Evaluation and correction of precipitation data obtained with different measurement methods using data from precision lysimeter network

Accurate precipitation measurements are essential for various applications such as determining the water balance of ecosystems and modelling soil-water fluxes in the earth critical zone. Gauge based point precipitation measurements are affected by wind, gauge design, and maintenance of the device. Ground-level gauges, like high precision weighing lysimeters, are less affected by environmental factors and thus provide more accurate data if well managed and the data are post-processed with filters. However, studies evaluating precipitation measuring methods with lysimeter references at multiple sites with high temporal resolution and detailed weather data are rare.

In the present study, high-precision weighing lysimeter precipitation data from four years of measurement with an hourly resolution were used as references to evaluate data from four different precipitation measurement methods at three sites under different climatic conditions. The methods were tipping bucket gauges (TB), weighing gauges (WG), acoustic sensors (AS), and laser disdrometers (LD). Different sites and climatic conditions were chosen to be able to draw conclusions as to whether deviations between the measurement and comparison data were environment-dependent or unit-specific. Methodically, the evaluation included correlation analyses, comparison of catch ratios, x-y scatter plots, and the application of correction schemes.

For the total period, all measurement methods recorded less precipitation than the lysimeters, with catch ratios between 33 to 92 % depending on the measuring method. Non-rainfall water inputs, like dew and fog, have been excluded for this study, therefore the measuring differences are attributed to the precipitation gauges. The bias of the hourly measurements varied between -0.69 to -0.01 mm h^-1 based on the measuring method and no site-specific influence on the data was detected. Correction algorithms reduced the bias and improved the catching ratios of hourly precipitation data with similar improvements at all sites for the same gauge models, thus one adequate correction scheme may be sufficient to be used for the same model under different climatic conditions and environments. The findings suggest that a correction of the data by empirical or mathematical models appears to be necessary to ensure the quality of the precipitation data and to reduce over- and underestimations, which is the prerequisite for environmental studies in the critical zone.