Weather or not—Global climate databases: Reliable on tropical mountains?

surface datasets are not necessarily the ’truth’ as many factors influence the true value (exploi- tation of data, measurement equipment), inconsistencies or errors can be reduced through careful selection of data and quality control [7]. For our data, we can therefore exclude this explanation. Based on such data, more meaningful models can be built, as has been done for Kilimanjaro for precipitation [17] (Hemp, 2006) and for precipitation, temperature and humidity [13] with a much higher resolution. The fact that the measured periods sometimes deviate considerably from the modelled data is hardly significant: firstly, the longest records and the highest deviations in the measurement periods occurred in the lowlands, where the models performed best, while at higher elevations, with the much higher agreement in the recording periods, the deviation in the amount of pre- cipitation values was greatest. Furthermore, precipitation changes have not been very pro- nounced in recent decades [18]. The observed discrepancies are thus more related to the downscaling approach used, to the lack of available data for the model and to the grid cell size of 1 km, which is much too large in complex terrain. In the case of C, where data from a downscaled satellite-gauge reanalysis (ERA5) is merged with further remote sensing data, possible sources of error are even more difficult to detect. One aspect that has not been addressed in this study is the impact of water input from fog. On Kilimanjaro, this input is of less importance due to the weak and moderate wind speeds

[20] (Hemp, 2005b). In the Pare Mountains, however, this input is important. On Mwala, a peak of the South Pare, for example, water input from fog can be more than twice the rainfall, increasing the MAP from 740 mm to 2303 mm, creating cloud forest habitats with unique Fig 5. Comparison of measured and modelled MAP maxima in relation to biodiversity patterns on the southern slope of Kilimanjaro. MAP data of WC and C cannot explain biodiversity patterns (e.g. of pteridophytes, green curve, from [25] and vegetation zonation (right images; from bottom to top: Colline savannah zone, submontane agroforestry zone, montane forest zone, subalpine and alpine zone) due to their incorrect rainfall maxima (blue zones). https://doi.org/10.1371/journal.pone.0299363.g005 PLOS ONE Global climate datasets on tropical mountains PLOS ONE | https://doi.org/10.1371/journal.pone.0299363 March 13, 2024 9 / 12