Hyperspectral Imaging of Adaxial and Abaxial Leaf Surfaces for Rapid Assessment of Foliar Nutrient Concentrations in Hass Avocado

Round 1

Reviewer 1 Report

Manuscript, entitled (Hyperspectral imaging of adaxial and abaxial leaf surfaces for rapid assessment of foliar nutrient concentrations in Hass avocado. Foliar potassium concentrations were predicted successfully only from the adaxial surface (R P = 0.56, RPD = 1.54). Therefore, adaxial surfaces can be used to predict most foliar nutrient concentrations. However, foliar sodium concentrations were predicted successfully (RP = 0.59, RPD = 1.58) only from the combined images of both surfaces. Hyperspectral imaging showed great potential as a rapid assessment tool for monitoring the crop nutrition status of avocado trees. There are several shortcomings that should be included in order to enhance the manuscript for the readers.

Abstract

·         Please do not use the pronouns in scientific writing?  Such as we aimed in line 16 and we successfully in line 22.

·         What is the conclusion of the abstract?

Introduction

·         Line 48. Please add the hyperspectral imaging can indirect predict.

·         The introduction is poor written and very short to present the novelty of this study as well as it should be improved. Several previous studies of this work should be presented with more details.

·         Please add the hint about the basic of remote sensing for detect foliar nutrient concentrations?

·         Please write about the advantage and disadvantage of ground-based hyperspectral imaging system and unmanned aerial vehicles since they are mentioned in this study.

·         Please highlight in introduction, what is the novelty (originality) of the work? And what is new in your work that makes a difference in the body of knowledge?

Materials and methods  

·         Scientific writing must improve; the authors several times used the pronouns in materials and methods with separate sentences.

·         Table 1 should be added in results section.   Results

·         I think that the data of PLSR model is not enough to do an integrated search in a satisfactory way for the reader. For that i suggest for the authors to do:

1-      Please could the authors support this study by extract the best spectral - -indices with R²

 

2-      Please could the authors present the other machine learning model such as ANN. I think the data could be improved Na and S.

    Discussion ·         The discussion is needed to improve by above suggestions.   Conclusions   ·         Please write about the limitations of this work in details in conclusion section.

Extensive editing of English language required

Author Response

Please see attached 

Author Response File: Author Response.docx

Reviewer 2 Report

I'm reviewing "Hyperspectral imaging of adaxial and abaxial leaf surfaces for rapid assessment of foliar nutrient concentrations in Hass avocado".

The exposed research aim to obtain leaves nutrient concentration from laboratory hyperspectral analysis. An empirical model is trained on samples measurements and analysis.

The manuscript is well written and provide the necessary background to frame the experiment.

 

You state that sampling took place on 30 tree taking 4 samples/tree. This fit with the resulting 120 samples mentioned.

Randomly subsetting calibration and validation set is most likely validating on a tree quadrant the model trained on the remaining three quadrants.

No clear mention is given about within within-tree heterogeneity across quadrants; in case quadrants were similar one the others the model validation performance could be affected by overfitting.

 

The proposed model performance would be more accurately measured by validating on six random trees (24 quadrants) the model trained on 24 trees (96 quadrants).

 

Except the point above the experiment is soundly conducted and the results clearly stated.

I expect minor reviews to be applied to the manuscript in order to get it publication ready.

 

Author Response

Please see attached 

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript reviewed (id remotesensing-2381583) is entitled "Hyperspectral imaging of adaxial and abaxial leaf surfaces 2 for rapid assessment of foliar nutrient concentrations in 3 Hass avocado".

The authors performed assessments in order to “monitor crop nutrient status and manage fertiliser applications in real time” using hyperspectral imaging (HSI). They aimed to “determine the potential of HSI for predicting foliar nutrient concentrations in avocado trees and establish whether imaging different sides of the leaves affects prediction accuracy”. To this purpose, “laboratory based hyperspectral images (400–1000 nm) were taken of both leaf surfaces collected from Hass avocado trees at 0, 6, 10 and 28 weeks after peak anthesis” and “developed partial least squares regression (PLSR) models to predict mineral nutrient concentrations using images from (a) abaxial surfaces, (b) adaxial surfaces, and (c) combined images of both leaf surfaces. According to the authors, “HSI has not been used to assess avocado crop-nutrient status”.

In terms of language, scientific vocabulary, expressions and syntax the manuscript does not require further editing by the authors. The bibliography and references to previous work on the field relatively adequately cover the subject.

 Specific comments/suggestions:

 Par. 2.1. The authors sampled one young fully-expanded leaf per quadrant of each one of the 30 Haas trees under study, at approximately 1.5 m from the ground for 0, 6, 10, 28 weeks after anthesis. Is the use of one leaf per quadrant representative of the whole tree foliage within each quadrant and why?

Par. 3.2. Please provide in a table the relative regression equations that you obtained for each nutrient and leaf surface.

 

Generally, the scientific work and results presented in this manuscript are interesting mainly due to (i) the use of hyperspectral data for this type of study (ii) the separate use of three leaf surfaces (adaxial, abaxial and combined). Although (i) the regression analysis that the authors adopted is not innovative and (ii) the authors did not exploit the wealth of spectral information they had at their disposal (400 spectral bands covering the spectral area of 400nm-1000nm) in order to conduct more advanced spectroscopy (e.g. investigate potential spectral features linked to nutrients), they achieved the purpose of their study as they described it and filled the relative gap in bibliography.

Therefore, provided that the authors take into consideration my two previous specific comments, I am in favor for the publication of this manuscript.

Author Response

Please see attached 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors did significant improvment. It can be accepted for publication.