At AgroCares we are often confronted with the question regarding accuracy and comparison with wet chemistry Laboratories.
Especially, the Handheld Scanner is positioned as a monitoring tool and therefore cannot be viewed as a replacement of wet chemistry methods. However, it is important to AgroCares that the quality standard of the analysis provides agronomic relevant results to the users.
Even though AgroCares is following regular quality assurance measures and participating in the Wepal proficiency scheme (the results can be fund on the website or requested separately from your account manager) as the very first sensor based analysis provider in often cases our customers wish to perform their own validation study and compare with local laboratories.
For a good and representative comparison study (for Scanner and Liab) it is important to consider sample taking procedure, sample preparation, repetition, comparable methods and statistical analysis of the results. The suggested guidelines are summarized in the below article.
Still, contacting your account manager or firstname.lastname@example.org before staring is the best way to move forward in order to adapt the comparison to your final objective of the study and guide you thought the process.
It should always be considered that the Handheld Scanner is a monitoring tool, and it is not a substitute for analytical laboratory techniques. Therefore, we suggest to perform comparisons based on classification rather direct comparison of the actual values.
Overall, it is to be expected that sensor based solutions will not perform at the same level as wet chemistry laboratories, simply because the predictions models for sensors are build on the reference data of wet chemistry laboratories. Thus, the error of the certified reference laboratory used for calibration is in addition to the error of the instrument itself.
AgroCares sensor technology has been calibrated at AgroCares laboratory GSL (Golden Standard Laboratory). GSL is one of the most accurate laboratories participating in on the global WEPAL ring test, while the Handheld Scanner and Liab are certainly competitive in accuracy with most of the participating labs. You may request WEPAL results here. WEPAL is the most independent validation procedure that can be provided. Performing a comparison in a laboratory which does not participate in ring test trials is not encouraged. As seen in the results from this ring test, the variation between wet chemistry laboratories is quite substantial.
Such a comparison to a single laboratory can not be viewed as representative, ideally several certified laboratories with comparable methods and sample preparation procedures are selected for comparisons.
Key points to consider
1. Soil sample. The reliability of a soil test is only as good as the sample you submit.
- Timing: sample collection should be performed before liming and application of fertilizers/manures (or minimum of 6 weeks after)
- Sampling strategy depending on the overall goal of the study: to analyze within field variation or take a representative field different strategies will have to be followed. Depth of the soil sample should be from 0 to 30 cm, as this is the depth the Scanner was calibrated for.
- Soil sample storage: fresh soil should always be stored in a fridge, scan the soil with less then 24h after soil collection and send the soil as soon as possible to the laboratory to avoid changes in chemical properties.
2. Scanning process. Following the scanning steps carefully and according to protocol is very important in order to obtain representative results of the soil sample.
- White calibration. Before each order, a white-background calibration needs to be performed with the calibration cap. Check if the sensor head and glass plate are clean from dirt and moist. Make sure the pins of the Scanner connect perfectly with the holes of the calibration cap. The Scanner head needs to be in contact with the white surface.
- Yellow calibration. At the beginning of each working day or when you connect your app to another device, a yellow standard sample needs to be calibrated with the yellow side of the calibration cap. Please follow the same steps as for the white calibration.
- Soil Sample Scanning. After calibration, the app will direct you to soil sample scanning. To scan your sample, please follow these steps:
- Mix the sample well in a bucket and take for each scan a full scoop of material (about 250gr of soil).
- Place the Scanner on top of the sample. Make sure the scanning surface of the sensor head is in contact with the sample.
- Press the ‘Scan’ button for 1 second to start the scanning process. Do not move the Scanner while it is scanning or while the scan (green) light is blinking.
- The mobile app will notify you when you can continue to the next scan. Three scans per sample are required.
- Clean the sensor head before proceeding with the second scan.
- Take for each scan a new full scoop of soil from the collected soil sample in the bucket, place the Scanner on it and press the Scan button.
- Repeat the process until all 3 required valid scans are finalized. During each scan, make sure the scanning surface of the sensor head is in contact with the sample, that there are no air bubbles and the light from the Scanner is not visible. Clean the sensor head between each scan, clean it with the brush and/or towel provided, you may also use paper towels.
For more detailed explanation please see this instruction video or read the document "Getting Started with AgroCares Scanner" send to your email after purchasing the Scanner (you may request this document at email@example.com or to your account manager).
3. Number of soil samples. Please consider including in your study multiple soil samples in order to have a better overview on the results. It is recommended to at least include five soil samples and repetitions. Each soil samples should be analyzed in multiple replicates to be statistically relevant.
4. Number of replicates. Each soil sample should ideally be summited three times to each laboratory, and produce at least three reports with the Scanner. This is required for statistical analysis. To detect outliers at least five replicates are necessary. Eventually, it is important to assess if the laboratory is meeting their quality standards. Good quality laboratories are accredited and follow certified ISO procedures and methods. Additionally, ideally the laboratories provide information about errors and bias of their analysis methods together with the analysis results. To avoid bias, the soil sample should be submitted without identifying labelling, ideally a numerical or alphabetical order is used for that which can be later on linked to the original barcodes.
5. Number of laboratories. As it is highlighted above, significant differences can be found between laboratories. Therefore, we strongly advise you to select at least three laboratories in order to assess the variability in the results between them and conduct a fair comparison. Each replicate send to each lab will be scanned with the Scanner (ideally three reports) in order to have the exact Scanner result of each specific samples sent to the laboratories. AgroCares is encouraging to search for laboratories with good quality standards, participate in ring test trials like WEPAL.
The weight of the soil sample collected must be estimated considering the amount of soil that each lab needs to receive. For instance, if each lab requires 0.5kg of fresh soil, three replicates per laboratory and three laboratories assessed. 0.5kg*3*3 = 4.5kg of fresh soil collected (very well mixed) per soil sample. We strongly advise you to keep at least 0.5 kg of soil stored per soil sample in case of further analysis will be required or any of the results coming either from a lab or the Scanner are questioned. Please dry the samples at 40 degrees Celsius and store it in an airtight bag with correct labeling in a cool and dry space.
6. Pool of nutrients & Methods. The Scanner provides results for specific nutrients, pools of nutrients and methods. For instance, for Nitrogen we provide results of the total pool, of Phosphorous we provide results for the available pool and for Potassium for the exchangeable pool. Making sure the lab is analyzing the same pool of nutrients is the first step, otherwise the results will not be comparable. Once the correct pool of nutrients has been match between the Scanner and the laboratory, it is crucial to check the methodology used by the laboratory.
The Scanner has been calibrated at GSL following strict protocols and methods for each parameter following ISO norms. Reviewing the methods reported by the Scanner (hence, GSL) and the methods available in the laboratory is key in order to choose which parameters can be compared and should be requested to the laboratory. For instance, AgroCares reference method for Total Nitrogen is dry combustion (Dumas), but other laboratories might instead use Kjeldahl method. The correlation between the two methods is satisfactory, however Dumas provides on average greater result than Kjeldahl method. This is important information to consider once you are choosing the lab method and further for data analysis.
Information on the methods used to calibrate the Scanner can be found here. Additionally, scientific literature can be found online that explain which parameters are comparable, which not and what are the expected differences. At AgroCares we are happy to guide you thought the process and share our experience, please contact your account manager or firstname.lastname@example.org and make sure that your comparison is scientifically correct.