Pendugaan Nilai Kemanisan dan Kekerasan Buah Anggur menggunakan NIRS (Near Infrared Spectroscopy)

Abstract

Anggur merupakan buah non klimaterik, yaitu buah yang pola respirasinya menurun dan tidak menunjukkan adanya kenaikan konsumsi O2 dan produksi CO2 yang mencolok. Hingga saat ini penentuan tingkat kematangan buah anggur masih dilakukan berdasarkan pengamatan visual tanpa bantaun alat. Oleh karena itu jika buah anggur dipanen belum cukup umur, maka akan mempunyai kualitas yang rendah karena belum matang sempurna. Deteksi tigkat kematangan yang cepat dan efisien dapat diwujudkan melalui penggunaan teknologi Near Infrared Spectroscopy (NIRS). Penelitian ini bertujuan mempelajari karakteristik kematangan buah anggur melalui parameter total padatan terlarut (TPT) dan kekerasan serta spektrum NIRS dari buah anggur dengan umur panen yang berbeda serta mengembangkan model kalibrasi untuk prediksi nilai TPT dan kekerasan berdasarkan reflektan spektra NIR meggunakan metode PLS. Sampel yang digunakan yaitu 240 buah anggur dengan tiga umur petik yang berbeda yaitu masing-masing 85, 90, dan 95 hari setelah bunga mekar (HSBM). Pengukuran reflektan buah anggur dilakukan menggunakan alat spektrometer NIRFlex N-500 fiber optic solid. Pengukuran total padatan terlarut dan kekerasan secara destruktif dilakukan menggunakan rheometer dan refraktometer. Kalibrasi data NIRS dan data total padatan terlarut serta kekerasan dilakukan menggunakan Partial Least Square (PLS) untuk mendapatkan model kalibrasi terbaik dalam memprediksi total padatan terlarut dan kekerasan. Pendugaan total padatan terlarut terbaik diperoleh melalui spektrum reflektan dengan pra-pengolahan data menggunakan Three points Smoothing Avarage pada faktor PLS 17 (r = 0,83; standard error calibration (SEC) = 0,47 %brix; standard error prediction (SEP) = 0,44 %brix; ratio of standard error prediction to deviation (RPD) = 1,89; konsistensi = 108,66 %). Pendugaan kekerasan terbaik diperoleh melalui spektrum reflektan original atau tanpa pra-pengolahan data dengan faktor PLS 14 (r = 0,89; standard error calibration (SEC) = 0,98 N; standard error prediction (SEP) = 1,04 N; ratio of standard error prediction to deviation (RPD) = 2,06; konsistensi = 94,33 %).

Grapes are non-climacteric fruits, meaning that the fruits whose respiration pattern decreases and does not show a significant increase in O2 consumption and CO2 production. Until now, the determination of the level of ripeness of grapes is still conducted based on visual observations without the help of tools. Therefore, if the grapes are harvested not yet old enough, they will have low quality because they are not yet fully ripe. Fast and efficient detection of ripeness levels can be realized through the use of Near Infrared Spectroscopy (NIRS) technology. This study aims to study the maturity characteristics of grapes through the parameters of total dissolved solids (TSS) and nirs hardness and spectrum of grapes with different harvest ages and develop a calibration model for the prediction of TSS values and hardness based on reflectance of NIR spectra using the PLS method. The samples used were 240 grapes with three different picking ages, namely 85, 90, and 95 days after blooming. The reflectance measurement of grapes was carried out using a solid fiber optic NIRFlex N-500 spectrometer. Measurements of total dissolved solids and firmness were destructively carried out using a rheometer and refractometer. Calibration of NIRS data and data on total dissolved solids and firmness were performed using Partial Least Square (PLS) to obtain the best calibration model in predicting total dissolved solids and firmness. The best estimate of total dissolved solids was obtained through the reflectance spectrum by data pre-processing using the Three-points Smoothing Avarage data at PLS 17 factor (r = 0.83; standard error calibration (SEC) = 0.47 %brix; standard error prediction (SEP) = 0.44 %brix; ratio of standard error prediction to deviation (RPD) = 1.89; consistency = 108.66 %). The best firmness estimation was obtained through the original reflectance spectrum or without pre-processing data with a PLS factor of 14 (r = 0.89; standard error calibration (SEC) = 0.98 N; standard error prediction (SEP) = 1.04 N; ratio of standard error prediction to deviation (RPD) = 2.06; consistency = 94.33 %).