Abstract:
The truncated ecosystems of the Tana River County are highly vulnerable to climate change and
variability due to their low adaptive capacities and high dependence on climate sensitive
resources. Inadequacy of long term climate information is a serious constraint for long-term
planning for enhanced food security and minimization of the adverse impacts of climate change
and variability. This study was motivated by the need to downscale climate information using
modelling procedures based on Regional Climate Models (RCMs). The objectives of the study
revolved around evaluations of the performance of Coordinated Regional Climate Downscaling
Experiment (CORDEX) RCMs in simulating rainfall and temperature conditions and use of these
data sets in projecting future climate change scenarios and their implications on agricultural
productivity and related resources. Assessments and validation tests were run to authenticate the
plausibility of CORDEX RCMs and the relevance of historical climate data in evaluations of the
impact of climate change and variability on agricultural productivity. Agricultural data (crops
and livestock) for more than 20 years collected from the Ministry of Agriculture, Livestock and
Fisheries (MALF) departments in Tana River County were utilized in the study. The gross yield
of five widely grown crops in the region comprising of maize, green grams, rice, cassava and
mangoes was collated. Livestock population data for specific livestock species was used.
Subjective sampling was applied for three focused group discussions conducted. Bi-variate
correlations and simple linear regressions were used to investigate crop/livestock production and
rainfall relationships. Combination of dynamical and statistical downscaling approaches were
used in RCMs evaluation and projecting the future climate scenarios for Tana River County.
RCMs simulated above 84% observed climatology in Tana River County making them valuable
tools for agricultural production planning. The ensemble model had better agreement with
ground data observations than individual models. Seasonal rainfall variability was of the order of
70% during short and long rains making rainfed agriculture unreliable. Crop yields showed low
correlations with March-May (MAM) seasonal rainfall (r = 0.3) as compared with OctoberDecember (OND) season (r = 0.55). Seasonal rainfall explained 8% of the variation in maize
yields and 40-56% in livestock numbers. The OND season is more reliable for agricultural
production activities in the region. A warming trend in the region of 3.0 to 3.5oC under
RCP4.5/8.5 scenarios is projected by the middle of 21st century. A warming climate in the region
will negatively impact food production, water availability and livelihood systems in the region.
