Remote Sensing of Lake Ice Phenology across a Range of Lakes Sizes, ME, USA

Shuai Zhang,Tamlin M. Pavelsky

doi:10.3390/rs11141718

Shuai Zhang, Tamlin M. Pavelsky

Open Access

https://doi.org/10.3390/rs11141718

Copy DOI

Journal: Remote sensing	Publication Date: Jul 20, 2019
Citations: 30	License type: CC BY 4.0

Affiliation: University of North Carolina at Chapel Hill

Abstract

Remote sensing of ice phenology for small lakes is hindered by a lack of satellite observations with both high temporal and spatial resolutions. By merging multi-source satellite data over individual lakes, we present a new algorithm that successfully estimates ice freeze and thaw timing for lakes with surface areas as small as 0.13 km2 and obtains consistent results across a range of lake sizes. We have developed an approach for classifying ice pixels based on the red reflectance band of Moderate Resolution Imaging Spectroradiometer (MODIS) imagery, with a threshold calibrated against ice fraction from Landsat Fmask over each lake. Using a filter derived from the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) surface air temperature product, we removed outliers in the time series of lake ice fraction. The time series of lake ice fraction was then applied to identify lake ice breakup and freezeup dates. Validation results from over 296 lakes in Maine indicate that the satellite-based lake ice timing detection algorithm perform well, with mean absolute error (MAE) of 5.54 days for breakup dates and 7.31 days for freezeup dates. This algorithm can be applied to lakes worldwide, including the nearly two million lakes with surface area between 0.1 and 1 km2.

Highlights

The timing of lake ice freeze and thaw is an indicator of climate change because it responds directly and rapidly to variations in air temperature [1,2,3,4,5]
Active sensors such as Synthetic Aperture Radar (SAR) have been employed for measuring the ice depth over lakes [10,11,12,13], and the visible or near infrared (VIS/NIR) imagery such as Moderate Resolution Imaging Spectroradiometer (MODIS) and AVHRR have been used for detecting ice phenology over large lakes [14]
When the lake ice fraction first dropped below 20% each spring, we identified the breakup date

Summary

Introduction

The timing of lake ice freeze and thaw is an indicator of climate change because it responds directly and rapidly to variations in air temperature [1,2,3,4,5]. It can affect its surrounding environmental and climate systems [3,6]. The smallest lakes observed in these studies range from 100 km2 [1] to 1 km2 [15]

Methods

Findings

Discussion

Conclusion