Abstract. The assessment of rainfall erosivity is one of the main inputs in determining soil erosion. To calculate mean annual rainfall erosivity (R), long-term high-resolution observed rainfall time series are required, which are often not available. To overcome the issue of limited data availability in space, four methods are employed: the direct regionalisation of R, the regionalisation of 5 minute rainfall, the disaggregation of daily rainfall into 5 minute timesteps, and the use of a

more »

... alised stochastic rainfall model. In addition the minimum time series length necessary to adequately estimate R is investigated. The impact of station density is considered for each of the methods. The study is carried out using 159 recording and 150 non-recording (daily) rainfall stations in the federal state of Lower Saxony, Germany. Results show that the direct regionalisation of mean annual erosivity leads to the best results in terms of relative bias and relative root mean square error (RMSE). This is followed by the regionalisation of the 5 minute rainfall data, which yields better results than the rainfall generation models, namely an alternating renewal model (ARM) and a multiplicative cascade model (Disagg). However, a key advantage of using regionalised rainfall models is the generation of rainfall time series that can be used for the estimation of the erosive event characteristics, which is not possible through the direct regionalisation of R. Using the stochastic ARM it can be shown that in most cases more than 60 years of data is needed in order to obtain a stable estimate of the annual rainfall erosivity. Estimation of soil erosion based on only 5 or 10 years of data can lead to uncertain values. Such short time series are often used when regionalisation is applied. Moreover, it was also found that resolution of measuring device has a significant effect on the rainfall erosivity and coarser data resolution can lead to high relative bias.