Estimation Method of Flood Damage to Rice Production

Estimation Method of Flood Damage to Rice Production

Katsuo Okamoto, Shuji Yamakawa, and Hiroyuki Kawashima
3-1-1 Kannondai, Tsukuba, Ibaraki 305 JAPAN
Tel: (81) -298-38-8171 Fax: (81)-298-38-8199
E-mail: Kokamoto@niaes.affrc.go.jp

Abstract
It was reported that North Korea (Democratic People's Republic of Korea) suffered from heavy flooding due to unprecedented rainfall from the middle of July to the beginning of August 1995. many rivers reportedly flooded and rice plants in paddy fields were washed away by deluges, although details have not been clarified outside North Korea. Through use of landsat Thematic Mapper (TM) data, the damage to paddy fields in the western part of North Korea was investigated. It was estimated that at least 2.5 x 105 ha of paddy fields, which corresponded to 42% of the total area of paddy fields in North Korea, were damaged by floods. The total rice crop in the western part of North Korea in 1995 was 2.21 x 105 metric tons, which was approximately 5.30 x 105 tons less than the total rice crops recorded in the 1980s.

Introduction
It was reported that global warming frequently gives rise to abnormal weather (Japan Meteorological Agency, 1994). As such weather causes considerable damage to agriculture, it is important to evaluate the damage quickly and accurately in order to secure food supply. Remote sensing techniques can collect information over large areas or in places inaccessible on the ground. Remote sensing provides the best means to estimate the damage resulting from a disaster because its accumulated findings are available.

As North Korea suffered extreme events since 1990, their food has been in short supply since 1991 (The Nihon Keizai Shimbun, 1995). The western part of North Korea suffered from unprecedented rainfall from July to August 1995 (The Asahi Newspaper, 1995; USDA, 1996b). Major granaries in Chagangdo Province, Pyonganpukto Province, Pyongannamdo Province and Hwanghaepukto Province suffered from flooding (The Asahi Newspaper, 1996b; USDA 1996c). However, details haven't been reported outside North Korea.

In this study, we estimated the total area and the 1995 flood damaged area of paddy fields in the western part of North Korea. Using the relation between the rice phenology and spectral characteristics, we estimated the difference between the rice crop in a normal year and in 1995.

Fig 1 Test site

Materials and Methods
We used the Landsat TM data (Path-Row : 117-32 through 34) acquired on 15 October 1992 (in the harvesting season) and 27 May 1993 (in the rice-planting season) to estimate the area of paddy fields. That is, candidates for paddy fields were extracted from the result of land-cover classification for data obtained during the rice-planting season and then open water, which were extracted from the data obtained during the harvesting season, were separated from candidates for paddy fields (Okamoto and Fukuhara, 1996)

We also used the TM data (Path-Row: 117-32 through 34) acquired on 16 September 1987, 22 September 1996 to estimate the rice crop. These data covered the western part of North Korea (Fig. 1) Since most of the paddy fields affected by the flooding are in this region (The Asahi Newspaper, 1996b; USDA, 1996c), we can estimate the flood damage from these data.
The difference in the total paddy field area in North Korea between 1993 and 1995 is considered to be negligible because the North Korean Government has no reason to decrease because paddy fields cannot be developed to any great extent in two years.

Table 1 Landsat TM calibration constants for conversion of digital values to radiances in W m-2sr-1 mm-1 (ERSDAC,1990).

Band	LMIN	LMAX
TM3	-1.20	204.35
TM4	-1.50	206.20

Conversion from the digital values (DN) of TM products to at-satellite spectral radiances (L) in W m-2sr-1 mm-1 was accomplished with the following equations:
L= Gain*DN +Bias
Gain= (LMAX-LMIN)/255,
Bias = LMIN,

Where LMAX is saturation radiance and LMIN threshold one (Table 1). We assumed that the atmospheric conditions on 16 September 1987, 22 September 1995 and 8 September 1996 were the same because data acquisition were within the same two-week period in each of the three years and because the weather at the test site was clear. We corrected for the effects of solar elevation

The spectral track of rice plants observed in japan is shows in Fig 2. paddy field classes just after rice- planting a lokared around the pl stang in a near infrared (TM4)- red (Tm3) future space plot. As rice plants grow normally, paddy field classes move toward the Hv stage along the spectral growth track (SGT) and each stage (?) on the SGT lasts about 10.

