The impact of lamp environment on prediction of peach TSS content

The implementation of NIRS in horticulture involves translation of a laboratory technique to packhouse and field conditions. Reference measurements are often difficult and inconvenient, therefore taken periodically. The variation in halogen lamp output with change in ambient temperature and lamp age, the later both in the short term (from lamp power up) and long term (warm up time and longevity) is documented. The impact of such changes on predictive performance (RMSEP and bias of a PLS model) for estimation of Brix content in intact peaches is evaluated. Some lamps demonstrated an initial ‘burn in’ period from first initial use, differing in length from 40 to 220 min. Long term drifts in lamp output over an eight month period caused bias changes of as much as 5 Brix units in model predictions. The implications for practical use of this technology are discussed.