Regressive Analysis on Leaf Nitrogen Content and Near Infrared Reflectance and Its Application for Agricultural Farm Monitoring with Helicopter Mounted Near Infrared Camera

Method for evaluation of nitrogen richness of tealeaves with near infrared reflectance is proposed. Also tea farm monitoring with helicopter mounted near infrared camera is proposed. Through experiments and regressive analysis, it is found that the proposed method and monitoring system is validated. Keywords-regressive analysis; nitrogen content; tealeaves; near infrared cameras; t


INTRODUCTION
Vitality monitoring of vegetation is attempted with photographic cameras [1].Grow rate monitoring is also attempted with spectral reflectance measurements [2].Total nitrogen content corresponds to amid acid which is highly correlated to Theanine: 2-Amino-4-(ethylcarbamoyl) butyric acid so that total nitrogen can be used for a measure of the quality of tealeaves.
It is well known that Theanine rich tealeaves taste good while fiber content in tealeaves is highly correlated to the grow rate of tealeaves.Both total nitrogen and fiber content in tealeaves are highly correlated to the reflectance in the visible and Near Infrared: NIR wavelength regions and vegetation index derived from visible and NIR data so that it is possible to determine most appropriate tealeaf harvest date using the total nitrogen and fiber content in the tealeaves which are monitored with ground based visible and NIR cameras and with Helicopter mounted NIR cameras.
It is obvious that nitrogen rich tealeaves tastes good while fiber rich tealeaves tastes bad.Theanine: 2-Amino-4-(ethylcarbamoyl) butyric acid that is highly correlated to nitrogen contents in new tealeaves are changed to catechin [3], [4], [5] due to sun light.In accordance with sun light, new tealeaves grow up so that there is a most appropriate time for harvest in order to maximize amount and taste of new tealeaves simultaneously.
Regressive analysis is conducted with measured total nitrogen and NIR reflectance measured with ground based spectral radiometer and helicopter mounted NIR camera data.Both total nitrogen contents and NIR reflectance derived from ground based and helicopter mounted NIR camera shows a good coincidence.Therefore, it is concluded that the proposed method and monitoring system is validated and appropriate.
The following section describes the proposed method and monitoring system for estimation of total nitrogen content and followed by some experiments.Then conclusion is described together with some discussions.

A. Regressive Analysis
Linear regressive equation is expressed in equation (1).
where N, R denotes measured Nitrogen content in leaves, and measured Near Infrared: NIR reflectance, respectively while a and b denotes regressive coefficients.There is well known relation between nitrogen content and NIR reflectance.Therefore, regressive analysis based on equation (1) is appropriate.

B. Proposed Method for Tealeaves Quality Evaluation of
Tealeaves quality can be represented Theanine content which is closely related to nitrogen content.Furthermore, it is well known that nitrogen content can be represented with NIR reflectance.Therefore, tealeaves quality can be evaluated with measured NIR reflectance based on the equation (1).
The proposed method and tea farm area monitoring system with helicopter mounted NIR camera is based on the aforementioned scientific background.www.ijarai.thesai.org

C. Tea Farm Area Monitoring System with Helicopter Mounted NIR Cameras
The helicopter used for the proposed system is "GrassHOPPER" manufactured by Information & Science Techno-Systems Co. Ltd.Outlook of helicopter is shown in Figure 1.Table 1 shows major specification of GrassHOPPER.Canon Powershot S100(focal length=24mm)is mounted on the GrassHOPPER.It allows acquire images with the following Instantaneous Field of View: IFOV at the certain altitudes, 1.1cm (Altitude=30m)3.3cm(Altitude=100m)and 5.5cm(Altitude=150m).
In order to measure NIR reflectance, standard plaque whose reflectance is known is required.Spectralon provided by Labsphere Co. Ltd. is well known as well qualified standard plaque.It is not so cheap that photo print papers are used for the proposed system.Therefore, comparative study is needed between Spectralon and the photo print papers.
The proposed system consist Helicopter, NIR camera, photo print paper.Namely, photo print paper is put on the agricultural plantations, tea trees in this case.Then farm areas are observed with helicopter mounted NIR camera.Nitrogen content in agricultural plants, tealeaves in this case, is estimated with NIR reflectance.

III. EXPERIMENTS A. Preliminary Experiments
There are some candidates of NIR Cameras, Ricoh GX200, Cannon PowerShot S100, ADC3 which allows not only NIR image but also Normalized Deviation of Vegetation Index: NDVI iamge, etc.Before acquiring NIR reflectance data with helicopter mounted NIR cameras, ground surface images are acquired with the candidate ground based cameras.Figure 2(a) and (b) shows examples of the acquired images with Ricoh GX200 and ADC3, respectively.Although Ricoh GX200 is just a visible camera, NIR image is also acquired with the camera if NIR filter is attached to the optical entrance of the camera.On the other hand, ADC3 has NIR channel.Therefore, NIR image is acquired as shown in Figure 2 (b).
From the acquired images of Figure 2, a small portion of image is extracted and enlarged as shown in the bottom image of Figure 2 (a) and (b).In the portion of image, there is name board of tealeaves.Due to the fact that standard plaque is required for estimation of NIR reflectance, and Field of View of the Cameras is limited, not so small number of standard plaques is required for the wide areas of tea farm area and agricultural farm areas.Therefore, relatively cheap photo print papers are used as standard plaque.Through a comparison between well qualified standard plaque of Spectralon and the proposed photo print papers, usability of the photo print papers is confirmed as a standard plaque.Outlook of the tea farm area and Spectralon of standard plaque and photo print paper which are put on the tea trees Fig. 7.
Rectified three strips of GrassHOPPER tranck.
As the results, it would be better to put photo print papers in all strips of GrassHOPPER traks.Then NIR reflectance can be estimated with a reference to the pixel value of the photo print paper in the image in the corresponding strip.

