Validation of the Accuracy of the GNSS RO Temperature Data for Climate Monitoring over Africa

Abstract

Africa is facing significant impacts from temperature and climate change, with a pronounced warming trend surpassing the global average over the past century. This trend is reshaping climates and ecosystems, leading to increased temperatures, more frequent heatwaves, and altered precipitation patterns across different regions. Severe weather events like droughts, cyclones, and floods pose risks to human lives and well-being. The decline in radiosonde stations across Africa and the lack of collocated meteorological sensors at a majority of GNSS ground-based stations present substantial hurdles in effectively monitoring and responding to climate change in the region. This study thus evaluates the validity of temperature data based on the GNSS Radio Occultation (RO) technique as an alternative to the in-situ radiosonde observations for climate change monitoring over Africa. Three datasets which include GNSS RO, Radiosonde, and ERA5, were analyzed using Python programming, focusing on temperature as the key variable. The validation process involved the use of ten (10) different performance metrics for assessing temperature data from the datasets, including NRMSE (%), Bias (mm), SDR, MAE, ACC, VC, MEF, RI, -squared, and SD Residuals (mm). Most of these metrics exhibited improved performance when dealing with a temperature discrepancy of approximately 1K, although a few stations showed weak or no correlation with certain metrics. Some stations displayed temperature differences as high as 7K due to the number of Radio Occultation Events (ROE) around the stations or problematic Radiosonde data, while most stations exhibited temperature variances of around 1K. These findings suggest that the GNSS RO technique could be considered for Africa due to the current status of the Radiosonde stations over the region.