Quantifying precipitation estimates and its uncertainty from a dense Commercial Microwave Link network in Nigeria

High-resolution accurate and timely rainfall estimates are essential in many hydrological applications, ranging from flood early warning to urban water management, and essential in many agricultural services. However, many regions in the world, predominantly in the Global South, lack sufficient coverage from dedicated ground-based rainfall sensors such as weather radars and rain gauge networks, and thus have to rely on satellite rainfall products. These products, however, have the downside that their spatial or temporal resolution is often too low for hydrological applications at the kilometer scale. Moreover, many of these products have limited accuracy with regards to measuring near-surface rainfall, especially in the tropics.

An ‘opportunistic’ alternative for high resolution near-surface rainfall estimates comes from the signal attenuation experienced by commercial microwave links (CMLs) in cellular communication networks. When it rains, the radio signal between two cell phone towers is (partially) attenuated, and this rain-induced attenuation can be used to infer the average rainfall intensity along the path. Typically, every 15 minutes the minimum and maximum received signal levels are stored in network management systems by mobile network operators for quality monitoring purposes. Based on these signal levels it is possible to estimate path-averaged rainfall intensities, which can be interpolated to produce high-resolution rainfall maps. Several studies have already shown the potential of this opportunistic measuring technique on the African continent, though most with a relatively small CML data set.

In this study we investigate the use of several thousands of CMLs in Nigeria, predominantly located in heavily urbanized areas across the country. We compare the path-averaged rainfall intensities from these CMLs to the few rain gauges in the area, and quantify the uncertainty range in such a dense CML network. We do a similar comparison by comparing interpolated rainfall maps from CMLs to available gridded (satellite) rainfall products on a seasonal basis. As such, we show the added value and the associated uncertainties in measuring rainfall using this opportunistic source of rainfall estimation in a region that typically lacks this hydrometeorological information.