Interpretation of mean‐field bias correction of radar rain rate using the concept of linear regression

In this study, the correction problem of mean‐field bias of radar rain rate was investigated using the concept of linear regression. Three different relationships were reviewed for their slopes to be used as the bias correction factor: Relationship 1 (R1) is based on the conventional linear regression, relationship 2 (R2) is forced to pass the origin and relationship 3 (R3) is the line whose slope is the G/R ratio. In other words, R1 is the regression line connecting the intercept and the mass centre of measurement pairs, R2 is the regression line forced to pass the origin, and R3 is the line connecting the origin and the mass centre. The slopes of all three relationships were reviewed analytically to compare them, and thereby, the effect of zero measurements could be evaluated. Additionally, the effect of using switched independent and dependent variables on the derived slopes was also evaluated. The theoretically derived results were then verified by analysing the rainfall event on 10–11 August 2010 in Korea. Finally, the difference between the bias‐corrected radar rain rate and the rain gauge rain rate was quantified by root mean square error and mean error so that it could be used as a measure for the evaluation of bias correction factors. In conclusion, the slope of R2 was found to be the best for the bias correction factor. However, when deciding the slope of this R2, the radar rain rate should be used as the independent variable in the low rain rate region, and the rain gauge rain rate in the high rain rate region above a certain threshold. Copyright © 2013 John Wiley & Sons, Ltd.     

Detection of mean-field bias of the radar rain rate using rain gauges available within a small portion of radar umbrella: a case study of the Donghae (East Sea) radar in Korea

Observation of a storm approaching from the ocean to the in-land area is very important for the flood forecasting. Radar is generally used for this purpose. However, as rain gauges are mostly located within the in-land area, detection of the mean-field bias of radar rain rate cannot be easily made. This problem is obviously different from that with evenly-spaced rain gauges over the radar umbrella. This study investigated the detection problem of mean-field bias of radar rain rate when rain gauges are available within a small portion of radar umbrella. To exactly determine the mean-field bias, i.e. the difference between the radar rain rate and the rain gauge rain rate, the variance of the difference between two observations must be small; thus, a sufficient number of observations are indispensable. Therefore, the problem becomes determining the number of rain gauges that will satisfy the given accuracy, that being the variance of the difference between two observations. The dimensionless error variance derived by dividing the expected value of the error variance by the variance of the areal average rain rate was introduced as a criteria to effectively detect the mean field bias. Here, the variance of the areal average rain rate was assumed to be the climatological one and the expectation for the error variance could be changed depending one the sampling characteristics. As an example, this study evaluated the rainfall observation over the East Sea by the Donghae radar. About 6.8 % of the entire radar umbrella covered in-land areas, where the rain gauges were available. It was found that, to limit the dimensionless error variance to 2 %, a total of 26 rain gauges are required for the entire radar umbrella; whereas, a total of 24 rain gauges would be required within the in-land area with available for the rain gauge data.     

Adjustment of wind-drift effect for real-time systematic error correction in radar rainfall data

An effective bias correction procedure using gauge measurement is a significant step for radar data processing to reduce the systematic error in hydrological applications. In these bias correction methods, the spatial matching of precipitation patterns between radar and gauge networks is an important premise. However, the wind-drift effect on radar measurement induces an inconsistent spatial relationship between radar and gauge measurements as the raindrops observed by radar do not fall vertically to the ground. Consequently, a rain gauge does not correspond to the radar pixel based on the projected location of the radar beam. In this study, we introduce an adjustment method to incorporate the wind-drift effect into a bias correlation scheme. We first simulate the trajectory of raindrops in the air using downscaled three-dimensional wind data from the weather research and forecasting model (WRF) and calculate the final location of raindrops on the ground. The displacement of rainfall is then estimated and a radar–gauge spatial relationship is reconstructed. Based on this, the local real-time biases of the bin-average radar data were estimated for 12 selected events. Then, the reference mean local gauge rainfall, mean local bias, and adjusted radar rainfall calculated with and without consideration of the wind-drift effect are compared for different events and locations. There are considerable differences for three estimators, indicating that wind drift has a considerable impact on the real-time radar bias correction. Based on these facts, we suggest bias correction schemes based on the spatial correlation between radar and gauge measurements should consider the adjustment of the wind-drift effect and the proposed adjustment method is a promising solution to achieve this.     

