**1. Introduction**

Insufficient water availability, owing to variability in rainfall patterns and frequent dry spells exacerbated by climate change [1,2], threatens food security and rural livelihoods in Sub-Saharan Africa (SSA) [3]. In SSA, more than 95% of arable land is under rainfed crop production, which contributes 81% to the regional food basket [4,5]. Because of variable rainfall and low-input cultivation [6,7], grain yields are only from 1 to 2 Mg ha−1, whereas attainable yields range between 4 and 5 Mg ha−<sup>1</sup> in SSA [5,8]. Furthermore, risks of crop failure in SSA have increased due to land degradation and soil nutrient depletion [9,10], signified by negative annual NPK balances with −26 kg ha−<sup>1</sup> N, −7 kg ha−<sup>1</sup> P2O5, and −23 kg ha−<sup>1</sup> K2O, as reported in [11]. On a continental scale, annual NPK losses averaged 54 kg ha−<sup>1</sup> (and ranged between 9 kg ha−<sup>1</sup> in Egypt and 88 kg ha−<sup>1</sup> in Somalia), resulting in land degradation in more than 40% of Africa's total farmland [12,13]. These risks have further reduced the already insufficient financial capacity of farmers to invest in sustainable land management (SLM) strategies [3,5]. However, such strategies are key for optimizing trade-offs between food production and other agro-ecosystem services [12]. In water-scarce environments such as the Upper East region of Ghana (UER), sustainable soil-water management has been identified as the most influential among agricultural management practices, including soil fertility management, selection of crop varieties, and control of pests and diseases [5,14], for enhancing food security as well as improving the smallholders' livelihoods [5,15–17].

The reservoir-based irrigation schemes in SSA, which store water (i.e., mostly surface runoff) in the rainy season, were originally designed to supply water for dry season crop irrigation, the livestock sector, fish farming, and domestic use, excluding supplemental irrigation in the rainy season. However, increasing climate variability calls for exploring the feasibility of supplemental irrigation for crop cultivation in the rainy season [3]. Supplemental irrigation has considerable potential to increase grain yield, particularly if provided during the critical stages of the crop growing cycle (i.e., booting and grain filling) [18].

Because of increasing competition for stored water in the dry season, the extra water demand for supplemental irrigation to bridge dry spells is likely to result in a mismatch between water supply and demand in the reservoir-based irrigation schemes. Thus, the requirement for supplemental irrigation might be satisfied with water saved through increased irrigation efficiency as a result of improving dry season irrigation scheduling [19]. As long as increased irrigation efficiency is accompanied by yield increments, this provides incentives for irrigators to engage in SLM [5,15,18,20]. Consequently, crop–water–soil–atmosphere models will be useful to determine the most appropriate irrigation schedules for the prevalent cropping practices and for assessing possible alternative scenarios [21–23]. Among the common models capable of simulating irrigated crop growth, those requiring large inputs of primary data, for instance APSIM [24] and CropSyst [25], and that are not available for free, such as the irrigation scheduling model ISAREG [26], might not be favorable for applications in SSA. The DSSAT model [27] has been commonly used to assess the impact of agronomic inputs on irrigated crop yield but at present is not suitable to evaluate the effectiveness of irrigation practices. Some other models, such as CROPWAT [28] do not distinguish between evaporation (i.e., non-beneficial water consumption) and crop transpiration, and do not provide an estimation of yield or, as with EPIC [29], apply simplified routines to evaluate the groundwater contribution to crop water use. Due to relatively modest data requirements, consideration of all major agro-hydrological processes, and its free availability, the AquaCrop model developed by the Food and Agriculture Organization of the Unites Nations (FAO) [22] has found many applications worldwide, including in SSA [30–32].

Current irrigation schedules in reservoir-based irrigation schemes in SSA are based on locally established rules governing access to water for irrigation, but with little consideration of crop- and site-specific water demands in terms of quantity and timing, resulting in the over-irrigation of crops [19]. For instance, in reservoir-based irrigation schemes in onion fields in the UER, the ratio of total water supply to gross irrigation demand ranged between 2.4 and 5.7 during dry season crop irrigation [33]. The problem of over-irrigation in reservoir-based irrigation schemes was further confirmed by gross irrigation amounts (GIAs) ranging from 380 to 852 mm for dry season tomato production in the UER [34], and between 274 and 838 mm for tomato cropping under groundwater irrigation in the same region [35]. Simulations have suggested that the net irrigation requirement (NIR) for dry season tomatoes ranges from 359 to 372 mm in the reservoir-based Koga irrigation scheme in Ethiopia [36], emphasizing the need as well as the potential to improve water management through irrigation scheduling to reduce water losses and increase productivity.

To the best of our knowledge, no study has attempted to develop an irrigation schedule for dry season cropping systems in the UER. Moreover, the limited number of studies on supplemental irrigation in SSA have not explored the feasibility of using dry season water savings in reservoir-based irrigation schemes. For example, Sanfo et al. [3] investigated the economic value of supplemental irrigation of grain crops using farm ponds of 300 m3 capacity in south-western Burkina Faso, and reported that in years of low rainfall, supplemental irrigation could be a cost-effective intervention to reduce risks of crop failure and increase farmers' incomes. Fox and Rockström [37] also assessed the effect of supplemental irrigation, based on 150 m3 capacity farm ponds, on the grain yield of sorghum in northern Burkina Faso and found that supplemental irrigation alone resulted in an approximately 56% increase in grain yield, making it a useful technology to mitigate dry spells and shorten the yield gap. Similarly, Mustapha [16] studied the water productivity of pearl millet under supplemental irrigation applied at five different crop growth stages in Nigeria and reported that the supplemental irrigation amount of 84 mm applied at booting and grain filling stages could result in a 69% increase in yields.

This study aims to improve the traditional dry season irrigation practices in reservoir-based irrigation schemes in the UER, and to assess the potential for introducing supplemental irrigation in the rainy season as an adaptation to climate change. To this end, we (i) parameterized and validated the AquaCrop model to render applications for irrigated crop production in the EUR of Ghana (ii) assessed the appropriateness of current water management in the typical tomato–maize rotational system; (iii) developed an improved irrigation schedule for dry season cultivation of tomato; and (iv) determined the requirement for supplemental irrigation of maize in the rainy season under different climate scenarios.