B. Comparison Between GrassHOPPER Mounted ADC3 Acquired Reflectance and Measured Reflectance with Ground Based MS-720 of Spectral Radiometer
A comparison between GrassHOPPER mounted ADC3 acquired reflectance and measured reflectance with ground based MS-720 of spectral radiometer manufactured by Eiko Co. Ltd., Japan is conducted.MS-720 is reliable radiometer with 5nm of wavelength resolution.This experiment is conducted at Saga Prefectural Tea Experiment Station of Ureshino City, Saga Japan, on April 9 2012.By comparing between the reflected radiance from the Spectralon and the reflected radiance from the tealeaves, the reflectance can be measure.On the other hand, GrassHOPPER mounted ADC3 derived reflectance is estimated by comparing the pixel values of the Spectralon image portion and that of the tealeaves of image portion.Table 2 shows the average and the www.ijarai.thesai.orgThe NIR reflectance (which is proportional to total nitrogen content and Theanine of amino acid in the tealeaves) is increasing in accordance with tealeaves grow as shown in Table 3 and Figure 9. Horizontal and vertical axis of Figure 9 shows different species and NIR reflectance, respectively.On the other hand, Blue, Red, and Green colored bar denotes, NIR reflectance which is measured on April 9, 10, and 23, respectively.Most of species of tea trees are grown up during this period and are harvested in the begging of May.
Total nitrogen content of the tealeaves is measured based on Kjeldahl method (a kind of chemistry method).The measured total nitrogen content is compared to the NIR reflectance.Figure 10 shows the relation between the measured total nitrogen content and the measured NIR reflectance.Both show a good coincidence.
As aforementioned, the pixels for standard plaque in the helicopter strip image are required for estimation of NIR reflectance.FOV of the helicopter mounted NIR cameras are limited.Therefore, not so small number of helicopter strips which is corresponding to the number of standard plaques are required for cover the entire tea farm areas.Then relatively cheap photo print paper is used on behalf of Spectralon of standard plaque.Therefore, it would be better to compare the mean and standard deviation of measured NIR reflectance between standard plaque and photo print paper.Table 4 shows the comparison between both.The mean of the NIR reflectance which is measured with photo print paper is a little bit greater than that with standard plaque while the standard deviation of the NIR reflectance which is measured with photo print paper is a little bit greater than that with standard plaque.During this period, measured NIR reflectance is increasing in accordance with the time duration.Black solid lines show linear approximated lines for the different species of tealeaves.Variability of the NIR reflectance may be caused by the weather condition (Illumination, waving tealeaves due to the wind, influence due to shadow and shade, etc.).

IV. CONCLUSION
Method for evaluation of nitrogen richness of tealeaves with near infrared reflectance is proposed.Also tea farm monitoring with helicopter mounted near infrared camera is proposed.Through experiments and regressive analysis, it is found that the proposed method and monitoring system is validated.
In particular, relation between measured NIR reflectance and total nitrogen content is clarified.Also it is found that photo print paper can be used as the reference target for estimation of NIR reflectance.

Fig
Fig. 1.Outlook Of The Grasshopper TABLE I. MAJOR SPECIFICATION OF GRASSHOPPER weight 2kg(Helicopter only)

Fig. 2 .
Fig. 2.Examples of the acquired images with Ricoh GX200 and ADC3The name board of Figure2(a) is much clear than that of Figure2 (b).Tea farm area is designated at Prefectural Tea Institute which is situated in Ureshino city, Saga, Japan.Preliminary tests are conducted for confirmation of the relation between helicopter altitude and IFOV with GrassHOPPER and ADC3.An example of acquired image is shown in Figure3.Also, Figure4shows examples of images which are acquired with GrassHOPPER mounted ADC3 camera with the different altitudes.

Figure 5
Figure 5 shows example of acquired image with GrassHOPPER mounted ADC3 camera.Standard plaque has to be acquired within a range of Field of View of the camera.Green rectangle shows four shots of acquired image with the camera (ADC3) from the 50 m of altitude.

Fig. 5 .
Fig. 5.Example of the acquired image with GrassHOPPER mounted ADC3 camera.

Figure 6
Figure 6 (a) shows outlook of the tea farm area while Figure 6 (b) shows Spectralon of standard plaque and photo print paper (Both are aligned together closely) which are put on the tea trees.Also Figure 7 shows rectified three strips of GrassHOPPER tranck.In the first strip, standard plaque and photo print paper can be seen in the acquired image.
Fig. 6.Outlook of the tea farm area and Spectralon of standard plaque and photo print paper which are put on the tea trees GrassHOPPER mounted ADC3 derived reflectance and ground based MS-720 derived reflectance.Both show a good coincidence between mean values as shown in Figure8.The horizontal axis of Figure8is GrassHOPPER mounted ADC3 derived reflectance while the vertical axis is ground based MS-720 derived reflectance.

Fig. 11 .
Fig. 11.Trend of NIR reflectance for the period starting from April 9 to April 23.