Modelling the soil water balance of potatoes for improved irrigation management

Abstract

Soil Water Balance (SWB), is a generic and mechanistic crop growth model that has been successfully used to model the water balance of several crops. Its ability to combine crop water modelling and irrigation scheduling approaches allows it to be used as a research tool and an irrigation management tool. Since SWB is a tool that could be used as decision making tool for farmers, its accuracy in simulating crop growth, development and soil water balance should be high. To highlight the importance of improved irrigation management for potato crop by the means of a mechanistic soil water balance model and the importance of the photoperiod factor in potato modelling in sub-tropical region, two potato experiments were carried out in two contrasting seasons, namely, spring and autumn. Growth and development responses of potato under both well irrigated and water stressed conditions for spring and autumn plantings were examined. This study successfully quantified the water use and potato growth responses to water stress. The water use efficiency varied with irrigation treatments and planting time, and autumn experiment had generally higher values than spring. Unstressed treatment gave the highest tuber yields irrespective of planting season and marketable tuber yield was higher in autumn than spring. Water stress imposed at tuber initiation until end of tuber bulking was revealed to be the most detrimental to biomass and tuber production. This suggests that water stress at tuber initiation and bulking stage should be avoided if high tuber yield is the target. Growth analysis data were used to determine crop parameters for SWB calibration and validation. The model simulated reasonably well growth, development and soil water balance in both unstressed and stressed conditions. However, simulations results of total and harvestable dry matter towards the end of the exponential tuber bulking stage (50 - 65 DAP) were deteriorated. As a result, the model did not simulate accurately the final yield. This is an indication that the model fails to simulate the size of the canopy and its duration. The time at which tuber initiation commences appeared not be affected by the planting seasons since variation of the duration between emergence and tuber initiation in different seasons was small. This small variation could be attributed to the fact that the potato growing season in South Africa (Pretoria) in spring 2004 and autumn 2005 experiences minimum and maximum temperatures which are acceptable for the growth of potato. In Pretoria, emergence and tuberisation take place under relatively cool temperatures late in September and also early in April when temperatures are relatively cool. Consequently, potato grown in this period may escape the early autumn and late spring high temperatures. However, autumn planting experiences an abrupt change of day lengths from long days to short days towards tuber initiation. This brusque change of day length may change the crop physiology and affect the subsequent normal course of plant growth. If the day length factor could be integrated into SWB, it appears that the model will better simulate potato growth and development. The poor simulation results of total dry matter and harvestable dry matter early in the growing season suggest that the model should be improved by allowing it to simulate the start of tuber initiation. A linear function of average temperature between a base and an optimal temperature corrected with photoperiod factor was found to be the most appropriate method to estimate thermal time required for tuber initiation. This method suggests that the time of tuber initiation can be estimated from its thermal time within two days.