Satellite based yield: water use relationships of perennial horticultural crops

Deposited with permission of the author. © 2011 Dr. Mark Glenn O'ConnellDrought and several seasons of historically low water allocations in the period, 1996 – 2010, created demands for reliable data relating yield and water requirement of perennial high value horticultural crops grown in the Murray Darling Basin of Australia. To cope with drought, irrigators were required to decide whether to buy or sell water on the water market, and to answer basic agronomic questions concerning the viability of crops and the amount of water to be applied to crops. The drought established the demand for detailed data describing the dependence of yield on water supply. This study focused on apple, peach, nectarine, pear, apricot, plum and wine grape crops grown in the Goulburn Valley Irrigation Region of northern Victoria. Doorenbos and Kassam (1979) produced a simple yield-water use relationship where relative yield is related to relative crop water use. Relative yield is measured as actual yield/YMAX, where YMAX is maximum attainable yield. Relative water use is measured as ET/ETMAX, where ET is actual evapotranspiration and ETMAX is evapotranspiration for standard conditions with no water stress (Crop Water Requirement; CWR). This thesis applied the Doorenbos and Kassam model, subject to the condition that increases in yield are constrained to the region, ET _ CWR. The approach required estimates of YMAX and CWR for each crop. Implementations of the Doorenbos and Kassam model are commonly based on the assumption that regional crop-specific estimates of YMAX and CWR apply. However, the major variables in the Doorenbos - Kassam model (actual yield, ET, YMAX and CWR) depend on crop vegetation cover. The yield-water use relationship of each crop/field therefore depends on the site-specific vegetation cover. Vegetation cover can be estimated using on-ground measures of fractional radiation interception (f) and/or derived from Normalised Difference Vegetation Index (NDVI) measured by satellites. NDVI data facilitate the extension of field scale findings to regional and industry scales. A model of vegetation cover was assumed whereby vegetation cover was maintained at its maximum value over a prolonged period in the midseason which encompasses yield processes and crop water use (Palmer et al. 2002). Yield can be estimated using the radiation use efficiency approach of Monteith (1977), whereby yield is related to growing season radiation interception. CWR depends on evaporative demand and the crop coefficient (Kc). Many reports have shown that Kc varies directly with vegetation cover. Field studies related yield to intercepted growing season radiation using the vegetation cover model applied to commercial crops grown in the Goulburn Valley. Estimates of YMAX were derived from the upper range of yield observations in those data. CWR was derived from analysis of satellite-based estimates of ET (METRIC; Allen et al. 2005a) and their relationship with NDVI (Tasumi et al. 2005a). The basal crop coefficient (Kcb) estimates of Tasumi et al. (2005a) were used in the derivation of CWR estimates in this study. In situ measures of f were linearly related to midseason satellite-derived measures of NDVI. Relationships conformed to responses reported for broad acre crops. Intra-seasonal variation in NDVI was also assessed, and found to conform with the adopted model for vegetation cover. The thesis successfully combined field studies and remote sensing approaches to produce water production functions for major horticultural crops grown in the Goulburn Valley. Satellite-derived measures of NDVI were found to provide an affordable, repeatable and reliable alternative to impractical ground-based observations of vegetation cover. This study implies a central role for NDVI in water productivity of perennial high value horticultural crops. Maps of NDVI thereby provide estimates of YMAX and CWR for all major horticultural crops in a region and irrigators can therefore access the data required for irrigation scheduling and undertake objective water management strategies based on the range of crop options available to the grower. The water production functions provide the ability to diagnose low water productivity providing direction for improved land and water management outcomes at farm and industry scales