Korean Journal of Remote Sensing, Vol.28, No.5, 2012, pp.509~520 http://dx.doi.org/10.7780/kjrs.2012.28.5.4 Estimating Rice Yield Using MODIS NDVI and Meteorological Data in Korea Suk Young Hong*, Jina Hur**, Joong-Bae Ahn**, Jee-Min Lee***, Byoung-Keol Min***, Chung-Kuen Lee****, Yihyun Kim*, Kyung Do Lee*, Sun-Hwa Kim*, Gun Yeob Kim* and Kyo Moon Shim* * National Academy of Agricultural Science (NAAS), RDA, ** Division of Earth Environmental System, Pusan National University *** B&T Solutions, **** National Institutel of Crop Science (NICS), RDA Abstract : The objective of this study was to estimate rice yield in Korea using satellite and meteorological data such as sunshine hours or solar radiation, and rainfall. Terra and Aqua MODIS (The MOderate Resolution Imaging Spectroradiometer) products; MOD13 and MYD13 for NDVI and EVI, MOD15 and MYD15 for LAI, respectively from a NASA web site were used. Relations of NDVI, EVI, and LAI obtained in July and August from 2000 to 2011 with rice yield were investigated to find informative days for rice yield estimation. Weather data of rainfall and sunshine hours (climate data 1) or solar radiation (climate data 2) were selected to correlate rice yield. Aqua NDVI at DOY 233 was chosen to represent maximum vegetative growth of rice canopy. Sunshine hours and solar radiation during rice ripening stage were selected to represent climate condition. Multiple regression based on MODIS NDVI and sunshine hours or solar radiation were conducted to estimate rice yields in Korea. The results showed rice yield of 494.6 kg 10a -1 and 509.7 kg 10a -1 in 2011, respectively and the difference from statistics were 1.1 kg 10a -1 and 14.1 kg 10a -1, respectively. Rice yield distributions from 2002 to 2011 were presented to show spatial variability in the country. Key Words : Rice yield, Remote sensing, MODIS NDVI, Solar radiation 509
Korean Journal of Remote Sensing, Vol.28, No.5, 2012 510
Estimating Rice Yield Using MODIS NDVI and Meteorological Data in Korea 511
Korean Journal of Remote Sensing, Vol.28, No.5, 2012 Fig. 1. Flow chart of MODIS data preprocessing. (NIR NDVI = _ RED) (-1 NDVI 1) (1) (NIR + RED) 512
Estimating Rice Yield Using MODIS NDVI and Meteorological Data in Korea Fig. 2. Parcel map based paddy mask(90 m). 513
Korean Journal of Remote Sensing, Vol.28, No.5, 2012 Fig. 3. Rice yield prediction schema using MODIS NDVI and climate data. Table 1. Yield statistics of Korea (Statistics Korea, http://kosis.kr) Year Unhulled rice Milled rice Year Yield(kg 10a -1 ) 2000 675 497 2001 693 516 2002 639 471 2003 609 441 2004 679 504 2005 661 490 2006 664 491 2007 630 466 2008 694 520 2009 706 534 2010 653 483 514
Estimating Rice Yield Using MODIS NDVI and Meteorological Data in Korea Table 2. Correlation coefficient between rice yield and MODIS products MODIS products Correlation Coefficient MODIS products 00~ 11 00~ 11 02~ 11 02~ 11 TNDVI209 (7.28) -0.25-0.33 TNDVI225 (8.13) -0.04-0.04 TEVI209 (7.28) -0.27-0.40 TEVI225 (8.13) -0.38-0.43 Terra TLAI201 (7.20) -0.05 0.01 TLAI209 (7.28) -0.15-0.24 TLAI217 (8.5) 0.23 0.30 TLAI225 (8.13) -0.05-0.11 TLAI233 (8.21) 0.41 0.33 TLAI241 (8.28) 0.63 0.62 ANDVI201 (7.20) -0.25 ANDVI217 (8.5) 0.28 ANDVI233 (8.21) 0.62 AEVI201 (7.20) 0.50 AEVI217 (8.5) 0.18 Aqua AEVI233 (8.21) -0.23 ALAI201 (7.20) -0.17 ALAI209 (7.28) 0.14 ALAI217 (8.5) 0.26 ALAI225 (8.13) 0.41 ALAI233 (8.21) 0.30 ALAI241 (8.28) 0.73 rainfall (1 ) -0.71-0.83 sunshine hours (1) 0.63 0.70 rainfall (2 )* -0.64-0.79 solar radiation (2) 0.73 0.86 A: Aqua, T: Terra, 1, 2: Meteorological data 1, 2, * 00-10(n=11) or 02-10(n=9) 515
