Evaluation of Time Series Trend of Total Phosphorus Concentration Distribution in Mekong River Floodplain Using Satellite Images

Cambodia is located in the downstream area of the Mekong River. Large flood in the Mekong River occurs annually and generally and forms the floodplain widely. In this floodplain, abundant nutrients become benefit for agriculture production. In recent years it has been necessary to evaluate the impact of flood and rapid urbanization on agriculture. Therefore, we focused on the evaluation of the impact of flood on agriculture using satellite images which can capture high temporal and spatial resolution. Thus the purpose of this study is to create distribution maps of nutrients in time series. This study selected phosphorus since it has considered as the limiting factor in the Mekong River floodplain in Cambodia. However various phosphorus types in water is difficult to be directly estimated by satellite images. Recently, the chlorophyll-a concentration which has been estimated from remote sensing and correlated with the phosphorus concentration can be used for evaluating distribution of phosphorus. The estimated total phosphorus concentration was calculated using the equation of Dillon et al (1974). Previous study reported that estimation of chlorophyll-a concentration using MODIS data can be estimated with high accuracy in the open ocean. However, applying this model has not been done to floodplains of rivers yet, and this means the application can estimate it with high accuracy is not known. This study was done by applying the MODIS data to the model of conventional chlorophyll a concentration and resulted in underestimation of the concentration obtained by field observation at some points. The underestimation is probably due to the influence of suspended solid. Therefore a correction model was built for the underestimation. In order to create the correction model, a regression analysis was performed on the error using the standardized MODIS band data. As a result, the correlation coefficient increased more than the original model. The correction model obtained was applied to MODIS composite data for one year, and the spatiotemporal distribution of total phosphorus concentration was estimated. From this concentration distribution map, we can understand the seasonal difference of water area and concentration distribution.