Rain near the coast behaves differently than over the open ocean or deep inland. By day, land heats faster than water, so moist air rises over the shore; at night the pattern reverses. Such sea-breeze circulations create characteristic bands of precipitation stretched along the coast. Yet until now, no way existed to separate these 'coastal' rains from background rainfall on a global scale. This work stands apart: the nearest studies in the archive deal with drought forecasting, local observations, or pixel-level downpours, but none have attempted to isolate the coastal contribution on a planetary scale.
The authors used CMORPH satellite precipitation estimates at 0.25° resolution from 1998 to 2013. On each three-hourly image, they identified rain areas that were brighter than the surrounding background, not too large (under 10⁶ km²), and elongated along the coast. To do this, each rain contour was fitted with an ellipse and its orientation checked against the coastline. Thresholds were not chosen by hand; instead, an ensemble of 27 parameter combinations was assembled to avoid subjective bias.
Over the Malay Archipelago, the Gulf of Panama, Madagascar, and the Mediterranean, 40–60% of all precipitation turned out to be coastally linked. In some parts of the archipelago, that share reaches 66%. The diurnal cycle of such rain is more pronounced than for ordinary rain: over land the peak comes in the evening, over the sea in the early morning. The spread among the 27 algorithm variants averaged 4% of the mean, indicating the method's robustness. During the suppressed phase of the Madden–Julian oscillation and in El Niño years, coastal rains kept precipitation over islands even as oceanic rainfall declined.
This work provides the first global map of precipitation generated by land–sea interaction. Unlike neighboring studies that either describe individual cases or predict droughts, it applies an objective geometric approach to the entire planet. This allows testing how well climate models reproduce coastal convection and identifying the conditions under which it occurs.
The algorithm can be run on any gridded precipitation data, from newer satellite products to high-resolution model output. This opens the way to studying how coastal rains respond to climate change and how accurately different models capture their diurnal cycle.
The method will help meteorologists evaluate forecasts in coastal zones and aid developers of convection parameterizations in climate models.
Next steps include applying the algorithm to data from other satellites and models, and investigating in detail the physical mechanisms that govern coastal rains in different regions.
An open question in atmospheric science is why models systematically err in the diurnal cycle of precipitation over coasts. The proposed dataset allows direct testing of how well models reproduce the coastally induced portion of rain, thereby narrowing down possible causes of error.
About 20% of the world's population lives in the zone influenced by coastal precipitation, and average population density near coasts is three times the global average.