Validation and trend analysis of satellite-derived surface water temperature observations over adirondack lakes

Abstract

This study aims to validate and evaluate satellite remote sensing observations from the Landsat series over 135 lakes in the Adirondack State Park, located in upstate New York, and to examine their surface temperature trends over the past 40 years. It utilizes data from the Moderate Resolution Imaging Spectroradiometer (MODIS), along with Landsat 5 and 7. Park-scale results were derived by extracting MODIS surface temperatures within the park boundary, while lake-scale results were estimated using Landsat 5 (1984-2012) and Landsat 7 (1999-2023) observations. In addition, field observations were utilized to perform a comprehensive validation and evaluation of satellite-based surface temperature data. The findings indicate that satellite-derived temperatures are within a reasonable range of in situ measurements, with RMSEs ranging from 1.97 to 2.08°C and correlation coefficients between 0.64 and 0.75. Comprehensive trend analyses conducted on an annual and monthly basis reveal that temperature values during the Landsat 5 period exhibited higher warming rates compared to the Landsat 7 period, which showed seasonal variations and even some cooling months. Monthly MODIS-derived park-wide trends ranged from −1.57 °C to 1.2 °C (March to November) and were generally lower in magnitude than those estimated from Landsat data. On average, lakes in the Adirondack lakes demonstrated a warming trend of about 0.89°C per decade. The seasonal analyses revealed an expansion of the summer season, with higher warming rates in May and October/September. These findings highlight the critical role of satellite observations in monitoring climate change impacts on lake ecosystems, especially in regions with limited ground-based data.