APHRODITE: Contribution for the early warning of heavy rainfall and consequence of landslides in Nepal

Numerical forecasts of precipitation are getting seamless, however, there are still some gaps between weather and seasonal prediction scores. Early warning of high-impact meteorological events such as heavy precipitation often linked to floods and landslides, hence, quantitative observation of precipitation and its link to disaster records as well as meteorological circulation fields may contribute for the warning of disasters caused by heavy precipitation. The Asian Precipitation - Highly Resolved Observational Data Integration Towards Evaluation (APHRODITE) has been used for improving daily as well as monthly (seasonal) precipitation forecasts by incorporating multi-model superensemble method.

Here APHRODITE precipitation data was used to investigate the relationship between patterns of monsoon rainfall and fatal landslide occurrence in Nepal. Rainfall from the South Asian Summer Monsoon (June to September) triggers on average 26 fatal landslide events each summer monsoon season. High peak annual rainfall corresponds with densely populated, mountainous districts in Nepal. Annual monsoon rainfall (1957-2016 ) was compared with the South Asian Summer Monsoon Index (SASMI). Results showed that when SASMI is negative (weak), monsoon rainfall is higher in central hill and mountain districts. The weak monsoon trough over Northern India, means moist air travels north and converges on central Nepal. Decomposing the July rainfall series using empirical orthogonal function (EOF) analysis identifies variability in the position of convergence over central Nepal when SASMI is negative. EOF 1 and 2 explain 52.8% and 15.5% of spatial variance in the data. High positive values in EOF 1 correspond with strong convergence and high rainfall in the west and south-west districts in Nepal. Negative values in EOF 2 correspond with high rainfall in the west, positive values correspond with high rainfall in the east. More fatal landslides occur when SASMI is negative, EOF 1 is positive and EOF 2 is negative. Intra-seasonal variations in the spatial pattern of monsoon rainfall may alter rainfall thresholds used for landslide warning, because both antecedent and peak rainfall are important in triggering landslides. Research is ongoing to explore the relationship between EOF mode and pre-landslide daily rainfall series.