Ana Catarina Rebelo, up201806239 Mariana Messias, up201806148 Rui Matias, up201806812
Abstract
Improving the quality and yeld of our agriculture is a must with the increasing population worldwide. One of the best ways to improve the intensive farming is by adding the exact amount of nutrients to the soil. However, the technologies available to measure the concentration of nutrients on the soil are usually too expensive to the common farmers. In this paper we will study the viability of a developed NIR-VIS spectrophotometer that presents a much reduced cost when compared to the solutions available on the market by studying the fiability of the data collected and the sensibility of the device.
Introduction
Why is our solution important?
Continuously growing population brings with it a low of problems. One of them is how to feed all that population. Having enough food to feed a continuously growing human population is one of the biggest concerns about the future. Intensive farming is a must to solve that problem. For that, the best efficiency on growing plants is desired. Intensive farming relies on continuously production without leaving any time for the natural regeneration of the soils. Because of constant use of the soils to achieve the biggest amount of food, soils are not able to naturally produce enough nutrients to feed humanity demands. So, addition of fertilizers is needed.
The most important nutrients for a good development of the plants are:
N, nitrogen, that is the biggest component of plant’s amino acids and chlorophyll, playing an important role in plant growth,
P, phosphorou, that is part of plants DNA and RNA, being part of seeds and roots formation.
K, potassium, that is important because it activates a big part of the enzymes needed to allow the growth of a healthy plant [1][2]
A small concentration of these nutrients on soil will lead to a reduced growth and lower yields while too high concentrations will also affect plant growth and yield and reduce plant quality[3] . So, it is important to reach a good concentration of NPK on the soil and, for that, it is crucial to have a good, cheap, and fast way to measure it, preferentially portable and in situ, allowing a proper dosage of fertilisers.
What is our solution?
As referred previously, solving the problem of being able to quantify the concentration of macronutrients present in the slurry and in the soil is a serious concern nowadays, which lead to multiple fabricants develop various solutions with different specialisations for the problem.
Multiple solutions from different fabricants caused to necessity to create a norm to normalize the method and format of how data is transferred between sensors, actuators, control elements, and how they are displayed and stored, regardless of the tractor or implement utilized by the agriculturist. The norm created is designated ISO 11783 and defines that data communications should be made by a single cable with four pairs of twisted wire and nine-pin terminal connector, commercially called the ISOBUS connector[4][5].
In this project, it is studied a solution for the problem, the AS7265x chipset, which would then be inserted in a tractor/cistern system to analyse the soil. The results obtained for this device will then be compared to an acquired device, USB2.0 Mini-Spectrometer C11697 MB, already on the market. Both the devices utilise NIR-VIS spectroscopy.
What is the chemistry involved?
The analysis methodologies used in this study, were ultraviolet-visible and near infrared spectrophotometry, which investigates the interaction of light radiation with physical matter in a wavelength range from 200 to 2500 nm. [6]
When light travels from one medium to another, in this case from air to soil, a part of the this light radiation can be transmitted through the medium, a part will be absorbed and a part will be reflected at the interface between the two media. Since the container in which the sample is placed is opaque, light is not transmitted, so we can consider this term to be negligible. [7]
The higher the concentration of the sample, the more radiation is absorbed by the sample, so the reflectivity decreases. Since the spectrophotometer used, relates reflectivity to wavelength, it is expected that the signal will decrease with increasing concentration. [7]
Procedure
FIgure 3: Representation of the procedure
Table 1-3: Added mass of each fertilizer and nutrient, for set 1,2 and 3
Table 4-7: Added mass of each fertilizer and nutrient, for set 4,5,6 and 7
Results
After measuring the sample’s signals with both spectrophotometers, the results obtained, were analyzed using spectra that relate reflectivity with wavelength.
Figures 4 and 5 show the spectra obtained from the 3 measurements of the 51 samples. It is relevant to note two recurring peaks in all samples that occur at a wavelength of, approximately, 440 nm and 560 nm.
Furthermore, the fact that there is a variation in the height of the peaks in the various samples, allow us to conclude that the spectrophotometers are sensitive to these nutrients in the soil, which means, they are able to detect them.
Figure 4 - Signal for all samples, USB2.0 Mini-Spectrometer
Figure 5 - Signal for all samples, developed spectrophotometer
Figure 6. Comparison of the spectrum of the control sample with the spectra of sectors 1-7, Developed spectrophotometer
To assess correctly that reflectivity decreases with the increase of concentration, one graph with spectra of three samples for each sector obtained with the developed spectrophotometer , was analysed.
