Evaluation of Spectroscopy and Methodological Pre-Treatments to Estimate Soil Nutrients in the Vineyard

The characterization of vineyard soil is a key issue for crop management, which directly affects the quality and yield grapes. However, traditional laboratory analysis properties tedious both time cost consuming, not suitable precision viticulture. For this reason, fast convenient technique needed assessment management. Here, spectroscopy appears as alternative to assist analysis. This work focuses on estimating by spectroscopy. Our study was carried out using 96 samples collected from three vineyards in Rias Baixas Designation Origen (Galicia, Spain). soils that were characterized include nitrogen (N), organic matter (OM) clay content (Clay). presented compared two regression techniques (partial least squares (PLSR) random forest (RF)) four spectral ranges: visible—VIS (350–700 nm), near infrared—NIR (701–1000 short wave infrared—SWIR (1001–2500 nm) VIS-NIR-SWIR (350–2500 order identify more prediction models. Moreover, effect pre-treatments reflectance data (smoothing Svitzky–Golay, SG, baseline normalization, BN, first derivative, FD, standard normal variate, SNV, logarithm 1/reflectance or (SP) detrending, SNV-D) evaluated. Finally, continuous maps created based estimated values results identified PLSR best technique, with less computation than RF. improved after applying transformation data, pre-treatment (logarithm 1/Reflectance). performances have obtained determination coefficients (R2) 0.69, 0.73 0.52 nitrogen, matter, clay, respectively, acceptable accuracy (RMSE: 0.03, 1.06 2.90 %) time. Furthermore, mapping generates information high interest viticulture well comparison between methodologies used.