Multi-scale climate change modeling study over the Greater Horn of Africa

Abstract

There has been limited regional climate modeling (RCM) studies of climate change over the Greater Horn of Africa because of challenges of modeling tropical precipitation with a limited observational rainfall network. This study customized a RCM model with particular interest in precipitation process using several precipitation data sets for validation. Various convective schemes and micro-physics sensitivities were performed. It was found that the convective scheme of MIT-Emanuel in conjunction with reducing the relative humidity threshold for cloud formation provided the most realistic simulation in terms of spatial distribution, convective partition, rainfall totals and temperature bias when compared with observations. The above RCM customization was run for approximately 40 years to determine the models ability to capture inter-annual variability and the possible climate change fingerprint over the region. The RCM is able to capture the inter-annual variability for all places and seasons for temperature. However, the positive precipitation bias limits the models ability to capture inter-annual variability of precipitation. Despite, the low inter-annual precipitation correlation, the RCM is able to simulate large scale changes in the rainfall pattern associated with the possible climate change fingerprint and the annual precipitation cycle associated with the monsoon. Since the model was able to capture possible changes associated with climate change, the model was downscaled for climate change simulations. The Finite Volume GCM (FVGCM) is used as the lateral boundary forcing for A2 scenario RCM climate change simulations. The FVGCM was compared with the other IPCC models and found to perform within the range during the contemporary climate for circulation, precipitation and temperature. Our analysis concluded that the FVGCM has a cool and wet bias compared to the other GCMs. The RCM future climate simulations, using an A2 emission scenario, show that average temperature patterns in arid to semi-arid regions are likely to have the largest temperature increases when coupled to increased drying. Coastal locations are likely to experience the smallest temperature increase in response to increased likelihood of enhanced precipitation east of the Great Lakes and the response to the ocean’s thermal inertia. Daily temperature mean increase of 2.5C is found with a shift toward more extreme heat waves. There is also a clear shift for more intense precipitation for the eastern GHA with some localized regions having a shift in the rainfall frequency. We caution the interpreation of the dynamical downscale results because we have only downscaled one GCM and biases in the GCM lateral boundary forcing and internal errors in the RCM itself. The approach was limited to one GCM because of the computational expense of dynamically downscaling and locating 6 hourly ICBC for both the contemporary and future climate. We test a “climatological†ICBC approach for climate change simulations to limit the cost factor of the simulations. We find the approach is likely to be of benefit in simulating the spatial distributions for the GHA region when the boundary is far removed from region of interest. This method may be applied to other regions within the tropics and likely useful dynamical downscale physics ensembles and mutliple GCMs. Based on the application of the dynamical downscale results, we have provided a framework to help provide a clear approach for future dynamical downscale climate change simulations.