The spectra shows a comparison between samples in which nutrients were added to the control sample, that is, the sample that contained only soil (the blue line). As expected, the reflectivity in the soil is higher than in samples concentrated with nutrients.
When analyzing the graphs for sector 3, in which calcium nitrate is added, it can be seen that it does not follow the same bias, since the value of the sample with the highest concentration of calcium nitrate has the highest peak. In the graph of sector 4, where two nutrients are mixed, there are overlapping lines, which does not follow the bias. This difference can be justified by the fact that, at these wavelengths, there is a lot of background related noise. For sector 5 the lines are also very close by, but for sector 6 and 7 the obtained spectra follows the bias, which means, that the signal decreases with the concentration.
The same procedure was performed for 3 samples from each sector in the USB2.0 Mini-Spectrometer, apparently following the same bias
Figure 7. Comparison of the spectrum of the control sample with the spectra of sectors 1-7, USB2.0 Mini-Spectrometer
Figure 8 - Pattern deviation vs wavelength
Analyzing the figure, it is possible to affirm that the obtained signal varies the most for wavelengths of 460 nm, around 560 nm and 610 nm. This observation leads to the conclusion that the developed device is more sensible to the variation of the nutrients’ concentration at these values of wavelength.
Furthermore, one can conclude that the device is more sensible to the variation of NK (KNO3) as the largest pattern deviation occurs for set 2. The worst response occurs for the mixture of MPK and NK, as set 5 presents the lowest pattern deviation.
Finally, it is possible to conclude that the device also presents a good signal response to the mixture of MPK (KH2PO4) and NCa (Ca(NO3)2) and the mixture of the three compounds as the signal for set 6 and 7 also varies significantly.
Conclusions
Sensibility:
We found that the device is able to measure concentrations above 0,249 mg(N)/g(soil), 0,493 mg(P)/ g (soil) and 0,382 mg(K)/g(soil). However, this data only represents the smallest ammount of added nutrients and does not have in concern the already existant nutrients.
We should do more samples at lower concentrations to see at what point the device does not detect any presence of added nutrients.
Nonetheless, this results are positive, specially when we compare them with the best average composition of a soil: 14mg (N)/g (soil), 1,86mg (P)/g(soil) and 10mg (K)/g (soil)[8].
Relation between concentration and the measured signal:
An increased concentration of nutrients in the sample causes a smaller pick on the signal, as expected.
This analysis does not allow us to find any mathematical relationship between the signal measured and the concentration of the soil. For that, we need an extra step. We propose the use of a machine learning program to treat all the data, aiming for a linear regression between the measured signal and the concentration of nutrients on the soil.
Additional Material
For further information, we present a video in which we explain the problematic, all the procedures, the results obtained and what we concluded with this project. It is also available an article with more detailed information about the procedure, the sensors that were used and the data analysis.
References
[1]Potdar, R. P.; Shirolkar, M. M.; Verma, A. J.; More, P. S.; Kulkarni, A., Determination of soil nutrients (NPK) using optical methods: a mini review. Journal of Plant Nutrition 2021, 1-14.
[2]Varsha A. Shukre , Supriya S. Patil, 2020, Comparative Study of Different Methodologies used for Measuring Soil Parameters: A Review, INTERNATIONAL JOURNAL OF ENGINEERING RESEARCH & TECHNOLOGY (IJERT) ICSITS – 2020 (Volume 8 – Issue 05).
[3]agrocares NPK: What is it and why is it so important? https://www.agrocares.com/2020/11/02/npk-what-is-it-and-why-is-it-so-important/ (accessed 18/02/2021).
[4]A. W Layton, A. D Balmos, S. Sabpisal, A. Ault, J. V Krogmeier, and D. Buckmaster, "ISOBlue: An Open Source Project to Bring Agricultural Machinery Data into the Cloud," presented at the 2014 Montreal, Quebec Canada July 13 – July 16, 2014, St. Joseph, MI, 2014. [Online]. Available: https://elibrary.asabe.org/abstract.asp?aid=45014&t=5.
[5] M. l. E. GmbH. "O padrão ISO11783." https://www.muellerelektronik.de/pt-br/isobus/die-norm/ (accessed 18-Mar-2021).
[6] C. E. Adeeyinwo, Okorie, N. N. , Idowu, G. O., "Basic Calibration of UV/ Visible Spectrophotometer," International Journal of Science and Technology, vol. 2, 2013.
[7] Callister, Jr William D. 1940-; Materials science and engineering
[8]Shah, D. R., and E. K. M. Pawar. 2009. Laboratory testing procedure for soil and water sample analysis, eds. W. R. D. D. P. Irrigation Research and Development. SSD/GL/01 (02): 1-134