Predicting freshwater demand on Irish dairy farms using farm data. (10th November 2017)

• Record Type: Journal Article
• Authors: Murphy, E.; de Boer, I.J.M.; van Middelaar, C.E.; Holden, N.M.; Curran, T.P.; Upton, J.
• Is Part Of: Journal of cleaner production. Volume 166(2017)
• Page Start: 58
• Page End: 65
• Publication Date: 2017-11-10
• Subjects: Freshwater use prediction -- Milk production -- Mixed models -- Profitability; Factory and trade waste -- Management -- Periodicals; Manufactures -- Environmental aspects -- Periodicals; Déchets industriels -- Gestion -- Périodiques; Usines -- Aspect de l'environnement -- Périodiques; 628.5
• DOI: 10.1016/j.jclepro.2017.07.240
• Languages: English
• ISSNs: 0959-6526

Abstract

Abstract: Freshwater use in agriculture is a matter of discussion due to rising concerns over water scarcity, availability and pollution. To make robust predictions of freshwater demand, a large dataset of agricultural data is needed to discern the relationships between production parameters and water demand. The objective of this research was to predict freshwater demand (L yr −1 ) on Irish dairy farms based on a minimal set of farm data. A detailed water footprint (WF) was calculated for 20 dairy farms for 2014 and 2015, and the relationships between the WF and agricultural inputs explored via a mixed modelling procedure, to develop a minimal footprinting solution. The WF comprised of the consumption of soil moisture due to evapotranspiration (green water, GW) and ground and surface water (blue water, BW). The performance of the models was validated using an independent data set of five dairy farms. The GW model was applied to 221 dairy farms to establish the relationship between the GWF of milk and economic performance. The average total volumetric WF of the 20 farms was 778 L/kg fat and protein corrected milk (L/kg FPCM) (range 415–1338 L/kg FPCM). Freshwater for pasture production made up 93% of the GW footprint. Grass grown, imported forages and concentrates fed were all significant predictors of GW. The relative prediction error (RPE) of the GW model was 11.3%. Metered on-farm water and concentrates were both significant predictors of BW. The RPE of the BW model was 3.4%. When applied to 221 dairy farms ranked by net margin per hectare, there was a trend (P < 0.05) towards higher profitability as the GWF decreased, indicating that the GWF of dairy farms can be improved by implementing good management practices aligned with improving profitability. Highlights: High resolution farm productivity data was used to develop water prediction equations. The RPE for the GW and BW prediction models was 11.3% and 3.4%, respectively. Grass growth had a standardised co-efficient of 0.92 for green water prediction. Farms which had a lower WF had the highest net profit per hectare.