Fig. 1. A map of the weather station sites where observation data were obtained. These sites consist of 34, 169 and 27 stations included in the synoptic observation system (SYNOP_SKOR), the meteorological information service system for disaster prevention (MISS_DP) in South Korea and the synoptic observation network in North Korea (SYNOP_NKOR), respectively.
Fig. 2. Nassi-schneider diagram of MODIS-PIPET that calculates mean temperature using MOD07 level 2 product. RF represents a random forest model to estimate air temperature.
Fig. 3. The proportion of cells where no missing data for the MODIS products were available by (a) month and (b) weather station. The weather station consists of the synoptic observation system (SYNOP_SKOR), the meteorological information service system for disaster prevention (MISS_DP) in South Korea and the synoptic observation network in North Korea (SYNOP_NKOR). MOD, MYD and MCD indicate Terra, Aqua and both Terra and Aqua respectively.
Fig. 4. The training results of random forest models to estimate daily mean temperature (Tmean) using surface air temperature product from (a) Terra (MOD), (b) Aqua (MYD), and (c) both Terra and Aqua (MCD) satellites, respectively. Ground measurements of temperature (OBS) were obtained from 34 synoptic observation stations in South Korea. N indicates the total number of training data that had no missing value for the weather stations during the period of 2009-2018.
Fig. 6. Probability density of (a) coefficient of determination (R2) and (b) root mean square error (RMSE) of mean temperature estimated from random forest models. R2 and RMSE were determined by weather station included in the meteorological information system for disaster prevention (MISS-DP) network in four provinces in South Korea. MOD, MYD and MCD indicate the random forest models of which training data were obtained from Terra, Aqua, and both Terra and Aqua satellites, respectively.
Fig. 7. Probability density of (a) coefficient of determination (R2) and (b) root mean square error (RMSE) for daily cooling degree-days (CDD). The CDD values were calculated using the mean temperature estimated from random forest models. R2 and RMSE were determined by weather station included in the meteorological information system for disaster prevention (MISS-DP) network in four provinces in South Korea. MOD, MYD and MCD indicate the random forest models of which training data were obtained from Terra, Aqua, and both Terra and Aqua satellites, respectively.
Fig. 8. Probability density of (a) coefficient of determination (R2) and (b) root mean square error (RMSE) for daily cooling degree-days estimated using random forest models. R2 and RMSE were determined by weather station included in the meteorological information system for disaster prevention (MISS-DP) network in four provinces in South Korea. MOD, MYD and MCD indicate the random forest models of which training data were obtained from Terra, Aqua, and both Terra and Aqua satellites, respectively.
Fig. 9. Comparison of cooling degree days (CDD) obtained from weather stations and satellite products in North Korea. The daily values of CDD were compared only for the days when no missing occurred for the satellite data. The daily CDD values were estimated using the random forest model for mean temperature (TMEANRF; a, c, and e) and CDD (CDDRF; b, d, and f), respectively. The random forest models were trained using data from Terra (a and b), Aqua (c and d) and both Terra and Aqua (e and f) satellites.
Fig. 5. The training results of random forest models to estimate daily cooling degree (CDD) using surface air temperature product from (a) Terra (MOD), (b) Aqua (MYD), and (c) both Terra and Aqua (MCD) satellites, respectively. Ground measurements of temperature (OBS) were obtained from 34 synoptic observation stations in South Korea. N indicates the total number of training data that had no missing value for the weather stations during the period of 2009-2018.
Fig. 10. Comparison of average cooling degree days (CDD) obtained from weather stations and satellite products in North Korea. The daily values of CDD were summed only for the days when no missing occurred for the satellite data. These cumulative CDD values for each year were averaged for all of the weather stations available in North Korea. The daily CDD values were estimated using the random forest model for mean temperature (a, c, and e) and CDD (b, d, and f), respectively. The random forest models were trained using data from Terra (a and b), Aqua (c and d) and both Terra and Aqua (e and f) satellites.
