Phenology Detection of Forest Area Using Satellite Data

Phenology Detection of Forest Area Using Satellite Data

Asako NAGAO, Koji KAJIWARA, Yoshiaki HONDA
Center for Environmental Remote Sensing, Chiba University
1-33 Yayoi-cho, Inage-ku, Chiba city, Chiba, 263-8522
Tel: (81)-43-290-3845 Fax: (81)-43-290-3845
Email: nagao@ceres.cr.chiba-u.ac.jp
JAPAN

Keywords: Forest, Remote Sensing, Phenology, Reflectance, NDVI

Abstruct:
Detecting Phenology of forest area using satellite data is effective way for the global environmental problems. We have established an observation tower in a forest area and been gotten ground truth data. By using these data, we decide the active period of forest area, and we could check satellite data how accurate data can detect this period.

1. Introduction
Global environmental problems, for example, greenhouse effect, Extinction of tropical rain forest, desertification have relevance to forest resource deeply and it is more and more necessary to monitor the condition of environment with a big scale and real time by satellite data. This method makes it possible to estimate vegetation amount of forest area but active period of forest area depends on the period of photosynthesis and we need to specify the time when photosynthesis is going active.

2 Objectives
To specify the period of forest is in active, it is important to detect the time when forest starts and finish their activities throughout the year. After all we need to know a germination period and the period when leaves fall and compare it with satellite data.

3 Measurement System of Ground Truth Data
For ground truth data, we have established an observation tower in Yatsugatake forest area, Kobuchizawa, Yamanashi Pref., Japan.(Fig.1) On the tower, we have set up spectrometer, CCD camera which catch the image of same place of spectrometer, and camera.(Fig.2)


Fig.1 Measurement Site		Fig.2 Measurement System of Observation Tower

The observation tower is about 25m tall and the camera and the spectrometer has been set at top of the tower. The camera takes photo once a day, and spectrometer and CCD camera take data since 6:00 until 18:00 every 15 minute. Measurement system of spectrometer is as follows (Fig.3).

Fig.3 Measurement System of Spectrometer on the Tower
4 Ground Truth Data

4.1 CCD camera image and camera image
CCD camera data and camera data shows as follows(Fig.4). From these data, we can specify the condition of increasing of vegetation.

	2000/04/25	2000/05/05	2000/05/09
CCD camera image

camera image

Fig.4 CCD camera image and Camera image
4.2 Spectrometer Data
Spectrometer is used to know the reflectance of leaves. Leaves show its reflectance highly at NIR, so we can know the time when leaves come out from spectrometer. By using this data, we have checked the value of NDVI. NDVI shows the amount of vagetation and it would be showed as follows:

NDVI = (NIR - RED) / (NIR + RED) (2.1)
This width of RED and NIR is depending on the sensor of satellite, but this time we use these values based on NOAA / AVHRR data. That is:

RED: 580-680nm NIR : 725-850nm
This figure(Fig.5) shows the change of NDVI at Yatsugatake forest area in April 1^st until May 15^th this year.

Fig.5 Change of NDVI of spectrometer at Yatsugatake

4.3 Satellite Data
For satellite data, we used NOAA/AVHRR data. NOAA/AVHRR data have 1km/pixel resolution and we used it of Yatsugatake Area for two months(Fig. 6).

2000/04/25	2000/05/05	2000/05/09

Fig.6 Satellite(NOAA/AVHRR) data
4.4 Result of analysis
For image data (CCD camera image and camera image), to extract the amount of vegetation, we used irvep program(Ichikawa, 2000). And for spectrometer data and satellite data, we calculated the values of NDVI for two months. The result of vegetation amount by each of the image data(CCD camera and camera) show as follows(Fig.7.1), and the result of NDVI(spectrometer data and NOAA/AVHRR data) shows as follows(Fig.7.2) :

Fig.7.1 Changes of Vegetation amount of CCD camera, camera and NDVI of spectrometer data

Fig7.2 NDVI of NOAA/AVHRR data
5. Conclution and Problems for the Future
Looking at the graph(Fig. 7.1), we can know the difference of recognition period by each of measurement system(camera, CCD camera, spectrometer). And the value of NDVI of NOAA/AVHRR is not so big compare with ground truth data(Fig.7.2). It is to say, NOAA/AVHRR data doesn't show accurately and it is clear that there is a gap between satellite data and ground truth data.

We can guess the reason why satellite data didn't show bigger value than ground truth data is because of the vapor and dust in the air over Japan. And these problems make it more difficult to specify phenology of forest area.

And, the data which we have used this time was only two months, so we have to check the past data and continue our research with a long period after this, and know the difference according to the year.

Acknowledgement
This work has been supported by CREST (Core Research for Evolutional Science and Technology) of Japan Science and Technology Corporation (JST).