Time series analysis of water surface temperature and heat flux components in the Itumbiara Reservoir (GO), Brazil

• Autor(a) principal: Alcântara,Enner Herenio de
• Data de Publicação: 2011
• Outros Autores: Stech,José Luiz, Lorenzzetti,João Antônio, Novo,Evlyn Márcia Leão de Moraes
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Acta Limnologica Brasiliensia (Online)
• Texto Completo: http://old.scielo.br/scielo.php?script=sci_arttext&pid=S2179-975X2011000300004
• topic: thermal remote sensing; time series; heat flux; physical limnology; MODIS imagery

AIM: Water temperature plays an important role in ecological functioning and in controlling the biogeochemical processes of the aquatic system. Conventional water quality monitoring is expensive and time consuming. It is particularly challenging for large water bodies. Conversely, remote sensing can be considered a powerful tool to assess important properties of aquatic systems because it provides synoptic and frequent data acquisition over large areas. The objective of this study was to analyze time series of surface water temperature and heat flux to advance the understanding of temporal variations in a hydroelectric reservoir. METHOD: MODIS water-surface temperature (WST) level 2, 1 km nominal resolution data (MOD11L2, version 5) were used. All available clear-sky MODIS/Terra images from 2003 to 2008 were used, resulting in a total of 786 daytime and 473 nighttime images. Time series of surface water temperature was obtained computing the monthly mean in a 3×3 window of three reservoir selected sites: 1) near the dam, 2) at the centre of the reservoir and 3) in the confluence of the rivers. In-situ meteorological data from 2003 to 2008 were used to calculate surface energy budget time series. Cross-wavelet, coherence and phase analysis were carried out to compute the correlation between daytime and nighttime surface water temperatures and the computed heat fluxes. RESULTS: The monthly mean of the day-time WST shows lager variability than the night-time WST. All time series (daytime and nighttime) have a cyclical pattern, passing for a minimum (June - July) and a maximum (December and January). Fourier and the Wavelet Analysis were applied to analyze this cyclical pattern. The daytime time series, presents peaks in 4.5, 6 12 and 36 months and the nighttime WST shows the highest spectral density at 12, 6, 3 and 2 months. The multiple regression analysis shows that for daytime WST, the heat flux terms explain 89% of the annual variation (RMS = 0.89 °C, p < 0.0013). For nighttime, the heat flux terms explain 94% (RMS = 0.53 °C, p < 0.0002). CONCLUSION: The daytime WST and shortwave radiation presents a good agreement for periods of 6 (with shortwave retarded) and 12 months (with shortwave advanced); For nighttime WST and longwave the good agreement is present for 1, 3, 6 and 12 months, all with longwave advanced in relation to WST.