Homogenised daily lake surface water temperature data generated from multiple satellite sensors: A long-term case study of a large sub-Alpine lake

Abstract

Availability of remotely sensed multi-spectral images since the 1980’s, which cover three decades of voluminous data could help researchers to study the changing dynamics of bio-physical characteristics of land and water. In this study, we introduce a new methodology to develop homogenised Lake Surface Water Temperature (LSWT) from multiple polar orbiting satellites. Precisely, we developed homogenised 1 km daily LSWT maps covering the last 30 years (1986 to 2015) combining data from 13 satellites. We used a split-window technique to derive LSWT from brightness temperatures and a modified diurnal temperature cycle model to homogenise data which were acquired between 8:00 to 17:00 UTC. Gaps in the temporal LSWT data due to the presence of clouds were filled by applying Harmonic ANalysis of Time Series (HANTS). The satellite derived LSWT maps were validated based on long-term monthly in-situ bulk temperature measurements in Lake Garda, the largest lake in Italy. We found the satellite derived homogenised LSWT being significantly correlated to in-situ data. The new LSWT time series showed a significant annual rate of increase of 0.020 °C yr−1 (*P < 0.05), and of 0.036 °C yr−1 (***P < 0.001) during summer.

Irregular changes in lake surface water temperature and ice cover in subalpine Lake Lunz, Austria

“…For example, when compared with Lake Baikal, for which a long-term average water temperature profile (since 1946) is available, the LSWT data of Lake Lunz differ again considerably; the surface water temperature in Lake Baikal has increased by 1.21 °C since 1946 (Hampton et al 2008), whereas the LSWT in Lake Lunz has actually decreased by 0.18 °C during the same period. When comparing LSWT of Lake Lunz with those of Lake Garda, Italy, investigated between 1986 and 2015 (Pareeth et al 2016), the LSWT of small Lake Lunz (0.046 °C/yr) increased more than twice as much as the LSWT of Lake Garda (0.020 °C/yr), which is much larger (368 km 2 ) and duration of full ice cover since 1921. The overall increase in LSWT between 1921 and 2015 (0.8 °C within 94 years) does not parallel other Austrian subalpine lakes investigated for a similar time period (1910-1990Livingstone and Dokulil 2001).…”

“…deeper (300-350 m maximum depth; Pareeth et al 2016). These differences in LSWT clearly suggest that lake location, morphometry, size, and regional climate strongly effect how LSWT develop over the same time period.…”

A Long-Term Fine-Resolution Record of AVHRR Surface Temperatures for the Laurentian Great Lakes

“…These gaps were typically over areas of the lakes that have experienced persistent cloudiness but sometimes occur because the data were not available for that day from the NOAA CLASS archive. Some previous studies have aimed to estimate LSWT under cloudy skies using purely time-series based interpolation techniques, like local regression [2,43] or harmonic analysis [15]. There are some cases where spatial information can improve the interpolated estimates from the time-series approaches.…”

“…While long-term in situ records of water temperature are ideal for detecting such changes in many cases, these records are not available for most of the world's lakes and are usually significantly lacking in spatial coverage of any one lake. Long-term satellite infrared observations have the ability to overcome this data limitation, as illustrated in several recent regional and global studies [1,2,[13][14][15][16].…”

The Relationship between Urban Land Surface Material Fractions and Brightness Temperature Based on MESMA

“…According to Planck's law, assuming that the content of vapor is relatively small and constant in the atmosphere (the radiation change affected by atmospheric conditions can only be ignored when there exists a homogeneous atmospheric environment), the at-satellite brightness temperature (in Kelvin) reflects the surficial temperature distribution condition on land [6,30]. Some researchers have proved that the at-satellite brightness temperature can be used to approximate the land surface brightness temperature (LSBT) instead [31][32][33], and some important parameters in [34] cannot easily be obtained in this research, so the at-satellite brightness temperature was used as the LSBT in our research.…”