Fig.2 Locations of paddy field classes in a near infrared (TM4)-red (TM3) feature space plot from the Landsat TM data acquired on 22 September 1995, and spectral growth track of rice based on data reported by Shibayama and Akiyama (1989)

**Table 2** 1995 growth stages of rice paddies in North Korea estimated from those in northeastern Japan
Abbreviation	Period stage	Cumulative temperature (^oC days)*	Phenological
P1	24 May to 2Jun.**	0	rice-planting season
MTN	12 Jul. to21 Jul.	930	maximum tiller number
Hd	10 Aug. to 19 Aug.	1600	heading time
LGF	4 Sep. to 13 Sep.	2100	linear grain filling
Hv	29 Sep. to 8 Oct.	2600	harvest time

*Values are quoted from the Research Team on Cool Summer Damage (1996).
**Source: The Asahi Newspaper 91996a).

Days (shibayama and Akiyama, 1989). If rice plants grow abnormally due to a cool summer or other adverse conditions, a paddy field class may take longer to move between stages on the SGT or may deviate from the SGT.

The period for rice cultivation on North Korea was considered to be the same as that in northeastern Japan (Table 2). According to Yamagata and Akiyama (1998), a TM4 value of around 67W m-2sr-1 mm-1 (84 in DN) corresponds to a crop of 3.0 metric tons ha-1, while a value of around 43 W m-2sr-1 mm-1 ((55 in DN) corresponded to be in straight head and the field is considered to have yielded no crop (Shiga and Asaka, 1994). Therefore we estimated that paddy field classes before the LGF stage (TM3<19 m-2sr-1 mm-1) on 22 September 1995 were expected to be no crop, since the harvest had to be finished by around 20 October in North Korea. Applying these relations, we divided the flood damage into 4 classes (Fig.2), i.e. no flood damage (N), slight (S), moderate (M) and heavy (H).

Result and Discussion
The area of paddy fields in the western part of North Korea was estimated to be 2.76*105 ha, with an estimation error of less than 5% (Okamoto and Fukuhara, 1996). The total area of paddy fields in North Korea reported by the FAO (1996) is 6.50*105 ha.

We produced a flood damage map of the paddy fields in the western part of North Korea in 1995. The paddy fields damaged by flooding were distributed around the confluences and the lower reaches rivers. The rice production in the areas (1.88*104 ha) around Sojoson-Man by and 50 km south of Pyongyang was estimated to be 6.39*104metric tons in 1987,1.67*104metric tons in 1995, and6.54*104metric tons in 1996. The result indicate that the 1995 flood damage in these areas was heavy but was recovered by 1996.

**Table 3** Extent of paddy fields damaged by floods and rice yield production in the western part of North Korea.
Damage	yield (metric tons ha^-1)	Area (ha^-1)	Area(%)	production (metric tons)
None or good harvest	>3	0.926x10³	0.3	3.65 x10³
Slight	>2 to 3	12.8x10³	4.6	31.9 x10³
Moderate	>1 to 2	75.2 x10³	27.3	113 x10³
Heavy	>0 to 1	145 x10³	52.5	72.3 x10³
Heavy (no crop)	0	42 x10³	15.3	0
Total		2.76 x10⁵	100	2.21 x10⁵

We estimated the area of flood damage in paddy fields (Table 3). We estimated that the paddy fields in the western part of North Korea yielded 7.51 x105 metric tons of rice in 1987 (a normal year). The corresponding figure for 1995 was 2.21 x105 metric tons. According to our estimation, an average yield of rice in the western part of North Korea was 2.73 metric tons ha-1 in in 1987: however, that was 0.80 metric tons ha-1in 1995. Assuming that the eastern part of North Korea was not affected by the flooding, 3.74 x105 has of paddy fields were identified there [=6.5 x105ha (the total area)- 2.76 x105 ha (the western part)] -2.76 x105 ha (the western part)], which yielded an average unit of 2.73 metric tons ha-1, the same as that in 1987; the total rice crop in North Korea in 1995 was at most 1.24 x106 metric tons. The rice crop in 1995 estimated by the USDA was 1.30 x106metric tons (USDA, 1996a). Our estimation tended to show that the crop damage due to flooding in 1995 was more serious than tha estimated by the USDA.

