عنوان مقاله [English]
Frost causes a lot of damage to the agricultural sector every year.From the meteorological point of view, when the temperature drops below a certain value, frost occurs. This threshold may vary from one crop to the other. Not much research has been done to predict frost using remote sensing technology. Most of the models used to predict frost have been provided by climatologists, geographers and meteorologists based on data collected at meteorological stations.The measurements at meteorological stations are at a point and the number of these stations are limited. Therefore, depending on the surface coverage and texture around the station, the air temperature would only be valid in certain and limited distance from the stations. On the other hand, satellite images have relatively acceptable spatial resolution specially for using in the environmental studies.This indicates the necessity of using remote sensing data in many occasions including frost prediction.This work tried to predict areas at risk of frost using the NEAT method in the state of Georgia, USA. For this purpose, the MODIS satellite data and the data collected in meteorological stations of AEMN network are used.
Materials and Methods
The State of Georgia, in the southern part of the United States between latitude of 30o31’ to 35o north, and longitude of 81o to 85o53’ west with an area of 154077 square kilometers, was chosen for this case study.The reason for choosing this region was merely because of accessibility and availability of surface collected data mostly in cultivating and agricultural zones.
In this study, data collected in 10 AEMN stations from 2005 to 2015 were used for modeling and evaluation. Also, data collected in 68 stations of AEMN were used for evaluation of model for two different periods.
The satellite images used in this study is collected by Moderate Resolution Imaging Spectroradiometer (MODIS) on board of Terra and Aqua platforms. The MODIS products used in this study consist of LST (MOD11 and MYD11), lifted index (MOD07 and MYD07), total precipitable water (MOD05 and MYD05), and normalized differential vegetation index (MOD13). Also, in this study, to estimate air temperature in each 1 by 1 km grid box, the method developed by Mobashari et al. (2018) was used. The method offered an accuracy of 2.33 °C and a correlation coefficient of 0.94.
Khesali and Mobasheri, 2019 presented Near-surface Estimated Air Temperature (NEAT) model in which extrapolation coefficients for air temperature to the next hours are calculated. To increase the accuracy of the NEAT model, it was recalculated using AEMN data at Aqua and Tera passing times.
The methodology in this study consists of the following steps.
• Selection of study area and collecting temperature data from AEMN meteorological stations,
• Reproducing NEAT model coefficients usinga set of AEMN data,
• Evaluating NEAT equation using another set of AEMN data,
• Receiving and preparation of MODIS products and calculation of air temperature at the passing time of Terra and Aqua,
• Applying NEAT to the MODIS images,
• Producing Frost map using temperatures estimated by NEAT
• Evaluation of frost prediction accuracy
Results and Discussion
In order to implement the model, Two periods were selected: 3–9 December 2006 and 3–11 April 2007 in which severe crop damage across the southeastern United States has happened (Prabha and Hoogenboom, 2008).
First, the NEAT model coefficients are calculated using the AEMN network data, and evaluated for air temperature extrapolation to the next hours. Then, the air temperature was extracted using MODIS products for Aqua and Terra night time sensors. Finally, the NEAT model was applied to the air temperature extracted from satellite images, and the nighttime temperature was predicted from approximately 22:30 pm to 7:30 am of next day at 15 minute intervals. Then in the extracted images the air temperature was classified into two degreeintervals. Areas with temperatures below zero degrees Celsius are considered frost zones. Data from 68 AEMN network stations were used for evaluation. Statistical parameters like RMSE and variations of User Accuracy and Overall Accuracy were analyzed over the night. The RMSE value for all data, which is 13,840, is estimated to be 2.5 degrees. This parameter has an increasing trend from the satellite passing time to 6 hours and varies from 0.1 to 2.5 degrees Celsius. The results show the effectiveness of the proposed model in frost prediction.
In this study, AEMN meteorological data and MODIS satellite images were used for frost prediction. The study area is located in the Georgia state in the southeast of the US. Using the Neat model, air temperature is extrapolated during night in 15 minute intervals. Air temperature maps for two periods of time are produced. The results and accuracy assessment parameters show the ability of the proposed model in air temperature prediction and its effectivenessin frost prediction
9. Adams, E. C., Nyaga, J. W., Ellenburg, W. L., Limaye, A. S., Mugo, R. M., Flores Cordova, A. I., ... & Sedah, A. (2017). Designing a Frost Forecasting Service for Small Scale Tea Farmers in East Africa. In AGU Fall Meeting Abstracts.
