Using a Reference Color Plate to Correct Smartphone-Derived Soil Color Measurements with Different Smartphones Under Different Lighting Conditions
, Yulia Kupriyanovich 1, David A. Lobb 2 and Masoud Goharrokhi 2
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis manuscript addresses an important issue in soil science: the variability of soil color measurements due to lighting conditions. Since soil color serves as an indicator of soil properties, this topic is of high relevance to the scope of Soil Systems. However, I recommend major revisions before reconsideration for publication. Below are specific comments and suggestions:
1. The authors should review previous publications that have addressed color calibration methods for soil images, and reflects on the insights from these. In particular, a recent study by Baek and colleagues has proposed similar calibration approaches using reference color plates:
- Baek, S. H., Jeon, J. S., Kwak, T. Y. (2023). Color calibration of moist soil images captured under irregular lighting conditions. Computers and Electronics in Agriculture, 214, 108299.
2. The effect of the size of the region of interest (ROI) on color measurements requires further discussion. For irregular and heterogeneous materials such as soil, color variation within the ROI can be significant. While the authors use the mean value within the ROI, in highly variable samples, the mode value may be more representative.
3. The author stated that soil-based reference materials might be preferable for calibration. However, the study does not conduct a direct comparison between manufactured color references and soil-based references. Therefore, this statement should not be interpreted as a confirmed outcome. It is therefore recommended that the authors revise the conclusion to clarify that this is a potential direction for future research, rather than a confirmed conclusion drawn from the current analysis.
4. To enhance the rigor of the validation process, the authors should consider incorporating a standard color difference (CIE, 2004), to quantify the accuracy of color correction before and after calibration. This would provide a more objective measure of the effectiveness of calibration method.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsReview
In their manuscript submitted for review, entitled "Using a reference color plate to correct smartphone-derived soil color measurements with different smartphones under different lighting conditions," the authors attempted to calibrate smartphones using a special reference color plate to test their suitability for determining soil color. Soil color, as is generally known, is an important parameter associated with various soil properties. However, in order to use this parameter to estimate, for example, humus content in soil, it must be very precisely determined. In my opinion, the authors raise an interesting issue in the manuscript concerning color and smartphone calibration to reduce errors. Furthermore, the authors explore many aspects that color depends on—for example, color measurement under different lighting conditions. In my opinion, before accepting the manuscript for further editorial review, the authors should revise the conclusions. The conclusions should summarize the research conducted and clearly address the research objective. In my opinion, some sentences included in the conclusions should not be there.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsUsing a reference color plate to correct smartphone-derived soil color
measurements with different smartphones under different lighting conditions
Journal: Soil Systems
By Li et al
At present, many people use smartphones; they can be used as important technological devices that enable the general public to collect data without much trouble. From a researcher’s perspective, obtaining accurate and reliable data from a smartphone is extremely important, as well as this approach must be efficient and cost-effective. However, this method should not put too much burden on the people using smartphones and should be easily usable. Therefore, it is crucial to have a proper way to assess the data collected by smartphones from the general public.
This study assesses the possible errors and misinterpretations that may arise when determining a soil colour from a smartphone image or photo. Three objects (a colour plate, Munsell color chips, and a soil sample) were analysed using four different smartphones under six different light conditions. The colour of an image taken from a smartphone can be varied based on the smartphone camera and the prevailing light conditions. Here, 6 colour parameters (red, blue, green, Munsell hue, chroma, and value) of a region of interest (ROI) of three objects (i.e. colour plate, Munsell chip, and soil sample) were determined using a smartphone image and a FieldSpec4 spectrometer. The data were calibrated using the linear regression model and using the colour plate as a reference.
Overall, the manuscript is well-written and arranged in a scientific way. The introduction contains necessary background information and highlights the importance of citizen science and this particular study. The research problem and objectives are clear. The Materials and Method section is also understandable and clear on what was carried out.
Results
This section has a few issues and is confusing. Firstly, this section was difficult to follow. I think the things can be expressed in a simpler manner. Secondly, there are several issues related to figures and tables.
