The diurnal and intra-annual distribution of extreme precipitation in the Himalayas

In south Asia, the Himalayas enhance and redistribute large-scale precipitation systems, both in winter months, associated with extratropical cyclones, and in summer months, associated with the south Asian monsoon. This precipitation is depended on by billions of people in south Asia for irrigation, power generation, and drinking in areas to which the precipitation drains. In order to predict moisture availability in this region in future years, a greater understanding is sought of the mesoscale processes determining the precipitation distribution and of the terrestrial processes determining where and when this precipitation drains.

Numerous studies have shown that high-resolution model simulations can effectively represent the mesoscale distribution of precipitation in the Himalayas. However, the skill of models in reproducing the true patterns of precipitation in these mountains remains poorly understood. To address this issue, a full year over the Himalayas is simulated continuously with the Weather Research and Forecasting (WRF) model at high resolution.

First, WRF is compared to high-altitude rain gauges. WRF is shown to simulate accurately the mean diurnal precipitation signature at high elevations for each calendar month, but over-estimates the magnitude by a factor of about 1.5. The large-scale distribution of simulated precipitation is then compared to TRMM over the year, according to which WRF accurately simulates the annual evolution of precipitation distribution along the Himalayas, as well as diurnal cycles of precipitation in different areas, but WRF greatly over-estimates the spatial coverage of the most extreme precipitation.

However, TRMM is itself an estimate and has known deficiencies, particularly with snowfall. To gain a clearer impression of the accuracy of WRF's total simulated precipitation over different parts of the Himalayas over a year, comparisons are made to observed river discharge data. WRF accurately simulates the annual cycle of river discharge in different drainage basins, but, despite WRF's over-estimation of precipitation magnitude over the year, WRF generally under-estimates surface runoff, according to the observations. This is because of the Noah land surface model (LSM), which does not accurately represent the evolution of snowpack or underground moisture storage diurnally or over the course of several months. An otherwise-equivalent one-year simulation is presented using the Noah-MP LSM, which more accurately represents these terrestrial moisture processes. The Noah-MP simulation reduces WRF's over-estimation of precipitation magnitude, according to rain gauges, but reduces WRF's under-estimation of surface runoff, according to river discharge stations. The comparison of these two simulations illustrates that a more sophisticated representation of terrestrial moisture processes better simulates not just the runoff of precipitation from the mountains, but also how much precipitation falls.