with respect to the position and the timing of the rain event, instead of its intensity. Warping is a field-deformation method that transforms an image into another one. We compared two methods, spatial warping focusing on the position errors and time warping for the timing errors. They were evaluated on two case studies: a synthetic rainfall event represented by an ellipse and a rain event in southern Ghana during the monsoon season. In both cases, the two warping methods reduced significantly the respective targeted (position or timing) errors. In the southern Ghana case, the average position error was decreased by about 45 km by the spatial warping and the average timing error was decreased from more than 1 h to 0.2 h by the time warping. Both warping methods also improved the continuous statistics on the intensity: the correlation went from 0.18 to at least 0.62 after warping in the southern Ghana case. The spatial warping seems more interesting because of its positive impact on both position and timing errors.

Automatic Quality Control of Crowdsourced Rainfall Data With Multiple Noises: A Machine Learning Approach

AbstractIn geophysics, crowdsourcing is an emerging nontraditional environmental monitoring approach that supports data acquisition from individual citizens. However, because of the involvement of undertrained citizens and imprecise low‐cost sensors, crowdsourced data applications suffer from different types of noises that can deteriorate the overall monitoring accuracy. In this study, we propose a machine learning approach for automatic crowdsourced data quality control (CSQC) that detects and removes noisy data inputs in spatially and temporally discrete crowdsourced observations coming from both fixed‐point sensors (e.g., surveillance cameras) and moving sensors (e.g., moving cars/pedestrians). We design a set of features from original and interpolated rainfall data and use them to train and test the CSQC models using both supervised and unsupervised machine learning algorithms. The performances of the CSQC models under various scenarios assuming no retraining are also tested (hereafter referred to as transferability). The results based on synthetic but realistic data show that the CSQC models can significantly reduce the overall rainfall estimate errors. Under the stationary assumption, the CSQC models based on both supervised and unsupervised algorithms perform well in noisy data identification and overall rainfall estimation error reduction; however, if the model is transferred to other cities with different rainfall patterns or noise compositions (without retraining), supervised multilayer perceptrons (MLPs) show the best performance.

Optimization of rain gauge network using multi-criteria decision analysis and entropy approaches: case of Tekeze River basin, northwestern Ethiopia

Rainfall is an important hydro-meteorological variable that has various applications in water resource planning and development. Adequate rain gauge network gives immediate and precise rainfall data that are crucial for effective and economical water resource management. This study describes a multi-criteria decision analysis (MCDA) in combination with kriging and entropy methods to optimize rain gauge networks in the Tekeze River basin, northwestern Ethiopia. The Tekeze River basin is a trans-boundary river basin in the northwestern part of Ethiopia having a drainage area 86 510 km2. The MCDA was used for selecting suitable locations for additional rain gauges in the basin. The kriging and entropy methods were used to determine the optimal number of rain gauge stations in the basin. Initial analysis showed that the existing 63 rain gauge stations in the basin are not adequate and result in about 12% rainfall estimation error. Optimization of rain gauge network in the Tekeze River basin resulted in 216 additional gauging stations. The total optimal number of rain gauges for the Tekeze River would thus be 279. The results of the study revealed that the combined use of the MCDA, kriging, and entropy methods is useful to optimize the spatial distribution and number of rain gauge stations in a basin.

Hydrologic evaluation of bias corrected CMORPH rainfall estimates at the headwater catchment of the Zambezi River

Uncorrected and bias corrected CMORPH satellite-based rainfall estimates were evaluated for application in the Representative Elementary Watershed (REW) modelling approach in the Kabompo Basin, a headwater catchment of the Zambezi River. Results of modelling (2008–2013) by use of raingauge rainfall served as benchmark to the simulation results. The uncorrected CMORPH rainfall show overestimation (>20%) and underestimation (<−15%) biases for the 2000–2015 period. A linear-multiplicative bias correction scheme was applied to remove the bias. When applied to the REW model, bias corrected CMORPH rainfall improved the match between observed and simulated streamflow. Improvements in streamflow simulation are reported for highest discharges and streamflow totals. Objective function values for Relative Volume Error and for Nash Sutcliffe Efficiency improved significantly by the bias correction. Errors in uncorrected CMORPH rainfall estimates cause large mismatch between simulated and observed streamflow, however mismatch reduces notably for bias corrected satellite rainfall. Optimized REW parameter values changed by > 50% when gauge-based rainfall inputs are replaced by uncorrected CMORPH estimates. However, parameter values, remained within physically acceptable ranges for bias corrected CMORPH, for intermediate flows and for the wet season. This study shows that bias correction of CMORPH rainfall in the Kabompo Basin is critical for successful application in rainfall-runoff modelling.

