Prediksi Sifat Fisiko Kimia Buah Markisa (Passiflora lingularis) Secara Nondestruktif Menggunakan Spektroskopi NIR

Abstract

Sifat fisiko kimia buah markisa seperti Total Padatan Terlarut (TPT), total asam dan kekerasan menjadi parameter yang penting dalam pemilihan buah markisa, namun selama ini penentuannya menggunakan metode destruktif. Analisis destruktif membutuhkan waktu yang lama dan menyebabkan kerusakan pada buah sehingga tidak sesuai untuk penanganan segar. Diperlukan metode penentuan sifat fisiko kimia buah markisa secara cepat dan nondestruktif, yakni spektroskopi Near Infrared (NIR). Penelitian ini bertujuan untuk memprediksi sifat fisiko kimia pada buah markisa secara nondestruktif menggunakan spektroskopi NIR. Sampel yang digunakan adalah 60 buah markisa dengan tiga umur panen, yakni 104,112, dan 120 hari setelah antesis (HSA). Reflektan buah markisa diukur menggunakan NIRFlex N-500 fiber optic solid, sementara TPT, total asam dan kekerasan secara destruktif diukur menggunakan refraktometer, acidity meter dan rheometer. Kalibrasi dan validasi spektra NIR dan data destruktif dilakukan menggunakan Partial Least Square (PLS) dan Principal Component Regression (PCR). Pre-tratment Baseline, D-Trending, Normalisasi, Multiplicative Scatter Correction (MSC), Standard Normal Variate (SNV), Smoothing Savitzky-Golay (SGs) dilakukan untuk meningkatkan akurasi NIRS dan dievaluasi berdasarkan nilai koefisien korelasi (r), standar error kalibrasi (SEC), standar error prediksi (SEP), dan Ratio of Standard Error Prediction to Deviation (RPD). Prediksi sifat fisiko kimia buah markisa (Passiflora lingularis) secara nondestruktif dapat dilakukan menggunakan spektroskopi NIR dengan metode PLS dan PCR dengan tingkat akurasi yang berbeda. Prediksi PLS terbaik untuk TPT didapatkan dari pre-treatment SGs faktor PLS 15 (r=0.86, SEC=1.10 oBrix, SEC=1.10, SEP=1.16 oBrix, RPD=1.77, konsistensi=94.53%), total asam dari pre-treatment SGs faktor PLS 15. (r=0.89, RPD=2.04, SEC=0.20%, SEP=0.21%, konsistensi=96.89%), kekerasan dari data original faktor PLS 10. (r=0.74, RPD=2.06, konsistensi=97.78%, SEC=4.89N, SEP=5.00N). Prediksi dengan metode PCR memiliki akurasi yang lebih rendah daripada metode PLS dan belum menghasilkan akuasi yang baik untuk TPT, total asam dan kekerasan. Prediksi TPT, total asam dan kekerasan pada metode PCR menunjukkan hasil yang tidak signifikan antara spektra original dan pre-treatment. Kata kunci: buah markisa, kekerasan, NIRS, total padatan terlarut, total asam

Physicochemical properties such as Total Dissolved Solid (TDS), total acid and hardness were the important things to choose passion fruit. However, destructive methods have been used to determine it. Destructive analysis is not suitable for handling fresh passion fruit because it takes a long time and damages the fruit. It is necessary to determine the physico-chemical properties of passion fruit quickly and non-destructively, such as Near Infrared (NIR) Spectroscopy. The aim of this research is to predict the physicochemical properties of passion fruit using NIR spectroscopy. The samples used were 60 fruits with harvest age of 104, 112, and 120 days after anthesis. The reflectance of passion fruit was measured with NIRFlex N-500 fiber optic solid, meanwhile TDS, total acid and hardness were measured with refractometer, acidity meter and rheometer. Calibration and validation of NIR spectra and destructive data were conducted with Partial Least Square (PLS) dan Principal Component Regression (PCR) method. Baseline pre-treatment, D-Trending, Normalize, Multiplicative Scatter Correction (MSC), Standard Normal Variate (SNV), Smoothing Savitzky-Golay (SGs) were conducted to increase the number of NIRS accuration and evaluated by the value of coefficient of correlation (r), Standard Error Calibration (SEC), Standard Error Perdiction (SEP), and Ratio of Standard Error Prediction to Deviation (RPD). Nondestructive prediction of physicochemical properties on passion fruit (passiflora lingularis) can be conducted using NIR Spectroscopy by PLS and PCR method with different accuration. The best total dissolved solid prediction produced by SGs’ pre-treatment with 15 PLS’ factors (r=0.86, SEC=1.10 oBrix, SEC=1.10, SEP=1.16 oBrix, RPD=1.77, consistency=94.53%.), total acid by SGs’ pre-treatment with 15 PLS’ optimum factors (r=0.89, RPD=2.04, SEC=0.20%, SEP=0.21%, consistency=96.89%.), and hardness by originial spectra with 10 PLS’ factors (r=0.74. RPD=2.06, consistency=97.78%, SEC=4.89N, SEP=5.00N). Accuration by PCR method is lower than PLS method and wasn’t produce good accuration for TDS, total acid and hardness. Prediction of TDS, total acid and hardness by PCR method showed insignificant result between original spectra and pre-treatment. Keywords: hardness, NIRS, passion fruit, total acid, total dissolved solid