Climate Signal or Network Artefact? A Relative-Homogeneity and Gauge-Support Screen for Areal Precipitation Trends, Applied to 73 Greek Lake Catchments

Areal precipitation is usually not measured directly. It is estimated from data collected at individual meteorological stations that open, close and move for administrative rather than climatic reasons. Therefore, an observed trend in precipitation may have two possible explanations: a real climate change or an artefact caused by changes in the observation network. For example, the data from one station may suddenly change, or the number and location of stations used in the calculation may change. Two stages of checking were carried out in the study. First, every gauge record is compared with a composite of its neighbors to check its consistency. Second, areal precipitation is recalculated from the gauges that actually enter each estimate and their annual weights. Across the 71 Greek lake catchments carrying trends, removing every suspicious gauge and rebuilding the series leaves 14 of the original 16 field-significant increases and creates new detections. That verdict holds under every removal rule but the most aggressive; how many new detections appear depends on the reference gate. The surviving increases are therefore probably climate signal rather than gauge artefact. One limitation matters: if all gauges change in the same way at the same time, the screen cannot detect it. There was also one case of a significant decrease in precipitation. The station data themselves passed the homogeneity test, but the decrease was largely caused by the calculation becoming more dependent on a drier station.