A comparison of gauge and radar precipitation data for simulating an extreme hydrological event in the Severn Uplands,UK

This paper provides a comparison of gauge and radar precipitation data sources during an extreme hydrological event. November–December 2006 was selected as a time period of intense rainfall and large river flows for the Severn Uplands, an upland catchment in the United Kingdom. A comparison between gauge and radar precipitation time‐series records for the event indicated discrepancies between data sources, particularly in areas of higher elevation. The HEC‐HMS rainfall‐runoff model was selected to assess the accuracy of the precipitation to simulate river flows for the extreme event. Gauge, radar and gauge‐corrected radar rainfall were used as model inputs. Universal cokriging was used to geostatistically interpolate gauge data with radar and elevation data as covariates. This interpolated layer was used to calculate the mean‐field bias and correct the radar composites. Results indicated that gauge‐ and gauge‐corrected radar‐driven models replicated flows adequately for the extreme event. Gauge‐corrected flow predictions produced an increase in flow prediction accuracy when compared with the raw radar, yet predictions were comparative in accuracy to those using the interpolated gauge network. Subsequent investigations suggested this was due to an adequate spatial and temporal resolution of the precipitation gauge network within the Severn Uplands. Results suggested that the six rain gauges could adequately represent precipitation variability of the Severn Uplands to predict flows at an approximately equal accuracy to that obtained by radar. Temporally, radar produced an increase in flow prediction accuracy in mountainous reaches once the gauge time step was in excessive of an hourly interval. Copyright © 2010 John Wiley & Sons, Ltd.     

Flood simulation using the gauge‐adjusted radar rainfall and physics‐based distributed hydrologic model

This paper reports the results of an investigation into flood simulation by areal rainfall estimated from the combination of gauged and radar rainfalls and a rainfall–runoff model on the Anseong‐cheon basin in the southern part of Korea. The spatial and temporal characteristics and behaviour of rainfall are analysed using various approaches combining radar and rain gauges: (1) using kriging of the rain gauge alone; (2) using radar data alone; (3) using mean field bias (MFB) of both radar and rain gauges; and (4) using conditional merging technique (CM) of both radar and rain gauges. To evaluate these methods, statistics and hyetograph for rain gauges and radar rainfalls were compared using hourly radar rainfall data from the Imjin‐river, Gangwha, rainfall radar site, Korea. Then, in order to evaluate the performance of flood estimates using different rainfall estimation methods, rainfall–runoff simulation was conducted using the physics‐based distributed hydrologic model, Vflo^?. The flood runoff hydrograph was used to compare the calculated hydrographs with the observed one. Results show that the rainfall field estimated by CM methods improved flood estimates, because it optimally combines rainfall fields representing actual spatial and temporal characteristics of rainfall. Copyright © 2008 John Wiley & Sons, Ltd.     

Spatial reconstruction of rainfall fields from rain gauge and radar data

Rainfall is a phenomenon difficult to model and predict, for the strong spatial and temporal heterogeneity and the presence of many zero values. We deal with hourly rainfall data provided by rain gauges, sparsely distributed on the ground, and radar data available on a fine grid of pixels. Radar data overcome the problem of sparseness of the rain gauge network, but are not reliable for the assessment of rain amounts. In this work we investigate how to calibrate radar measurements via rain gauge data and make spatial predictions for hourly rainfall, by means of Monte Carlo Markov Chain algorithms in a Bayesian hierarchical framework. We use zero-inflated distributions for taking zero-measurements into account. Several models are compared both in terms of data fitting and predictive performances on a set of validation sites. Finally, rainfall fields are reconstructed and standard error estimates at each prediction site are shown via easy-to-read spatial maps.     

