| 13 Apr 2026
European daily dataset of soil moisture from in situ meteorological observations
Abstract. This paper documents the development and validation of an in situ dataset for daily soil moisture based on observed meteorological variables from the European Climate Assessment & Dataset (ECA&D). These data are input to the SoilClim model which provides soil water contents values for the upper 10 cm, the root zone and the deeper layer of the soil. Soil characteristics are taken from the widely used SoilGrids system. The meteorological input data are daily precipitation totals, daily averaged temperature, and global radiation sums. The pan-European dataset for soil water content results in a set of 5758 stations for which daily soil water content values are produced, for the surface zone (upper 10 cm) and for the root zone (upper 40 cm), for the period for which all input data is available. Model results are validated against observations of actual evapotranspiration (ETa) from the FluxNet experiment. Calculated soil water content are compared against a selection of the observations of the International Soil Moisture Network (ISMN). In the comparisons against observations, calculated values of ETa and soil water content (SWC) from the state-of-the-art soil dynamics model SWAP (Soil-Water-Atmosphere-Plants), which is more advanced than SoilClim, are used as well. The comparisons show that SoilClim and SWAP show generally the same behaviour albeit that SoilClim saturates quicker than SWAP. On average, the similarity between the models and observation for ETa is higher (correlations 0.77 and 0.78 for SWAP and SoilClim respectively) than for the observed soil water content, where average correlations for the surface and rootzone vary between 0.52 and 0.62.
Based on the rootzone soil water content, four climate indices are calculated. These are the Soil Moisture Index (SMI) which is a metric to compare soil water content values across the diverse climatic zones which are present in Europe. This index relates to water content levels that are above or below the 50 % level of the field capacity. In addition, an index is introduced that simply counts the number of days where the soil water content drops below the 50 % level of the field capacity. In addition, two indices assess more extreme soil water content values: the number days where the soil water content is either nearly at field capacity or close to the wilting point. The 1991–2020 spring and summer climatology is provided for this dataset in terms of these indices. The relatively recent 2018 drought is put in a historical perspective by a comparison against the 1976 drought and a trend analysis of summer desiccation since 1979 is provided.
The use of in situ observations, the daily resolution and the climate indices included in the described dataset make it suitable as validation of other soil moisture datasets and the assessment of changes in the occurrence of extremes in soil moisture over the European domain. The dataset of daily soil moisture from in situ meteorological observations is available at https://doi.org/10.21944/phk8-at71 (van der Schrier et al. 2026).
The aim of this paper is to describe a combined data set with homogenised climate observations and simulated soil moisture for a large part of Europe. In principle, this is an interesting idea and paper, and it should at some point get published. However, it does still need significant work to make this suitable for ESSD.
Partly, this is because the manuscript could benefit from some good English and other editing, I made some suggestions on the attached pdf, but more work can be done. In particular, all the labels and symbols on the figures are too small and are difficult to read. Also, please think about using colour blind friendly pallettes for your figures rather than everything in blue.
In terms of the content, I feel the manuscript needs to be more explicit at the start (abstract and maybe the title) that this is a “simulated” soil moisture dataset. The abstract should also highlight that the dataset is mainly based on grassland locations.
I think the fact that the models are “uncalibrated” is a strength, which is not emphasised enough. This is simulated soil moisture purely from the data using current scientific knowledge as captured in models.
The strongest part of the dataset is the homogenised dataset for temperature, rainfall and radiation. This is valuable as this is the aligning a large dataset that was until now relatively inaccessible. This is definitely valuable as published dataset. I am less convinced about the derived soil moisture unless it comes with the above highlighted disclaimers: simulated data from uncalibrated models. Overall, the accuracy of the soil moisture data is limited unless the reasons for the model misfits can be further investigated (but that is should not be in this data paper)
I am don’t think the the drought analysis (section 5.3) should be part of this paper as this is a data paper. At the moment this leans too far into a research paper. The drought index section should simply report on the data and maybe use a recent drought mapping as verification rather than trying to analyse trends or spatial patterns.
There are several overlaps and repeats in the text that should be better managed