Conclusion
It is estimated that at least 2.75 x105 ha of paddy fields, which corresponds to 42% of the total area of paddy fields in North Korea, were affected by the flooding. The total rice crop in the western part of North Korea in 1995 was 2.21 x105 metric tons, which is approximately 5.30 x105 metric tons less than the total rice crops estimated for 1987 (a normal year). Assuming that the eastern part of North Korea was not affected by the flooding, the total rice crop in North Korea in 1995 was at most 1.24 x106 metric tons. Our estimation shows that the crop damage due to flooding in 1995 was serious.

Acknowledgement
The US government retains ownership of the Landsat TM data. EOSAT/NASDA supports us in acquiring the Landsat TM data at marginal cost.

Reference

ERSDAC (Earth Resources Satellite Data Analysis Center), 1990, Image data processing and analysis (2). (Tokyo: ERSDAC), 76p. (in Japanese)
FAO, 1996, FAOSTAT Database Results, http://apps.fao.org/. (Rome: The Food and Agriculture Organization of the United Nations).
Japan Meterological Agency (editor), 1994, Circumstances of Extreme Events and Climatic Variation. In The Report '94 of Extreme Events, (Tokyo: Printing Bureau, Ministry of Finance), 57-129. (in Japanese)
Okamoto, K., and Fukuhara, M., 1996, Estimation of paddy field area using the area ratio of categories in each mixel of Landsat TM., International Journal of Remote Sensing, 17, 1735-1749.
Research Team on Cool Summer Damage, 1996, Data regarding the phonological state. http://ss.tnaes.affrc.go.jp/~kanda/seiiku.html..(Morioka, Japan: The Tohoku Agriculture Experimental Station). (in Japanese)
Shibayama, M., and Akiyama, T., 1989, Seasonal visible, near-infrared and mid-infrared spectra of rice canopies in relation to LAI and above-ground dry phytomass. Remote Sensing of Envioronment, 27,119-127.
Sigha, H., and Asaka, D., 1994, The use of MOS-1 MESSR data for estimation of rice yielddamaged by cold weather in 1993. Journal of the Remote Sensing Society of Japan, 14,54-60. (in Japanese with English abstract)
The Asahi Newspaper, 1995,25 August 1995(Tokyo: The Asahi Shimbun). (in Japanese)
The Asahi Newspaper, 1996a, 17June 1996. (Tokyo: The Asahi Shimbun).(in Japanese)
The Asahi Newspaper, 1996b, 3 August 1996. (Tokyo: The Asahi Shimbun).(in Japanese)
The Nihon Keizai Shimbun, 1995, 17 August 1995. (Tokyo: The Asahi Shimbun).(in Japanese)
USDA, 1996a, Korea Home Page (data). http://ffas.usda.gov/ffas/fas-programs-services-resources/fas-programs/fas-commodity/remote-sensing /korea/kordata.htmal. (Washington: United States Department of Agriculture)
USDA, 1996b, Korea Home page (weather). http://ffas.usda.gov/ffas/fas-programs-services-resouces/fas-programs/fas-commodity/remote-sensing/korea/korweather.Html. (Washington : United States Department of Agriculture)
USDA, 1996C, Korea Home Page (data). http://ffas.usda.gov/ffas/fas-programs-services-resources/fas-programs/fas-commodity/remote-sensing/nkorfood.giɦ. (Washington: United States Department of Agriculture)
Yamagata, Y., and Akiyama, T., 1988, Flood damage analysis using multitemporal Landsat Thematic Mapper data. International Journal of Remote Sensing, 9,503-514.

Page 1 of 1