Comparison of rainfall generators with regionalisation for the estimation of rainfall erosivity at ungauged sites

Abstract. Rainfall erosivity values are required for soil erosion prediction. To calculate the mean annual rainfall (R), long-term high-resolution observed data required, which often not available. overcome issue of limited availability in space and time, four methods were employed evaluated: direct regionalisation R, 5 min rainfall, disaggregation daily into time steps, a regionalised stochastic model. The impact station density is considered each methods. study carried out using 159 recording 150 non-recording (daily) stations around federal state Lower Saxony, Germany. In addition, minimum record length necessary to adequately estimate R was investigated. Results show that best terms both relative bias root square error (RMSE), followed by data, yields better results than generation models, namely an alternating renewal model (ARM) multiplicative cascade However, key advantage models ability generate series can be used estimation erosive event characteristics. This possible if regionalising only R. Using ARM, it assessed more 60 years needed most cases reach stable erosivity. Moreover, temporal resolution measuring devices found have significant effect on with coarser leading higher bias.