Evaluation of the climate prediction center (CPC) morphing technique (CMORPH) rainfall product on hourly time scales over the source of the Blue Nile River

Limited availability of surface‐based rainfall observations constrains the evaluation of satellite rainfall products over many regions. Observations are also often not available at time scales to allow evaluation of satellite products at their finest resolutions. In the present study, we utilized a 3‐month rainfall data set from an experimental network of eight automatic gauges in Gilgel Abbay watershed in Ethiopia to evaluate the 1‐hourly, 8 × 8‐km Climate Prediction Center morphing technique (CMORPH) rainfall product. The watershed is situated in the Lake Tana basin which is the source of the Blue Nile River. We applied a suite of statistical metrics that included mean difference, bias, standard deviation of differences and measures of association. Our results indicate that the accuracy of the CMORPH product shows a significant variation across the basin area. Its estimates are mostly within ±10 mm h^?1 of the gauge rainfall observations; however, the product does not satisfactorily capture the rainfall temporal variability and is poorly correlated (<0.27) to gauge observations. Its poor rain detection capability led to significant underestimation of the seasonal rainfall depth (total bias reaches up to ?52%) with large amounts of hit rain bias as well as missed rain and false rain biases. In the future refinement of CMORPH algorithm, more attention should be given to reducing missed rain bias over the mountains of Gilgel Abbay, whereas equal attention should be given to hit, missed rain and false rain biases over other parts of the watershed. Copyright © 2012 John Wiley & Sons, Ltd.     

Comparison of radar and gauge precipitation data in watershed models across varying spatial and temporal scales

Precipitation is a key control on watershed hydrologic modelling output, with errors in rainfall propagating through subsequent stages of water quantity and quality analysis. Most watershed models incorporate precipitation data from rain gauges; higher‐resolution data sources are available, but they are associated with greater computational requirements and expertise. Here, we investigate whether the Multisensor Precipitation Estimator (MPE or Stage IV Next‐Generation Radar) data improve the accuracy of streamflow simulations using the Soil and Water Assessment Tool (SWAT), compared with rain gauge data. Simulated flows from 2002 to 2010 at five timesteps were compared with observed flows for four nested subwatersheds of the Neuse River basin in North Carolina (21‐, 203‐, 2979‐, and 10 100‐km² watershed area), using a multi‐objective function, informal likelihood‐weighted calibration approach. Across watersheds and timesteps, total gauge precipitation was greater than radar precipitation, but radar data showed a conditional bias of higher rainfall estimates during large events (>25–50 mm/day). Model parameterization differed between calibrations with the two datasets, despite the fact that all watershed characteristics were the same across simulation scenarios. This underscores the importance of linking calibration parameters to realistic processes. SWAT simulations with both datasets underestimated median and low flows, whereas radar‐based simulations were more accurate than gauge‐based simulations for high flows. At coarser timesteps, differences were less pronounced. Our results suggest that modelling efforts in watersheds with poor rain gauge coverage can be improved with MPE radar data, especially at short timesteps. Published 2013. This article is a U.S. Government work and is in the public domain in the USA.     

Investigation of the scale-dependent variability of radar-rainfall and rain gauge error covariance

There is a significant spatial sampling mismatch between radar and rain gauge data. The use of rain gauge data to estimate radar-rainfall error variance requires partitioning of the variance of the radar and rain gauge difference to account for the sampling mismatch. A key assumption in the literature pertaining to the error variance separation method used to partition the variance is that the covariance between radar-rainfall error and the error of rain gauges in representing radar sampling domain is negligible. Our study presents the results of an extensive test of this assumption. The test is based on empirical data and covers temporal scales ranging from 0.25 to 24 h and spatial scales ranging from 1 to 32 km. We used a two-year data set from two high quality and high density rain gauge networks in Oklahoma and excluded the winter months. The results obtained using a resampling procedure show that covariance can be considerable at large scales due to the significant variability. As the variability of the covariance rapidly increases with larger spatial and shorter temporal scales, applications of the error variance separation method at those scales require more caution. The variability of the covariance and one of its constituting variables, the variance ratio of radar and gauge errors, shows simple scaling behavior well characterized by a power-law.     

