Development of a decision support system for crop disease monitoring, surveillance and prediction in Bomet county, Kenya

Abstract:

The study proposes to use Geographic Information Systems and Remote Sensing techniques to spatially model Maize Lethal Necrosis (MLN) disease in maize growing areas in Kenya. Results from this work will be used for prediction, monitoring and to guide intervention on MLN. This will minimize maize yield losses resulting from MLN infestation and thus safeguard the livelihoods of maize farmers in Kenya. MLN was first reported in Kenya in September 2011 in Bomet county. It then subsequently spread to other parts in Kenya. Maize crops are susceptible to MLN at all growth stages. Once infected the only option left for the farmers is to burn their maize plantations. Infection rate and damage is very high affecting yields and sometimes causing complete loss of maize yield.The modelling exercise will cover the period prior to and after the incidence of MLN. Specifically, the analysis will integrate spatio-temporal information on maize phenology and field surveys with the intention of delineating the extent of MLN infestation and the degree of damage as a result of MLN. Additionally, the task will identify potential predisposing factors leading to MLN resurgence and spread and to predict potential areas where MLN is likely to spread and to estimate the potential impact of MLN on the farm holders. The area of study for this task will be Bomet County. Historical and current environmental and spatial indicators including temperature, rainfall, soil moisture, vegetation health and crop cover will be fed into a model in order to determine the main factors that aide the occurrence and the spread of MLN. Multi-spectral image processing will be used to produce indices to study maize crop health whilst image classification techniques will be used to identify crop cover clusters by differentiating the variations in spectral signatures in the area of study and hence distinguish infected, unaffected maize crops and other crop cover classes. Variables from these indicators will then be weighted in a spatial model and be used as a basis for generating site-specific MLN prediction maps that will guide policy on MLN management in Kenya. The broader
objective is to document a model that can be up-scaled and replicated in other maize producing areas in Kenya affected by MLN.