Evaluation of satellite-based products for extreme rainfall estimations in the eastern coastal areas of China

ABSTRACTRemotely sensed rainfall plays an important role in providing efficient approaches for global or regional rainfall analysis. However, the accuracy of satellite-based products is mainly affected by the errors in sensor observation and retrieval algorithms, particularly with respect to extreme rainfall estimates. The objective of this study is to evaluate the accuracy of satellite-based products in capturing rainfall extremes. The eastern coastal areas of China were chosen as the case study area to compare the accuracy of three mainstream satellite-based products with respect to extreme rainfall estimates during 2003–2015 period. This included the Tropical Rainfall Measurement Mission (TRMM) rainfall product 3B42V7, the Climate Prediction Centre Morphing technique RAW (CMORPH-RAW), and the CMORPH bias-corrected product (CMORPH-CRT). In general, all satellite-based products demonstrated numerous errors in extreme rainfall estimates. Based on three different indices of extreme rainfall, it was observed that the satellite-based products underestimated the amounts of rainfall extremes and their respective average values. It was noted that CMORPH-RAW demonstrated the largest relative bias (RB) and underestimated the average extreme rainfall by −31% to −35%. Additionally, all satellite-based products exhibited poor capabilities in capturing the variations in hourly extreme rainfall processes. Finally, a simple potential flood index was developed to simulate the potential flood areas in the eastern coastal areas of China. We found that the potential flood areas can be simulated by combining the potential flood index with the amounts of rainfall derived by satellite-based products.

A Step towards Integrating CMORPH Precipitation Estimation with Rain Gauge Measurements

Accurate daily rainfall estimation is required in several applications such as in hydrology, hydrometeorology, water resources management, geomorphology, civil protection, and agriculture, among others. CMORPH daily rainfall estimations were integrated with rain gauge measurements in Brazil between 2000 and 2015, in order to reduce daily rainfall estimation errors by means of the statistical objective analysis scheme (SOAS). Early comparisons indicated high discrepancies between daily rain gauge rainfall measurements and respective CMORPH areal rainfall accumulation estimates that tended to be reduced with accumulation time span (e.g., yearly accumulation). Current results show CMORPH systematically underestimates daily rainfall accumulation along the coastal areas. The normalized error variance (NEXERVA) is higher in sparsely gauged areas at Brazilian North and Central-West regions. Monthly areal rainfall averages and standard deviation were obtained for eleven Brazilian watersheds. While an overall negative tendency (3 mm·h−1) was estimated, the Amazon watershed presented a long-term positive tendency. Monthly areal mean precipitation and respective spatial standard deviation closely follow a power-law relationship for data-rich watersheds, i.e., with denser rain gauge networks. Daily SOAS rainfall accumulation was also used to calculate the spatial distribution of frequencies of 3-day rainfall episodes greater than 100 mm. Frequencies greater than 3% were identified downwind of the Peruvian Andes, the Bolivian Amazon Basin, and the La Plata Basin, as well as along the Brazilian coast, where landslides are recurrently triggered by precipitation.

