Trend Analysis of Total Suspended Solids in Inland Waters Using the BFAST Algorithm on MOD09GA Products in Porto Primavera Reservoir – Brazil

Abstract

Satellite remote sensing, particularly using the MODIS orbital platform, is crucial for large-scale lake monitoring, allowing the observation of optically active components with suitable spatial and temporal resolution for lakes with surfaces greater than 40 hectares. In this context, the objective of this article was to propose a methodology that improves the retrieval and monitoring of data related to Total Suspended Solids (TSS) in large lakes. The methodology employed involved defining a baseline to map the spatiotemporal dynamics of TSS using MODIS band 1, which generates information in the infrared spectrum and is centered at 645 nm. The method was tested in the Porto Primavera Reservoir (PPR), Brazil. To validate the model, two fieldwork campaigns were conducted in the PPR, where radiometric and water quality data were collected. An empirical model was fitted between reflectance and the TSS data set (r = 0.93, R2 = 0.85, p < 0.01, n = 25). This empirical model was applied to a time series of MODIS images from May 2000 to April 2020. Using the spatial distribution maps, a time series was created from an average pixel of the sampling stations, and then this time series was analyzed to separate the trend and seasonality. The results showed that the average TSS values observed in the time series were 5.79 mg/L. The seasonality of the time series revealed that the highest concentration is recorded in the austral summer (December-February), the rainiest season. The trend component indicates that variations in TSS concentration coincide with exceptional events of increased precipitation and with a homogenization interval of the waters following the reservoir's construction.

Keywords

• bfast algorithm
• surface reflectance
• temporal series
• remote sensing
• optical active components