Table 1. Options for Random Forest (RF) models and variable importance of RF models
References
- Ban, H.-Y., J.-B. Ahn, and B.-W. Lee, 2019: Assimilating MODIS data-derived minimum input data set and water stress factors into CERES-Maize model improves regional corn yield predictions. Plos One 14(2), e0211874. https://doi.org/10.1371/journal.pone.0211874
- Breiman, L., 2001: Random forests. Machine learning 45(1), 5-32. https://doi.org/10.1023/A:1010933404324
- Busetto, L., and L. Ranghetti, 2016: MODIStsp : An R package for automatic preprocessing of MODIS Land Products time series. Computers & Geosciences 97, 40-48. https://doi.org/10.1016/j.cageo.2016.08.020
- Famiglietti, C. A., J. B. Fisher, G. Halverson, and E. E. Borbas, 2018: Global validation of MODIS near‐surface air and dew point temperatures. Geophysical Research Letters 45(15), 7772-7780. https://doi.org/10.1029/2018GL077813
- Gao, F., T. Hilker, X. Zhu, M. Anderson, J. Masek, P. Wang, and Y. Yang, 2015: Fusing landsat and MODIS data for vegetation monitoring. IEEE Geoscience and Remote Sensing Magazine 3(3), 47-60. https://doi.org/10.1109/MGRS.2015.2434351
- Hong, S. Y., S.-I. Na, K.-D. Lee, Y.-S. Kim, and S.-C. Baek, 2015: A study on estimating rice yield in DPRK using MODIS NDVI and rainfall data. Korean Journal of Remote Sensing 31(5), 441-448. https://doi.org/10.7780/kjrs.2015.31.5.8
- Huang, R., C. Zhang, J. Huang, D. Zhu, L. Wang, and J. Liu, 2015: Mapping of daily mean air temperature in agricultural regions using daytime and nighttime land surface temperatures derived from TERRA and AQUA MODIS data. Remote Sensing 7(7), 8728-8756. https://doi.org/10.3390/rs70708728
- Jang, K., S. Kang, J. Kimball, and S. Hong, 2014: Retrievals of all-weather daily air temperature using MODIS and AMSR-E data. Remote Sensing 6(9), 8387-8404. https://doi.org/10.3390/rs6098387
- Lee, B., K. Jang, E. Kim, M. Kang, J.-H. Chun, and J.-H. Lim, 2019: Predicting forest gross primary production using machine learning algorithms. Korean Journal of Agricultural and Forest Meteorology 21(1), 29-41. https://doi.org/10.5532/KJAFM.2019.21.1.29
- Lee, J.-H., S.-K. Kang, K.-C. Jang, J.-H. Ko, and S.-Y. Hong, 2011: The evaluation of meteorological inputs retrieved from MODIS for estimation of gross primary productivity in the US corn belt region. Korean Journal of Remote Sensing 27(4), 481-494. https://doi.org/10.7780/kjrs.2011.27.4.481
- Lee, K.-D., S.-I. Na, S.-Y. Hong, C.-W. Park, K.-H. So, and J.-M. Park, 2017: Estimating sorn and soybean yield using MODIS NDVI and meteorological data in Illinois and Iowa, USA. KOREAN Journal of Remote Sensing 33(5), 741-750.
- Lee, K. J., S. Lee, B. W. Lee, and K. S. Kim, 2013: Implementation of GrADS and R scripts for processing future climate data to produce agricultural climate information. Atmosphere 23(2), 237-243. https://doi.org/10.14191/Atmos.2013.23.2.237
- Liaw, A., and M. Wiener, 2002: Classification and regression by randomForest. R news 2(3), 18-22.
- Luus, K. A., R. Commane, N. C. Parazoo, J. Benmergui, E. S. Euskirchen, C. Frankenberg, J. Joiner, J. Lindaas, C. E. Miller, W. C. Oechel, D. Zona, S. Wofsy, and J. C. Lin, 2017: Tundra photosynthesis captured by satellite‐observed solarinduced chlorophyll fluorescence. Geophysical Research Letters 44(3), 1564-1573. https://doi.org/10.1002/2016GL070842
- Neteler, M., 2010: Estimating daily land surface temperatures in mountainous environments by reconstructed MODIS LST data. Remote sensing 2(1), 333-351. https://doi.org/10.3390/rs1020333
- Prasad, A. K., L. Chai, R. P. Singh, and M. Kafatos, 2006: Crop yield estimation model for Iowa using remote sensing and surface parameters. International Journal of Applied Earth Observation and Geoinformation 8(1), 26-33. https://doi.org/10.1016/j.jag.2005.06.002
- Ramirez-Villegas, J., A. Jarvis, and P. Laderach, 2013: Empirical approaches for assessing impacts of climate change on agriculture: The EcoCrop model and a case study with grain sorghum. Agricultural and Forest Meteorology 170, 67-78. https://doi.org/10.1016/j.agrformet.2011.09.005
- Turner, D. P., W. D. Ritts, W. B. Cohen, S. T. Gower, S. W. Running, M. Zhao, M. H. Costa, A. A. Kirschbaum, J. M. Ham, S. R. Saleska, and D. E. Ahl, 2006: Evaluation of MODIS NPP and GPP products across multiple biomes. Remote Sensing of Environment 102(3-4), 282-292. https://doi.org/10.1016/j.rse.2006.02.017
- Wei, S., C. Yi, W. Fang, and G. Hendrey, 2017: A global study of GPP focusing on light‐use efficiency in a random forest regression model. Ecosphere 8(5).
- Wilmers, C. C., B. Nickel, C. M. Bryce, J. A. Smith, R. E. Wheat, and V. Yovovich, 2015: The golden age of bio‐logging: how animal‐borne sensors are advancing the frontiers of ecology. Ecology 96(7), 1741-1753. https://doi.org/10.1890/14-1401.1
- Yoo, B. H., and K. S. Kim, 2017: Development of a gridded climate data tool for the COordinated Regional climate Downscaling EXperiment data. Computers and Electronics in Agriculture 133, 128-140. https://doi.org/10.1016/j.compag.2016.12.001
- Yoo, B. H., K. S. Kim, and J. Lee, 2019: Comparison of the wall clock time for extracting remote sensing data in Hierarchical Data Format using Geospatial Data Abstraction Library by operating system and compiler. Korean Journal of Agricultural and Forest Meteorology 21(1), 65-73. https://doi.org/10.5532/KJAFM.2019.21.1.65