Assessing spatially dependent errors in radar rainfall estimates for rainfall-runoff simulation

Weather radar been widely employed to measure precipitation and to predict flood risks. However, it is still not considered accurate enough because of radar errors. Most previous studies have focused primarily on removing errors from the radar data. Therefore, in the current study, we examined the effects of radar rainfall errors on rainfall-runoff simulation using the spatial error model (SEM). SEM was used to synthetically generate random or cross-correlated errors. A number of events were generated to investigate the effect of spatially dependent errors in radar rainfall estimates on runoff simulation. For runoff simulation, the Nam River basin in South Korea was used with the distributed rainfall-runoff model, Vflo™. The results indicated that spatially dependent errors caused much higher variations in peak discharge than independent random errors. To further investigate the effect of the magnitude of cross-correlation among radar errors, different magnitudes of spatial cross-correlations were employed during the rainfall-runoff simulation. The results demonstrated that a stronger correlation led to a higher variation in peak discharge up to the observed correlation structure while a correlation stronger than the observed case resulted in lower variability in peak discharge. We concluded that the error structure in radar rainfall estimates significantly affects predictions of the runoff peak. Therefore, efforts to not only remove the radar rainfall errors, but to also weaken the cross-correlation structure of the errors need to be taken to forecast flood events accurately.

Merging radar and in situ rainfall measurements: An assessment of different combination algorithms

Merging radar and gauge rainfall estimates is an area of active research. Since rain gauges alone are often limited at representing the complete spatial distribution of rainfall, a combination of radar derived rainfall with spatially interpolated gauge estimates using alternate weighting approaches is investigated. This paper examines several merging methods that differ in the consideration of correlation among the estimation errors, their distribution and the application of dynamic and static weighting. The merging process has been applied to the radar data from Terrey Hills radar located in Sydney, Australia and spatially interpolated gauge rainfall on the same area. The performance of the merging methods is assessed by comparing the combined estimate with the gauge observation. It is however clear from our findings that rainfall estimation from any of the combination approaches assessed contains less error than any of the non-combination approaches. The results show that the correlation between these two rainfall estimation errors plays a significant role in the performance of the merging methods. The combination method should be chosen depending on the purpose, accuracy of the estimate and complexity of the method. This article is protected by copyright. All rights reserved.

Evaluating the potential of rainfall product from Indian geostationary satellite for operational agromet advisory services in India

Satellite remote sensing technology is increasingly gaining recognition as an important source of operational agro meteorological services. Spatial daily rainfall product from geostationary satellite is one of the important data source for quick evaluation of suitability of sowing conditions and other economically relevant agricultural operations (irrigation, fertilizer applications, spraying etc.) by farmers as well as disaster (drought, flood) warning causing crop loss and also to derive weather derivatives for crop insurance. In view of that a study was taken up to explore the use of satellite based rainfall data at different temporal scales. Under the present study, comparison has been made between Kalpana-1 high resolution (0.25° x 0.25°) rainfall estimates for four south-west monsoon months (June-September) in two contrasting monsoon periods of 2008 (normal) and 2009 (drought) with in situ measurements andforecast (1° x 1°) at nine AMFUs (Agro-Meteorological Field Units) in different agro climatic zones.Initial analysis show less correlation and large root mean square error (RMSE) of Kalpana-1 daily rainfall estimates with measurements and forecast. However, correlation was found to increase significantly from 60 to 80% over weekly to fortnightly scale with concomitant decrease in RMSE.

Space–time organization of debris flows-triggering rainfall and its effect on the identification of the rainfall threshold relationship

Summary Debris flow occurrence is generally forecasted by means of empirical rainfall depth–duration thresholds based on raingauge observations. Rainfall estimation errors related to the sparse nature of raingauge data are enhanced in case of convective rainfall events characterized by limited spatial extent. Such errors have been shown to cause underestimation of the rainfall thresholds and, thus, less efficient forecasts of debris flows occurrence. This work examines the spatial organization of debris flows-triggering rainfall around the debris flow initiation points using high-resolution, carefully corrected radar data for a set of short duration ( On average, triggering rainfall presents a local peak corresponding to the debris flow initiation point, with rain depth at 5 km (10 km) distance being on average around 70% (40%) of rain depth observed at the debris flow initiation points. The peak is consistently enhanced for events characterized by short durations and causes a systematic underestimation of the rainfall depth–duration thresholds when rainfall is measured away from the debris flow initiation points. We develop an analytical framework that exploits the general characteristics of the spatial rainfall organization to predict the systematic underestimation of the depth–duration thresholds when rainfall is sampled away from the initiation points. Predictions obtained based on this analytical framework are assessed using a Monte Carlo sampling technique.