Radar rainfall uncertainty modelling influenced by wind

Radar‐based estimates of rainfall are affected by many sources of uncertainties, which would propagate through the hydrological model when radar rainfall estimates are used as input or initial conditions. An elegant solution to quantify these uncertainties is to model the empirical relationship between radar measurements and rain gauge observations (as the ‘ground reference’). However, most current studies only use a fixed and uniform model to represent the uncertainty of radar rainfall, without consideration of its variation under different synoptic regimes. Wind is such a typical weather factor, as it not only induces error in rain gauge measurements but also causes the raindrops observed by weather radar to drift when they reach the ground. For this reason, as a first attempt, this study introduces the wind field into the uncertainty model and designs the radar rainfall uncertainty model under different wind conditions. We separate the original dataset into three subsamples according to wind speed, which are named as WDI (0–2 m/s), WDII (2–4 m/s) and WDIII (>4 m/s). The multivariate distributed ensemble generator is introduced and established for each subsample. Thirty typical events (10 at each wind range) are selected to explore the behaviours of uncertainty under different wind ranges. In each time step, 500 ensemble members are generated, and the values of 5th to 95th percentile values are used to produce the uncertainty bands. Two basic features of uncertainty bands, namely dispersion and ensemble bias, increase significantly with the growth of wind speed, demonstrating that wind speed plays a considerable role in influencing the behaviour of the uncertainty band. On the basis of these pieces of evidence, we conclude that the radar rainfall uncertainty model established under different wind conditions should be more realistic in representing the radar rainfall uncertainty. This study is only a start in incorporating synoptic regimes into rainfall uncertainty analysis, and a great deal of more effort is still needed to build a realistic and comprehensive uncertainty model for radar rainfall data. Copyright © 2014 John Wiley & Sons, Ltd.     

Estimation of rainfall from infrared‐microwave satellite data for basin‐scale hydrologic modelling

The infrared‐microwave rainfall algorithm (IMRA) was developed for retrieving spatial rainfall from infrared (IR) brightness temperatures (TBs) of satellite sensors to provide supplementary information to the rainfall field, and to decrease the traditional dependency on limited rain gauge data that are point measurements. In IMRA, a SLOPE technique (ST) was developed for discriminating rain/no‐rain pixels through IR image cloud‐top temperature gradient, and 243K as the IR threshold temperature for minimum detectable rainfall rate. IMRA also allows for the adjustment of rainfall derived from IR‐TB using microwave (MW) TBs. In this study, IMRA rainfall estimates were assessed on hourly and daily basis for different spatial scales (4, 12, 20, and 100 km) using NCEP stage IV gauge‐adjusted radar rainfall data, and daily rain gauge data. IMRA was assessed in terms of the accuracy of the rainfall estimates and the basin streamflow simulated by the hydrologic model, Sacramento soil moisture accounting (SAC‐SMA), driven by the rainfall data. The results show that the ST option of IMRA gave accurate satellite rainfall estimates for both light and heavy rainfall systems while the Hessian technique only gave accurate estimates for the convective systems. At daily time step, there was no improvement in IR‐satellite rainfall estimates adjusted with MW TBs. The basin‐scale streamflow simulated by SAC‐SMA driven by satellite rainfall data was marginally better than when SAC‐SMA was driven by rain gauge data, and was similar to the case using radar data, reflecting the potential applications of satellite rainfall in basin‐scale hydrologic modelling. Copyright © 2010 John Wiley & Sons, Ltd.     

Using microwave links to adjust the radar rainfall field

The final stage in processing radar data so as to arrive at an estimated rain field typically involves a comparison of the preliminary radar-derived estimates of hourly rainfall with those observed by ground-based gauges. Often a mean field bias adjustment will then be applied using an age-weighted average of the individual gauge–radar comparisons. In this paper, a mean field bias adjustment is presented that uses the path-integrated rainfall estimates provided by microwave links together with information from gauges. It is shown to be at least as efficient as the current gauge-based procedure used by the UK Met Office to improve the accuracy of radar-based estimates of rainfall at the ground.     

A multi‐source precipitation estimation approach to fill gaps over a radar precipitation field: a case study in the Colorado River Basin

