I am pleased to invite you to our fifteenth Sussex Data Science Meetup. As is the way of things at the moment, this meeting has moved online and will be conducted remotely via zoom. We are fortunate enough to have Edward Efui Salakpi, a doctoral researcher from the University of Sussex who will be talking about:
Forecasting the Grassland Vegetation Condition with Bayesian Auto-Regression Distributed Lags Models
Amongst all the natural disasters reported yearly, drought events are the most prevalent. Their devastating effects are mostly felt in the grass and shrublands of arid and semi-arid lands (ASALs) within the East African region. The people living in these ASAL regions are mostly pastoralists who depend solely on the grass and shrubs as fodder for animals.
Unfortunately, the recurring drought conditions in these areas adversely affect livestock farming resulting in limited cash flow and loss of livelihoods. Addressing this challenge requires Early Warning Systems (EWS) that do not only monitor but accurately forecast drought events.
Current research on drought forecast with univariate indicators has worked well for short-range (4-6 weeks) forecasts. However, drought events are known to be influenced by other environmental and biophysical factors that occur before and during these drought events. The aim of our current research to build on the existing drought models and extend the forecast range by including exogenous factors like precipitation, soil moisture and land surface temperature derived from Earth Observation datasets.
The talk will be based on work done by researchers working on drought forecast under the University of Sussex AstroCast project. It will cover the use of an Auto-regressive distributed lag (ARDL) model implemented within a Bayesian Framework for modelling and forecasting vegetation based drought condition.
We will close the meeting with open discussions and announcements from the membership.