10. Anismov O. (2001) Prediction pattern of near surface air temperature using empirical data. Climatic Change. 50:297–315.
11. Bannayan M., and Hoogenboom G. (2008). Daily weather sequence prediction realization using the non-parametric nearest-neighbor re-sampling technique. Int. J. Climatol. 28 (10), 1357-1368.
12. Bannayan M., and Hoogenboom G. (2008). Weather Analogue: A tool for lead time simulation of daily weather data based on modified K-nearest-neighbor approach. Env. Modeling and Software 23, 703-713.
13. Bhutiyani M.R., Kale V.S. and Pawar N.J. (2007). Long-term trends in maximum, minimum and mean annual air temperatures across the Northwestern Himalaya during the twentieth century. Climatic Change, 85:159–177.
14. Bolstad P.V., Swift L., Collins F. and Regniere J. (1998). Measured and predicted air temperatures at basin to regional scales in the southern Appalachian Mountains. Agricultural and Forest Meteorology, 91:161-176.
15. Didari, S., Norouzi, H., Zand-Parsa, S., & Khanbilvardi, R. (2017). Estimation of daily minimum land surface air temperature using MODIS data in southern Iran. Theoretical and applied climatology, 130(3-4), 1149-1161.
16. Hoogenboom, G., (2000). The Georgia Automated Environmental Monitoring Network. In: Preprints of the Proceedings of the 24th Conference on Agricultural and Forest Meteorology, August 14–19, 2000, Davis California, American Meteorol. Soc., Boston, MA.
17. Hoogenboom, G., (2005). The Georgia Automated Environmental Monitoring Network: experiences with the development of a state-wide automated weather station network. Proceedings of the 13th Symposium on Meteorological Observations and Instrumentation & 15th Conference on Applied Climatology, American Society of Meteorology, Boston, MA (Abstracts) (www.ametsoc.org/meet/index.html).
18. Kalma J.D., Laughlin G.P., Caprio J.M., and Hamer P.J.C. 1992. Advances in Bioclimatology, 2. The Bio climatology of Frost. Berlin: Springer-Verlag.144p.
19. Khesali, E., Mobasheri, M.R,. (2018), Air Temperature image production at different times from a satellite pass time using weather stations data. 13th Symposium on Advances in Science and Technology.
20. Kotikot, S.M. and Onywere, S.M., (2015). Application of GIS and remote sensing techniques in frost risk mapping for mitigating agricultural losses in the Aberdare ecosystem, Kenya. Geocarto International, 30(1), pp.104-121.
21. Kotikot, S. M., Flores, A., Griffin, R. E., Sedah, A., Nyaga, J., Mugo, R., ... & Irwin, D. E. (2018). Mapping threats to agriculture in East Africa: Performance of MODIS derived LST for frost identification in Kenya’s tea plantations. International journal of applied earth observation and geoinformation, 72, 131-139.
22. Lomme J.P., and Guilioni L. 2004. A simple model for minimum crop temperature forecasting during nocturnal cooling. Agricultural and Forest Meteorology, 123 (1-2): 55-68.
23. Mobasheri, M.R, Khesali, E., Ghorbani, Alireza. (2018), An Attempt in Presenting a Model for Temperature Determination at Near Land Surface Using MODIS Images. 13th Symposium on Advances in Science and Technology.
24. Pouteau R., Rambal S., Ratte J.P., Goge F., Jore R., Winlel T. (2011). Downscaling MODIS-derived maps using GIS and boosted regression trees: the case of frost occurrence over the Arid Andean highlands of Bolivia. Remote Sensing of Environment, 115 (1): 117-129.
25. Prabha, T., & Hoogenboom, G. (2008). Evaluation of the Weather Research and Forecasting model for two frost events. Computers and electronics in agriculture, 64(2), 234-247.
26. Simões, D. D. S., Fontana, D. C., & Vicari, M. B. (2015). Use of LST images from MODIS/AQUA sensor as an indication of frost occurrence in RS. Revista Brasileira de Engenharia Agrícola e Ambiental, 19(10), 920-925.
27. Snyder R.L., and Paulo de Melo-Abreu J. )2005). Frost Protection: fundamental, practice and Economics, Vol. 1, FAO.
28. Tait, A. and Zheng, X., (2003). Mapping frost occurrence using satellite data.Journal of Applied Meteorology, 42(2), pp.193-203.