Effect of sowing date distributions on simulation of maize yields at regional scale – A case study in Central Ghana, West Africa. (September 2016)

• Record Type: Journal Article
• Authors: Srivastava, Amit Kumar; Mboh, Cho Miltin; Gaiser, Thomas; Webber, Heidi; Ewert, Frank
• Is Part Of: Agricultural systems. Volume 147(2016)
• Page Start: 10
• Page End: 23
• Publication Date: 2016-09
• Subjects: Maize -- Sowing date -- Probabilistic modelling, sub-Saharan Africa; Agricultural systems -- Periodicals; Agriculture -- Environmental aspects -- Periodicals; 338.16
• DOI: 10.1016/j.agsy.2016.05.012
• Languages: English
• ISSNs: 0308-521X

Abstract

Abstract: In sub-Saharan Africa, with its high rainfall variability and rainfed agricultural production system of maize ( Zea mays L), the estimation of its sowing date is a crucial decision for farmers. To support decision making in rainfed agriculture, different methods using "probabilistic approaches" for the selection of the sowing dates at the regional level has been developed where most of the times we only have information about the probable sowing period. The crop model LINTUL5 embedded into a general modelling framework, SIMPLACE (Scientific Impact Assessment and Modelling Platform for Advanced Crop and Ecosystem Management) has been combined with a multilayer soil water balance model (SLIM) to simulate maize yields in Central Ghana. Different assumptions about the sowing date distributions at the regional level were compared to the corresponding deterministic approaches. The simulated regional maize yields with the probability-based approaches showed always the lower RMSE compared to the deterministic approaches, although significant in all cases. The approach A4-S4, where we assumed that sowing dates are normally distributed around the sowing day estimated with a rainfall based rule were the best approach in capturing the spatial and temporal variability of maize yields at the regional level. The assumption of a probabilistic distribution of sowing dates within a given sowing period tends to be superior to deterministic sowing date selection because the decisions about sowing dates are often driven by factors like availability of labor, capital or seeds and are hence much more complex than those assumed in existing crop models. Highlights: Effect of sowing date distribution on simulated crop yield is estimated. Probability approach of estimating plating date has been proposed. Probability approach tends superior to deterministic planting date methods.