The present work develops an approach to seamlessly blend satellite, available radar, climatological and gauge precipitation products to fill gaps in ground‐based radar precipitation field. To mix different precipitation products, the error of any of the products relative to each other should be removed. For bias correction, the study uses an ensemble‐based method that aims to estimate spatially varying multiplicative biases in SPEs using a radar precipitation product. A weighted successive correction method (SCM) is used to make the merging between error corrected satellite and radar precipitation estimates. In addition to SCM, we use a combination of SCM and Bayesian spatial model for merging the rain gauges (RGs) and climatological precipitation sources with radar and SPEs. We demonstrated the method using a satellite‐based hydro‐estimator; a radar‐based, stage‐II; a climatological product, Parameter‐elevation Regressions on Independent Slopes Model and a RG dataset for several rain events from 2006 to 2008 over an artificial gap in Oklahoma and a real radar gap in the Colorado River basin. Results show that: the SCM method in combination with the Bayesian spatial model produced a precipitation product in good agreement with independent measurements. The study implies that using the available radar pixels surrounding the gap area, RG, Parameter‐elevation Regressions on Independent Slopes Model and satellite products, a radar‐like product is achievable over radar gap areas that benefit the operational meteorology and hydrology community. Copyright © 2013 John Wiley & Sons, Ltd.     

Use of mixed bivariate distributions for deriving inter‐station correlation coefficients of rain rate

Even though rain rate is notorious for its spatial and temporal intermittency, its effect on the second‐order statistics of rain rate, especially the inter‐station correlation coefficients, has not been intensively evaluated before. This study has derived and compared the inter‐station correlation coefficient of rain rate for three cases of data: (1) only the positive measurements at both locations; (2) the positive measurements at either one or both locations; (3) all the measurements including zero measurement at both locations. For these three cases, the inter‐station correlation coefficients are analytically derived by applying the mixed bivariate log‐normal distribution. As an application example, the model parameters are estimated using the rain rate data collected at the Geum River basin, Korea, and the resulting inter‐station correlation coefficients are evaluated and compared with those estimated by applying the Gaussian distribution. We could find that highly biased inter‐station correlation coefficients are unavoidable when simply estimating them under the assumption of Gaussian distribution, or even when using the log‐transformed rain rate data. Copyright © 2007 John Wiley & Sons, Ltd.     

The Use of Radar and Gauge Measurements to Estimate Areal Precipitation for Several Czech River Basins

Two methods estimating areal precipitation for selected river basins in the Czech Republic are compared. The methods use radar precipitation (the radar-derived precipitation estimate based on column maximum reflectivity) and data from 81 on-line rain gauges routinely provided by the Czech Hydrometeorological Institute. Data from a dense network of climatological rain gauges (the average inter-station distance is approximately 8 km), the measurements of which are not available in real time, are utilized for the verification. The mean areal precipitation, which is used as the ground truth, is obtained by the weighted interpolation of the dense rain gauge network. The accuracy of the methods is evaluated by the root-mean-square-error.The first, pixel-related method merges radar precipitation with rain gauge data to obtain adjusted pixel values. The adjusting procedure combines radar and gauge values in one variable that is interpolated into all radar pixels. The adjusted pixel precipitation is calculated from radar precipitation and from the value of the combined variable. The areal estimates are determined by adding the corresponding pixel values. The second method applies a linear regression model to describe the relationship between the areal precipitation (dependent variable) and its estimates, which are determined from (i) non-adjusted radar precipitation and (ii) on-line rain gauge measurements interpolated into pixels. Classical linear regression, ridge regression and robust regression models are tested.Both the methods decrease the average areal error in comparison with the reference method, which uses the on-line rain gauge data only. The decrease is about 10% and 15% for the pixel-related and regression methods, respectively. When the estimates of the pixel-related method are included as predictors into the regression method then the improvement of accuracy is almost 25%.     

Empirically-based modeling of spatial sampling uncertainties associated with rainfall measurements by rain gauges

In the quantitative evaluation of radar-rainfall products (maps), rain gauge data are generally used as a good approximation of the true ground rainfall. However, rain gauges provide accurate measurements for a specific location, while radar estimates represent areal averages. Because these sampling discrepancies could introduce noise into the comparisons between these two sensors, they need to be accounted for. In this study, the spatial sampling error is defined as the ratio between the measurements by a single rain gauge and the true areal rainfall, defined as the value obtained by averaging the measurements by an adequate number of gauges within a pixel. Using a non-parametric scheme, the authors characterize its full statistical distribution for several spatial (4, 16 and 36 km²) and temporal (15 min and hourly) scales.     

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

Rainfall 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.     