Korean Journal of Remote Sensing, Vol.28, No.5, 2012 Table 3. Rice yield prediction model based on Aqua NDVI 233 (x 1 ) and sunshine hours 1(x 2 ) or rainfall 1(x 2 ) of ripening stage( 02~ 10, n=9) Yield Prediction Model 2011 Rice Yield predicted(kg/10a) Rice Yield = (NDVI 233 (x 1 ), sunshine hours 1(x 2 )) P=0.0274 RSq=0.70 * RMSE=17.9 kg 10a -1 494.6 Rice Yield = (NDVI 233 (x 1 ), rainfall 1(x 2 )) P=0.0054 RSq=0.82 * RMSE=13.7 kg 10a -1 515.5 Table 4. Rice yield prediction model based on Aqua NDVI 233 (x 1 ) and insolation 2(x 2 ) or rainfall 2(x 2 ) of ripening stage( 02~ 10, n=9) Yield Prediction Model 2011 Rice Yield predicted(kg/10a) Rice Yield = (NDVI 233 (x 1 ), solar radiation 2(x 2 )) P=0.0103 RSq=0.78 * RMSE=15.2 kg 10a -1 509.7 Rice Yield = (NDVI 233 (x 1 ), rainfall 2(x 2 )) P=0.0078 RSq=0.80 * RMSE=14.5 kg 10a -1 515.0 (a) Fig. 4. Comparision of predicted rice yield based on MODIS NDVI and sunshine hours 1 (a) and MODIS NDVI and solar radiation 2 (b) and rice yield statistics from 2002 to 2011. (b) 516
Estimating Rice Yield Using MODIS NDVI and Meteorological Data in Korea 2011 2010 2009 Fig. 5. Rice yield map of Korea from 2002 to 2011. 2008 517
Korean Journal of Remote Sensing, Vol.28, No.5, 2012 2007 2006 2005 2004 Fig. 5. Continued 2003 2002 518
Estimating Rice Yield Using MODIS NDVI and Meteorological Data in Korea Ahn, J.B., C.K. Park, and E.S. Im, 2002. Reproduction of regional scale surface air temperature by estimating systematic bias of mesoscale numerical model, Journal of Korean Meteorological Society, 38(1): 69-80. Becker-Reshef, I., E. Vermote, M. Lineman, and C. Justice, 2010. A generalized regression-based model for forecasting winter wheat yields in Kansas and Ukraine using MODIS data, Remote Sensing of Environment, 114: 1312-1323. Chang, J., M.C. Hansen, K. Pittman, M. Carroll, and C. DiMiceli, 2007. Corn and soybean mapping in the United States using MODIS time-series data sets, Agronomy Journal, 99: 1654-1664. Cock, J.H. and S. Yoshida, 1972. Accumulation of 519
Korean Journal of Remote Sensing, Vol.28, No.5, 2012 14 C-labelled carbohydrate before flowering and its subsequent redistribution and respiration in the rice plant, Proceedings of the Crop Science Society of Japan, 41: 226-234. Cressman, G.P., 1959: An operational objective analysis system. Mon. Wea. Rev., 87, 367-374. Doraiswamy, P.C, T.R. Sinclair, S. Hollinger, B. Akhmedov, A. Stern. and J. Prueger, 2005. Application of MODIS derived parameters for regional crop yield assessment, Remote Sensing of Environment, 97: 192-202. FAO (GIEW), http://www. fao.org/giews/english/index.htm Hong, S.Y., J.T. Lee, S.K. Rim, and J.S. Shin, 1997. Radiometric estimates of grain yields related to crop aboveground net production (ANP) in paddy rice. Proc. of 1997 International Geoscience and Remote Sensing Symposium, Singapore, Aug. 3-8, pp. 1793-1795. Jordan, C.F., 1969. Derivation of leaf area index from quality of light on the forest floor, Ecology, 50: 663-666. Myneni, R.B., G. Asrar, and F.G. Hall, 1992. A three dimensional radiative transfer modelsalgorithms-experiments. Remote Sensing of Environment, 51: 3-26. Narongrit, C. and K. Chankao, 2009. Development and validation of rice evapotranspiration model based on Terra/MODIS remotely sensed data, Journal of Food, Agriculture & Environment, 7: 684-689. Rasmussen M.S., 1997. Operational yield forecast using AVHRR NDVI data: reduction of environmental and inter-annual variability, International Journal of Remote Sensing, 18(5): 1059-1077. Rouse, J.W, R.H. Haas, J.A. Schell, and D.W. Deering, 1973. Monitoring vegetation systems in the great plains with ETRA. In third ETRS Symposium, NASA SP-353. U.S. Govt. Printing Office, Washington D.C., 1: 309-317. Ren, J., Z. Chen, Q. Zhou, and H. Tang, 2008. Regional yield estimation for winter wheat with MODIS-NDVI data in Shandong, China, International Journal of Applied Earth Observation and Geoinformation, 10: 403-413. Sun, J., 2000. Dynamic monitoring and yield estimation of crops by mainly using the remote sensing technique in China, Photogrammetric Engineering and Remote Sensing, 66(5): 645-650. 520