Reporting the rapid warming of italian lakes derived from homogenized multi-sensor satellite data.

Recent studies, using a combination of long term in-situ and satellite data indicate that lakes are warming rapidly at the global scale. Lake Surface Water Temperature (LSWT) being sensitive to long term modifications in the thermal structure of lakes is a good indicator of changes in lake characteristics. Further studies are needed to understand variation in trends, impacts and consequences at the regional scale. However, long term in-situ data recorded with high temporal frequency are often lacking. Remote sensing is considered a promising approach to reconstruct complete time series of LSWT where direct measurements are missing. Temperature of land/water surfaces is one of the direct and accurate measurement using the satellite data acquired in thermal infra-red spectral region. Furthermore, some measurements are even available since the 1980 at the daily temporal resolution and 1 km moderate spatial resolution. In this study, we aim to i) reconstruct a homogenized LSWT database for six large Italian lakes by combining thermal data from multiple satellites, ii) assess the quality of satellite derived LSWT using long term in-situ data collected from these lakes, iii) report the seasonal and annual trends in LSWT using robust statistical tests. We compiled the LSWT database using daily data spanning 29 years (1986 - 2015) at a spatial resolution of 1 km, time standardised to 12:00 UTC using a new methodology, which combine data from multiple sensors. We used dual thermal channels (10.5 – 11.5 μm and 11.5 – 12.5 μm) and a split window algorithm with sensor specific coefficients to derive LSWT. We calibrated the thermal data acquired by six sensors on-board thirteen satellites and corrected them geometrically before deriving the LSWT. The sensors used for this study were (satellites in bracket) – AVHRR/2 (NOAA-9/11/12/14), AVHRR/3 (NOAA-16/17/18/19), ATSR1 (ERS-1), ATSR2 (ERS-2), AATSR (ENVISAT) and MODIS (AQUA/TERRA). We applied a modified Diurnal Temperature Cycle (DTC) model to correct the LSWT for different acquisition time of the satellites. Using this new LSWT dataset we are studying the long-term annual and seasonal trends in the peri-alpine lakes - Garda, Iseo, Como, Lugano, Maggiore, and the Lake Trasimeno in Central Italy. We found good agreement between LSWT and in-situ data with an average R2 and RMSE of 0.90 and 1.5 K, respectively. With regard to seasonal and annual trends, in Lake Garda, surface water mean temperature showed a significant (**P < 0.05) increase with annual rate of 0.013 ºC yr-1 and of 0.03 ºC yr-1 during summer (Kendall tests and Theil-Sen’s slope estimates). The statistical analysis of other lakes are in progress and these results will allow us to evaluate further the accuracy of the LSWT reconstructed using satellite data and assess the regional coherence of the trends detected in lake Garda.