Evaluating the potential of radar-based rainfall estimates for streamflow and flood simulations in the Philippines

This case study evaluates the suitability of radar-based quantitative precipitation estimates (QPEs) for the simulation of streamflow in the Marikina River Basin (MRB), the Philippines. Hourly radar-based QPEs were produced from reflectivity that had been observed by an S-band radar located about 90 km from the MRB. Radar data processing and precipitation estimation were carried out using the open source library wradlib. To assess the added value of the radar-based QPE, we used spatially interpolated rain gauge observations (gauge-only (GO) product) as a benchmark. Rain gauge observations were also used to quantify rainfall estimation errors at the point scale. At the point scale, the radar-based QPE outperformed the GO product in 2012, while for 2013, the performance was similar. For both periods, estimation errors substantially increased from daily to the hourly accumulation intervals. Despite this fact, both rainfall estimation methods allowed for a good representation of observed streamflow when used to force a hydrological simulation model of the MRB. Furthermore, the results of the hydrological simulation were consistent with rainfall verification at the point scale: the radar-based QPE performed better than the GO product in 2012, and equivalently in 2013. Altogether, we could demonstrate that, in terms of streamflow simulation, the radar-based QPE can perform as good as or even better than the GO product – even for a basin such as the MRB which has a comparatively dense rain gauge network. This suggests good prospects for using radar-based QPE to simulate and forecast streamflow in other parts of the Philippines where rain gauge networks are not as dense.

Estimation of debris flow triggering rainfall: Influence of rain gauge density and interpolation methods

The forecast of debris flow occurrence relies mainly on empirical rainfall intensity–duration thresholds, which are based on rain gauge observations. This work focuses on the effect of rainfall estimation uncertainty on the estimation of debris flow triggering rainfall events and on the identification of rainfall thresholds for debris flow occurrence. Specifically, the influence of rain gauge network density and the interpolation method on the estimation of debris flow triggering rainfall is investigated. These questions are examined using high-resolution, carefully corrected radar data to represent space–time patterns of true precipitation at the debris flow initiation points and in the surrounding area. Radar rainfall fields are sampled by simulated rain gauge networks, stochastically generated with varying rain gauge densities. Based on these networks, rainfall is estimated by using three rainfall interpolation methods: nearest neighbor (NN), inverse distance weighting (IDW) and ordinary kriging (OK). Results show that NN provides the estimates with bias smaller than IDW and OK but larger estimation variance. Overall, decrease in gauge density leads to increased underestimation and increased estimation variance of debris flow triggering rainfall. Rainfall estimation error leads to large underestimation of the intensity–duration thresholds. However, comparison of results shows that no particular benefit in intensity–duration threshold estimation is obtained by using approaches that are more complex than the NN method.

An algorithm for optimisation of a rain gauge network based on geostatistics and entropy concepts using GIS

There is a high level of spatio-temporal variability in rainfall events over a region. Rain gauges provide point measurements that are interpolated to estimate the rainfall in other points of a region or in order to estimate the regional depth of rainfall. Due to limited numbers of rain gauges that can be deployed, it is important to plan for the optimal number and locations of rain gauges in a region. Because of several factors that can affect the optimal number of stations as well as their locations, development of optimisation models for determining the number of rain gauges and their locations within a region is necessary. In this study, an optimisation model for determining the number and location of the rain gauges is proposed. The objectives of the model are maximising the minimum transformation entropy and minimising the rainfall estimation error. To determine the optimal number of new rain gauges, the candidate points, which are those with the maximum variance of Kriging error and the minimum information transformation entropy, are prioritised according to their participation in the transformation. A mountainous sub-basin has been considered as the case study. Applying the proposed approach to the case study, 17 candidate points are determined for deployment of new rain gauges. Finally, 10 of the candidate points are selected to be added to the current rain gauge network. The results of the study show that joint consideration of spatial analysis error and information transformation entropy can highly improve the optimal design of a rain gauge network.

