Malaria model outputs (LMM and VECTRI). Talib et al., 2024

Two climate-sensitive dynamical malaria models, the Liverpool Malaria Model (LMM) and Vector-Borne Disease Community Model of the International Centre for Theoretical Physics (VECTRI), are used to investigate the effect of explicitly representing convection on simulated malaria transmission. Both models are driven by the same climate variables: daily-accumulated precipitation; and daily-mean near-surface (2 m) air temperature. Whilst the LMM can only be used to investigate the climate suitability for malaria transmission, VECTRI also considers the impact of surface hydrology and population density.In addition to driving malaria models with observations, we perform a set of malaria transmission experiments using climate model outputs from Met Office Unified Model (MetUM) pan-African simulations. The MetUM is a non-hydrostatic model with a semi-implicit, semi-Lagrangian dynamical core. Full details of the model specifications and setup are provided by Stratton et al., 2018 and Kendon et al., 2019.To compare differences in simulated malaria transmission when using observations or climate model data, we perform three historical simulations for each malaria model. These simulations are either driven by observations, CP4 or R25 historical data, and enable us to understand the sensitivity of simulated malaria transmission to observed and model input data. For both VECTRI and the LMM, we also perform a set of simulations using future climate model data. The use of future climate model simulations enables us to investigate the effect of explicitly representing convection in both present and future climates. In particular, it enables us to assess whether explicitly representing convection changes predictions of malaria transmission. To isolate the importance of differences in simulated temperature and precipitation when changing the representation of convection, we perform simulations with either temperature or precipitation sourced from CP4, with the remaining variable sourced from R25. We also perform sensitivity experiments with which we only use precipitation or temperature from the future climate. Whilst we note that precipitation and temperature are not independent of each other, sensitivity experiments enable an initial understanding of the key drivers responsible for differences in malaria predictions. Due to limited computational resources, sensitivity experiments are only performed using the LMM.Given substantial errors in simulated precipitation and temperature in both climate model configurations, we also perform LMM experiments driven with bias-corrected, linearly-scaled simulation data. For temperature and precipitation data, we applied an additive and relative adjustment respectively. The use of linear-scaling only corrects time-mean biases, hence, variability errors still persist.