Assessment of the Efficiency of Using Modis MCD43A4 in Mapping of Rice Planting Calendar in the Mekong Delta

Abstract

This study used 487 MODIS MCD43A4 images, 1-day resolution, 500-meter spatial resolution to evaluate the efficiency of using the MODIS image series in monitoring the rice planting calendar in the Mekong Delta in 2019. The Savitzky-Golay filter method was applied to eliminate noise and smooth the time series of NDVI imagery. The rice plating date was then determined using TIMESAT software based on the threshold of increasing NDVI to 20% of the amplitude each season. The reliability of the model for mapping rice planting calendar was evaluated by compared to field data and local statistics. The results showed that the series of MODIS MCD43A4 images was used effectively in map of rice planting calendar for the Winter-Spring season, with area error was less than 10% and different of sowing date was less than 15 days (N = 21, MAE = 15 days, RMSE = 18 days). However, the series of MCD43A4 images was limited in monitoring the rice crop calendar of the other two remaining seasons (Summer-Autumn and Autumn-Winter) with an area error was greater than 30% and a higher difference in the time of sowing compared to the Winter-Spring season, especially in the Autumn-Winter season (N = 20, MAE = 22 days, RMSE = 23 days). The above results showed that the MODIS image series was effectively used in the dry season (Winter-Spring) and limited in the rainy season due to the influence of weather and clouds. Further studies combining the use of other types of satellite images or methods to replace cloud-affected pixels needed to be taken into consideration.