A non-parametric automatic blending methodology to estimate rainfall fields from rain gauge and radar data

Quantitative estimation of rainfall fields has been a crucial objective from early studies of the hydrological applications of weather radar. Previous studies have suggested that flow estimations are improved when radar and rain gauge data are combined to estimate input rainfall fields. This paper reports new research carried out in this field. Classical approaches for the selection and fitting of a theoretical correlogram (or semivariogram) model (needed to apply geostatistical estimators) are avoided in this study. Instead, a non-parametric technique based on FFT is used to obtain two-dimensional positive-definite correlograms directly from radar observations, dealing with both the natural anisotropy and the temporal variation of the spatial structure of the rainfall in the estimated fields. Because these correlation maps can be automatically obtained at each time step of a given rainfall event, this technique might easily be used in operational (real-time) applications. This paper describes the development of the non-parametric estimator exploiting the advantages of FFT for the automatic computation of correlograms and provides examples of its application on a case study using six rainfall events. This methodology is applied to three different alternatives to incorporate the radar information (as a secondary variable), and a comparison of performances is provided. In particular, their ability to reproduce in estimated rainfall fields (i) the rain gauge observations (in a cross-validation analysis) and (ii) the spatial patterns of radar fields are analyzed. Results seem to indicate that the methodology of kriging with external drift [KED], in combination with the technique of automatically computing 2-D spatial correlograms, provides merged rainfall fields with good agreement with rain gauges and with the most accurate approach to the spatial tendencies observed in the radar rainfall fields, when compared with other alternatives analyzed.     

Characteristics of rain events at an inland locality in northeastern Borneo,Malaysia

Understanding the intensity and duration of tropical rain events is critical to modelling the rate and timing of wet‐canopy evaporation, the suppression of transpiration, the generation of infiltration‐excess overland flow and hence to erosion, and to river responsiveness. Despite this central role, few studies have addressed the characteristics of equatorial rainstorms. This study analyses rainfall data for a 5 km² region largely comprising of the 4 km² Sapat Kalisun Experimental Catchment in the interior of northeastern Borneo at sampling frequencies from 1 min^?1 to 1 day^?1. The work clearly shows that most rainfall within this inland, forested area is received during regular short‐duration events (<15 min) that have a relatively low intensity (i.e. less than two 0·2 mm rain‐gauge tips in almost all 5 min periods). The rainfall appears localized, with significant losses in intergauge correlations being observable in minutes in the case of the typical mid‐afternoon, convective events. This suggests that a dense rain‐gauge network, sampled at a high temporal frequency, is required for accurate distributed rainfall‐runoff modelling of such small catchments. Observed rain‐event intensity is much less than the measured infiltration capacities, and thus supports the tenet of the dominance of quick subsurface responses in controlling river behaviour in this small equatorial catchment. Copyright © 2006 John Wiley & Sons, Ltd.     

Validation of official erosion modelling based on high‐resolution radar rain data by aerial photo erosion classification

The universal soil loss equation (USLE) is the most frequently applied erosion prediction model and it is also implemented as an official decision‐making instrument for agricultural regulations. The USLE itself has been already validated using different approaches. Additional errors, however, arise from input data and interpolation procedures that become necessary for field‐specific predictions on a national scale for administrative purposes. In this study, predicted event soil loss using the official prediction system in Bavaria (Germany) was validated by comparison with aerial photo erosion classifications of 8100 fields. Values for the USLE factors were mainly taken from the official Bavarian high‐resolution (5 × 5 m²) erosion cadastre. As series of erosion events were examined, the cover and management factor was replaced by the soil loss ratio. The event erosivity factor was calculated from high‐resolution (1 × 1 km², 5 min), rain gauge‐adjusted radar rain data (RADOLAN). Aerial photo erosion interpretation worked sufficiently well and average erosion predictions and visual classifications correlated closely. This was also true for data broken down to individual factors and different crops. There was no reason to assume a general invalidity of the USLE and the official parametrization procedures. Event predictions mainly suffered from errors in the assumed crop stage period and tillage practices, which do not reflect interannual and farm‐specific variation. In addition, the resolution of radar data (1 km²) did not seem to be sufficient to predict short‐term erosion on individual fields given the strong spatial gradients within individual rains. The quality of the input data clearly determined prediction quality. Differences between USLE predictions and observations are most likely caused by parametrization weaknesses but not by a failure of the model itself. Copyright © 2017 John Wiley & Sons, Ltd.