Synoptic-scale dynamics and modeling of extreme precipitation events in Kenya

Abstract

Extreme and contiguous precipitation events over Kenya are usually catastrophic leading to loss of lives, destruction of infrastructure and impact food security. This work documents synoptic-scale dynamics of these events using case studies with a view to understanding their initiation, evolution and dissipation including representation in models that resolve convection explicitly. As such, various data platforms are used ranging from satellite-based products e.g the Tropical Rainfall Measuring Mission (TRMM-3B42), Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO)/CloudSat overpasses, Advanced Microwave Sounding Unit (AMSU) and Microwave Humidity Sounder (MHS), Infrared brightness temperatures at 10.8 microns from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) as well as surfaces rain-gauges, Meteorological Terminal Air Report (METAR) and European Center for Medium-Range Weather Forecasts (ECMWF) reanalysis products. The latter is used to diagnose the synoptic-scale dynamics of extreme precipitation events. A convection-permitting model known as the ICOsahedral Nonhydrostatic modelling framework (ICON) is used to simulate these events. Atmospheric preconditions to deep moist convection and viability of the ICON nested set-up to represent the 24-hour accumulated precipitation between 0600–0600 UTC (A06) were the main goals of this work. Three cases were chosen based on an objective method that compared the extremity and contiguity of a precipitation event from both satellite and surface observations where agreement between both observations was strictly required. However, where this agreement was not attainable especially due to scarcity of surface observations, satellite-based observations were adopted. The first and the second cases were selected from the Coastal and Lowlands regions of Kenya on 14 May 2007 and 29 October 2011 respectively while the third case was selected from the Central Highlands of Kenya on 29 December 2012. The Coastal and Lowlands cases were found to be dynamically similar in that they were found to have been associated with large-scale atmospheric processes. The events appeared to have been either initiated or amplified by a superposition between a Kelvin wave and an equatorial Rossby wave creating a low pressure system. In contrast, the Central Highlands case appeared to have been a locally forced system characterised by intense daytime heating that created a low pressure system. The low pressure system observed from all cases induced low-level winds towards the region resulting in convergence that promoted the ascent of air to the Level of Free Convection (LFC). Within the limitations of ERA5 data in terms of reproducing thermodynamic instabilities and local forcings, the atmosphere appeared to have been conditionally unstable in all the cases studied. This was inferred from the high amounts of warm and unsaturated air in the lower levels of the atmosphere advected from the anomalously warm ocean requiring just a trigger mechanism to release instability. Indeed, a moisture flux between the surface and the mid-levels advected sufficient moisture promoting accumulation of total column water that surpassed the 98th percentile. The persistent moisture flux aided the longevity of the event for hours. For the Lowlands and Central Highlands cases, frictional convergence was found to have played a key role while for the Central Highlands case, orographic lifting was found to have been the major dynamic process that rendered the warm and unsaturated parcels in the lower levels positively buoyant. This study revealed the underestimation of upward and downward motion by ERA5 reanalysis model where Convective Available Potential Energy (CAPE) amounts were largely underestimated. However, CAPE anomalies in the period 1979–2018 revealed that for these events, CAPE was anomalously higher relative to climatology of ERA5. A Lagrangian analysis of moisture in all cases indicated that moisture uptake took place in the neighbouring anomalously warm western equatorial Indian Ocean. This analysis further revealed the role of diabatic processes in the development and growth of the convective systems through the exchange of moisture between the Ocean and the atmosphere leading to release of instability over the region. The convection-permitting ICON model, however, could not adequately represent A06 in all cases as it produced very intense but isolated convection. However, the model reproduced the location of the precipitation object to a considerable degree of skill. Representation of Cloud Top Temperatures (CTT) observed from satellite failed as well with the model producing too warm clouds with temperatures above 250 K during the time window of the most intense convection. This was partly attributed to the quality of initial data from ECMWF in forcing the model. It is suggested that, perhaps with a higher quality initial and boundary data, the results might improve. It is therefore left as an open question and perhaps a future endeavor, whether a different ICON model set-up would greatly improve the results.