1.) It is difficult to grasp why the x and y axes of Figures 3, 4, and 5 are FieldSpec 4 Munsell Value and SmartphoneMunsell Value, respectively. Aren’t three figures representing three different objects, i.e., a soil color plate, a Munsell chip, and a soil sample?
2.) Figure 4 can be put in the Supplementary Section, as it is a combination of data from Figures 3Ac, d, e, f and 3Bc, d, e, f?
3.) Please check whether the data in the firstTables 3 and 4 match with the corresponding table caption. The current Table 3 caption should be that of Table 4 and vice versa.
4.) Please include the other data in the supplementary section.
Discussion and conclusion are ok.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsThis study effectively solved the key error problem in soil color measurement of smart phones through multi-color reference plates. Although the absolute accuracy of soil samples is still limited by the material, the scheme significantly improves the repeatability of measurements and is of great value for promoting citizen science and rapid field monitoring. This manuscript has publication potential, but the following concerns need to be clarified.
The authors aimed to solve the problems in the application of citizen science, but the experimental design was not close enough to the actual scenarios (L145-154). In the experiment, a fixed shooting height, professional image preprocessing (zooming and rotating with GIMP software), and samples with reflections were excluded. These operations are beyond the reach of ordinary users. Even more paradoxically, while the authors claim that citizen science needs to be simplified in operation, they also require users to carry large 24-color color palettes. How could volunteers in reality carry such professional equipment with them?
The applicability of linear regression is worth considering. Figure 3 shows that the original data has a nonlinear relationship (the slope of the regression line varies significantly under different lighting conditions), but the authors insist on using linear correction, arguing that "polynomial fitting would cause excessive weights of extreme values", and this explanation lacks quantitative evidence (L 522-529). If nonlinearity is indeed the main source of error, then linear correction may instead introduce new deviations.
Some of the data in Tables 3 and 4 are elusive, and the error values after calibration are even larger than those before calibration (for example, the R value error of the Munsell color card measured by the Samsung S23 mobile phone increased from -30.09 to -34.86), but these anomalies were not fully discussed in the main text. Strangely enough, when the authors used the color plate itself as the test object, the effect was excellent, but when applied to real soil samples, the accuracy improvement was limited. However, this was simply attributed to "material differences" (L 584-595), which essentially evaded the core issue: the applicability of the color plate correction method to actual soil.
When it comes to statistical analysis methods, it is questionable to use R² as the main evaluation index. This is because an increase in R² may only indicate a decrease in dispersion (an improvement in accuracy), but it does not necessarily mean an improvement in accuracy (systematic error). In fact, the HVC parameters of the soil samples still had significant deviations after calibration (with an average error ranging from -3.73 to -0.21) (Table 3), but the authors downplayed this issue by citing that "citizen science pays more attention to relative differences".
In the Discussion Section, the authors suggested using the "stained soil color scale" (L 596-598) during field measurements, but this would introduce a new problem: How to ensure the stability of the self-made color scale? How to establish its "true value"? Paradoxically, the previous text emphasized that the method should be simplified, but here it is suggested that users fix the measurement time and bring an umbrella - these operational complexities actually weaken the practicality of the method.
Notably, the title of the paper claims to evaluate "different mobile phones and lighting conditions", but all the experimental light sources were artificially controlled (only two indoor lights + four specific outdoor scenes), without taking into account the variations in real environments such as dawn, dusk, and under the shade of trees. The four mobile phones are all high-end models (including the 200MP Samsung flagship), and the low-end models have not been tested. Such samples are not representative enough. Therefore, further clarification is needed in the Discussion Section.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThis manuscript addresses an important issue in soil science, the variability of soil color measurements under different lighting conditions. Since soil color is widely used as an indicator of various soil properties, the topic is highly relevant to the scope of Soil Systems.
Although the authors have not fully addressed the concerns I previously raised, they have made certain improvements in the revised version. Given the revisions and the overall quality of the manuscript, I consider it acceptable in its current form.
Reviewer 4 Report
Comments and Suggestions for AuthorsThe authors carefully revised the manuscript, and the revised version is now acceptable.
