Validity and reliability of drought reporters in estimating soil water content and drought impacts in central Europe

Abstract

Increasing drought is considered one of the major threats associated with climate change in central Europe. To provide an objective, quantitative tool that represents current drought conditions, the Czech Drought Monitor System (CzechDM) was established in 2012. Like other drought monitoring systems worldwide, the CzechDM uses several approaches to provide drought data. However, the CzechDM is unique internationally due to its utilization of a network of voluntary reporters (farmers) who complete a weekly online questionnaire to provide information about soil water content and the impacts of drought on crop yield. In this study, the results from the questionnaires from individual farms were aggregated by district. Reporters' data were compared and validated with the outputs of the SoilClim model (a core tool of the CzechDM) and with other drought monitoring tools, such as the water balance model, the soil water index and the evaporative stress index. The soil water content estimated by the reporters was significantly correlated (on average r = 0.8) with the outputs of the SoilClim model. Conversely, the correlation between the drought impacts on yield estimated by the reporters and the SoilClim outputs was lower (on average r = 0.4), suggesting that in situ observations by farmers provide additional insights into the occurrence of drought impacts. Importantly, it was found that farmers reported significant drought impacts on yield earlier in the season than any other methods (models or remote sensing). The main findings of this study are that the drought monitoring provided by reporters is a useful and reliable component of the CzechDM. We conclude that weekly reports by farmers represent a significant enhancement to drought monitoring and have potential for use in developing automated approaches that combine in situ, modeling and remote sensing data within a data fusion or machine learning framework.