Using the new dual-polarimetric capability of WSR-88D to eliminate anomalous propagation and wind turbine effects in radar-rainfall

This study addresses the effect that the interaction between anomalous radar beam propagation (AP) and wind turbines that are located far from the radar has on radar-rainfall estimates. The interference of wind turbines in radar observations may lead to significant errors in rainfall estimates since wind turbines are often clustered to form wind farms. In this study, we propose a novel approach – based on the polarimetric capability recently added to the WSR-88D NEXRAD radars – that identifies and eliminates wind turbine clutter along with common ground clutter AP effects. Our primary objective is to devise a physically meaningful and fully automated dual-polarimetric method that effectively handles clutter features, which are hard to detect using single-channel reflectivity data alone. To address this issue, we explore the feasibility of using polarimetric variables such as differential reflectivity (ZDR), copolar correlation (RHO), and differential phase (PHIDP). Accordingly, we developed three new approaches using polarimetric variables, which are combined with the AP detection algorithm that uses a three-dimensional structure of reflectivity. We evaluate the new algorithms in terms of both eliminating non-meteorological radar returns and preserving returns from actual rain. The proposed algorithm, which uses RHO conditioned on horizontal reflectivity values while also accounting for the variation of ZDR or PHIDP, shows good performance for the presented cases.

Influence of rainfall estimation error and spatial variability on sewer flow prediction at a small urban scale

Legislative drivers for water quality and urban flood risk are driving a growing need to accurately determine the performance of urban drainage systems in near real time. Rainfall data are clearly a key input to urban drainage system models. Historically rain gauge data have been used, however radar rainfall data are now widely available and benefits from significantly higher spatial coverage than rain gauges in most UK urban catchments. This paper describes a detailed study based on a small (11km2) urban catchment in West Yorkshire, England. Radar and rain gauge data have been compared and used as the input to hydrodynamic sewer flow simulations, and the results of these simulations have been compared with measured flows in the sewer system. The results showed that for this size of catchment, there can be significant differences in simulated peak flows and combined sewer overflow spill volumes due to inherent uncertainties between the two rainfall estimates.

Quantitative precipitation estimates from radar reflectivity corrected by the SMA method

In order to improve the accuracy of quantitative precipitation estimates, we correct radar reflectivity measurements by a “sorting and moving average” (SMA) method and with a POSS (precipitation occurrence sensing system) disdrometer. The correction procedure, optimized by a polynomial least-square fit of the data, greatly reduces errors in rainfall estimates. The Zc–R relationships for different cloud types are calculated using a classification algorithm. A VPR threshold algorithm is used to investigate the accuracy of rainfall estimates depending on the cloud type. For stratiform and convective cloud types, rainfall estimates were more accurate when the correction was taken into account, with statistical errors substantially reduced.The developed algorithm successfully reduced errors in the rainfall estimates and improved their accuracy. This new quantitative precipitation estimate (QPE) algorithm will improve the reliability of radar-based quantitative rainfall measurements and the accuracy of weather forecasts.

Uncertainties on mean areal precipitation: assessment and impact on streamflow simulations

Abstract. This paper investigates the influence of mean areal rainfall estimation errors on a specific case study: the use of lumped conceptual rainfall-runoff models to simulate the flood hydrographs of three small to medium-sized catchments of the upper Loire river. This area (3200 km2) is densely covered by an operational network of stream and rain gauges. It is frequently exposed to flash floods and the improvement of flood forecasting models is then a crucial concern. Particular attention has been drawn to the development of an error model for rainfall estimation consistent with data in order to produce realistic streamflow simulation uncertainty ranges. The proposed error model combines geostatistical tools based on kriging and an autoregressive model to account for temporal dependence of errors. It has been calibrated and partly validated for hourly mean areal precipitation rates. Simulated error scenarios were propagated into two calibrated rainfall-runoff models using Monte Carlo simulations. Three catchments with areas ranging from 60 to 3200 km2 were tested to reveal any possible links between the sensitivity of the model outputs to rainfall estimation errors and the size of the catchment. The results show that a large part of the rainfall-runoff (RR) modelling errors can be explained by the uncertainties on rainfall estimates, especially in the case of smaller catchments. These errors are a major factor limiting accuracy and sharpness of rainfall-runoff simulations, and thus their operational use for flood forecasting.