Evaluating GPM precipitation estimates in the Swiss Alps and Plateau

Owing to problems of accessibility, the harshness of the weather conditions and terrain limiting the coverage of ground-based radars to higher altitudes, precipitation measurements in mountainous terrain are more challenging than in flatter areas. At the same time, precipitation in mountainous regions is both complex and less well understood than elsewhere, and is of vital importance. Mountains act as a key source of water for large parts of the world: for irrigation, navigation, potable water and hydropower schemes.

The Dual-frequency Precipitation Radar (DPR) on the core satellite of the Global Precipitation Measurement (GPM) mission provides an interesting new perspective for precipitation measurements in mountainous regions. Its near nadir-pointing orientation means that occlusion is not the same issue as it is for ground-based radars, and it is unaffected by difficulties of ground access or harsh climatic conditions. However, its limited lateral resolution (5x5 km² at ground level) means that in mountainous terrain its lower level measurement volumes will often consist of a mixture of both terrain and precipitation, limiting the advantages gained from the nadir-pointing aspect. It is also limited in that it provides only brief snapshots of precipitation over a given area, not the continuous cover provided by gauge networks and ground-based radars. It can nevertheless provide useful complementary information to augment that from ground-based instruments, and may be of particular value in less well instrumented mountainous areas. The comparatively well instrumented nature of Switzerland makes it an excellent test bed for such measurements.

In this contribution, the level-2 GPM precipitation rate measurements are compared against MeteoSwiss radar mosaic precipitation rates (produced at 2.5 minute intervals), for data covering the first 15 months of GPM's operation. A number of approaches to this comparison are presented, including decomposition into hit error (difference between measured precipitation rates), missed precipitation and false precipitation, and also into their random and systematic components. The errors are sliced in many different ways: for variability with precipitation intensity, season (and hence dominant precipitation type) and, most importantly, location.

The key results come from comparing the relative errors in both the mountainous and flatter regions of Switzerland, where the terrain has a strong influence on both sets of measurements. Additional comparisons are made with MeteoSwiss's network of over 100 weighing gauges, which provide precipitation accumulation measurements at 10-minute intervals. Particularly in the mountains, these gauges make it possible to ascertain areas and times where GPM correctly reports precipitation but the operational radar network does not, as well as when both remote sensing methods miss precipitation, allowing an estimation of the amount of additional information added by GPM overpasses in Switzerland.