Mesoscale precipitation systems in the Tibetan Plateau region as represented by global reanalysis, regional downscaling and satellite data between 2001 - 2016

Abstract

The Tibetan Plateau (TP) is one of the world’s most important water towers. Convective precipitation systems play an important role by providing freshwater for the highly densely populated downstream regions, but they can also cause hazardous floods. Most of the convective processes occur at mesoscale where a spatiotemporal knowledge gap still exists in climate model simulations in mountainous regions. This report aims to bring new information to this knowledge gap by analyzing how convective precipitation systems are represented in two Regional Climate Model (RCM) simulations: the Weather Research and Forecasting simulation from the University of Gothenburg (WRF_GU) and the High Asia Refined analysis version 2 (HAR). The datasets differ in the techniques used to resolve mesoscale convective processes and in how they incorporate largescale circulations. We have also conducted analyses using a coarser global reanalysis dataset (ERA5) and satellite observations (GPM) for comparison. We have defined a convective precipitation system as Mesoscale Precipitation Systems (MPSs) to include all convection on a scale larger than local. These have later been tracked using an object-based approach, which enables a more detailed temporal analysis of their characteristics and evolvement as we can follow their every timestep during a lifecycle. Our results showed that the two RCM simulations detected more MPSs over the TP compared to ERA5 and GPM. In the larger domain, HAR detected most while WRF_GU the least. We could also see that MPSs contributed with a significant amount of precipitation, even in drier regions. MPSs tended to peak post-midnight in the larger domain while during the afternoon over the TP. Due to its coarse resolution, early convection was not detected in ERA5. This was evident in its delayed timing of MPS convection and low number of detected MPSs over the TP. These findings suggest an added value of using a higher spatial resolution when analyzing convection in mountainous regions. Furthermore, our results implicate that an RCM simulation using CP and re-initialization caused excessive convection while a non-CP RCM simulation using spectral nudging resulted in the opposite. By tracking MPSs at every timestep of their lifecycle, it was also found that longer lived systems were more prone to produce intense precipitation with distinct peaks and lows during a day compared to smaller ones. We believe the findings of our research could be valuable in further analyses of mesoscale convection using RCM simulations of different spatial and temporal resolution, with and without CP. We also believe that more research such as this can help fill the spatiotemporal knowledge gap in modelling mesoscale convective processes. In the end, this could aid in accurately modelling convective systems in water scarce regions dependent on them for freshwater and affected by them in terms of hazardous floods.