CROP SPECIES RECOGNITION AND DISCRIMINATION PADDY-RICE-GROWING-FIELDS FROM REAPED-FIELDS BY THE RADAR VEGETATION INDEX ( RVI ) OF ALOS-2 / PALSAR 2

The Japanese ALOS-2 satellite was launched on May 24th, 2014. It has the L-band SAR, PALSAR-2. Kim,Y. and van Zyl, J.J. proposed a kind of Radar Vegetation Index (RVI) as RVI = 8 * σ0hv / (σ0hh + σ0vv + 2*σ0hv) by L-band full-polarimetric radar data. Kim, Y. and Jackson,T.J., et al. applied the equation into rice and soybean by multi-frequency polarimetric scatterometer above 4.16 meters from the ground. Their report showed the L-band was the most promising wave length for estimating LAI and NDVI from RVI. The author tried to apply the analysis to the actual paddy field areas, both Inashiki region and Miyagi region in the eastern main island, “Honshu”, areas of Japan by ALOS-2/PALSAR-2 full-polarimetry data in the summer season, the main crop growing time, of 2015. Judging from conventional methods, it will be possible to discriminate paddy rice growing fields from reaped fields or the other crops growing fields by the PALSAR-2 data. But the RVI value is vaguely related to such land use or biomass at the present preliminary experiment. The continuous research by the additional PALSAR-2 full-polarimetry data should be desired.


INTRODUCTION
Most of the earth-observation-satellite in operating now, are the satellite equipped with SAR sensors.Japanese SAR satellites are the L-band SAR.But as its polarization is only HH polarization and its wave length is rather long, the estimation of paddy-rice-planting-area is only for practical use.
As SAR data of earth surface can be obtained in case of cloudy weather, this is an advantage of agricultural applications, such as crop species recognition or estimation of crop biomass.All-weather data collection characteristic using the L-band SAR is important for agricultural applications.

RADAR VEGETARTION INDEX (RVI)
2.1 History of RVI Kim and van Zyl (2001) proposed a kind of radar vegetation index (RVI).RVI = 4*min(λ 1, λ 2, λ 3) / (λ 1 + λ 2 + λ 3) They modified the equation as follows.( 2009) RVI = 8 * σ 0 hv / (σ 0 hh + σ 0 vv +σ 0 hv ) by L-band full-polarimetric SAR data.They applied it into rice crop and soybean.(Kim, Jackson et al., 2012) Their experiment was performed by the L-band full polarization sensor above the field collecting data every 10 minutes during the crop grown season.They compared RVI for L-, C-and Xbands to crop growth data, LAI and NDVI.They found L-band RVI was well correlated with Vegetation Water Content, LAI and NDVI.But the field data were collected by the multifrequency polarimetric scatterometer.The platform height was 4.16 meters from the ground.And they reported that the RVI is effective to estimate VWC, vegetation water content.Their VWC is useful for the estimation of soil moisture or drought, according to Kim and Jackson, et al.

Method of RVI caliculation
The ALOS-2/PALSAR-2 JAXA processing level 1.1 data and level1.5 are converted from the DN (digital number) into backscattering values by the following equation.

STUDY AREA
The field survey area was both sides along the lower course of stream of the" Tone" river in the Kanto plain of Central part of Japan.There are upland wheat fields or lotus paddy fields scattered among paddy rice fields in this area.Lotus roots are a kind of foodstuff of Japanese dishes.Geographical features are flat and somewhere its drain is bad.Therefore there are some drainage pump stations in this area.And field data were collected on August 17 th and 24 th , 2015.Its items were whether the paddy rice was still growing or already reaped, what kind of crop was, such as paddy rice, barley, wheat, lotus roots.Paddy rice is including cereals, forage-grain-rice, whole-cropsilage rice plant.(Figure2., Figure 3.) Lotus root is a kind of ingredients for Japanese cuisine.

SATELLITE DATA
The ALOS-2/PALSAR2 satellite was launched on May 24 th , 2014 and its data distribution to the public was started on Nov. 25 th , 2014.The opportunity of full-polarimetry data acquisition around Japan is limited as the following The ALOS-2/PALSAR2 data on August 25 th , 2015 were used in this study.The data processing levels were level 1.1 and level 1.5 by JAXA and RESTEC format (CEOS format).

CONVENTIONAL METHOD
The Entropy-Alpha(H-alpha) decomposition method is one of the conventional analytical method for in-coherent scattering.Its result both sides along the lower 'Tone' river course was shown in Figure 4. as follows.This calculation process was done by the polSARpro software, ver 5.04.The result image was rather rough, so its version might have some software bugs.

FINDINGS AND DISCUSSION
The Radar Vegetation Index (RVI) was promising and great expectative index for agriculture.Although the RVI result of scatter meter over the rice and soy beans in the near length were good agreement with plant's biomass in the past research paper (Kim, Jackson, 2012), the RVI from ALOS-2/PALSAR2 satellite radar was not so good.
But this is only one-year and once result.Further research should be desired.

Figure1.Figure 3 .
Figure1.Study area near Tokyo and ALOS-2 observation strip of land (red-coloured-band-image, on the Google earth)

Figure 7 .
Figure 7. RVI result at the north side of the Tone river and the overlaid-gound-survey-map

Table 1 .
table at the first observation year.ALOS-2/PALSAR2 observation plan around Japan