Monitoring networks set up by national weather services often follow generalized guidelines for instrument density. The World Meteorological Organization as well as the TAHMO initiative suggest station densities depending on the geographical setting. Especially in Africa, optimal densities are often not met, as a result of management issues but to a large degree in available resources required to install, operate and maintain dense monitoring networks. In such context of resource scarcity, knowledge of where and when additional monitoring adds the highest value not only improves data quality but also assists in planning networks more efficiently.  

Methods to estimate the optimal number of samples are often used in designing field campaigns. The aim of these methods is to determine the number of samples required to assess spatially averaged parameters (such as soil moisture or precipitation) at an accepted spatial accuracy. They are based on a priori information regarding the statistical distribution and spatial variability of the parameter in question, which may be available from the literature or previous campaign data.

We adopt this methodology to determine the optimal number of rainfall stations for Ghana and South Africa to achieve predetermined levels of resolution and spatial accuracy of estimated statistical modes of the spatiotemporal rainfall pattern. In order to keep the approach transferrable to other countries rainfall products such as TAMSAT or the South African Weather Service satellite rainfall product were utilized as a priori information. In addition, rainfall data from TAHMO and SAWS stations as well as the station locations were used to identify where additional stations are required and where sufficient monitoring data is available.