GEOSPATIAL DATA FUSION AND MODELING FOR IMPROVED IRRIGATION OF RICE (ORYZA SATIVA) IN THE MWEA IRRIGATION SCHEME, KENYA.

A B S T R A C T Water being the most perilous abiotic stress to crop growth, the performance of irrigation systems in the largely arid and semi-arid Kenya is critical in increasing agricultural production. The study aimed at improving the efficiency of irrigation systems for rice in the Mwea Irrigation Scheme during the major growing season from August to December 2022. The study used datasets which include: Vegetation Health Index (VHI), Leaf Area Index (LAI), climate data, crop data, soil data. Random Forest Algorithm was used to fuse VHI and LAI to obtain the crop water content and using the evapotranspiration calculated based on Penman Monteith Algorithm using Climate data, crop coefficient KC values at the different growth stages of rice is obtained. The CROPWAT model was used to integrate the datasets to estimate irrigation water requirements at the different growth stages and develop an irrigation schedule. From the results, the spectral indices ranged from 0-1with values near zero indicating adequate crop evapotranspiration, good crop condition implying no water stress while higher values close to 1 may indicate crop water stress. The temporal fluctuation in estimated CWC-derived Kc values showed moisture stress events during the season where reduced CWC caused a corresponding drop in Kc values. From the calculation of irrigation water demand, irrigation water increases gradually and reaches its peak at the mid/grain filling stage and decreases in the late stage. From the irrigation schedule developed, schedule based on daily monitoring of soil moisture balance is more robust in contrast with a schedule based on regular time interval period which may lead to over-irrigation or under-irrigation. A validation approach for the applicability of CROPWAT in Mwea Irrigation Scheme should however be developed.