# Determining the cause of inconsistent onset-season trends in the Northern Hemisphere snow cover extent record

## TL;DR

Satellites detect more snow cover now than 40 years ago due to improved technology, but this optical illusion needs correction to accurately track climate change effects.

## Contribution

The paper introduces a method to correct artificial snow cover trends in satellite records using a reanalysis-driven snow model.

## Key findings

- Artificial increasing snow cover trends in NOAA's CDR are due to changes in snow detection sensitivity.
- Merging recent CDR estimates with model data reveals decreasing Northern Hemisphere snow cover trends in most months.
- Offline snow models can serve as benchmarks to identify biases in observational datasets.

## Abstract

While seasonal snow cover extent (SCE), an essential climate variable, has broadly declined as a response to global warming, notable inconsistencies remain among long-term satellite-based estimates of SCE change during the Northern Hemisphere snow onset season. SCE datasets from a reanalysis-driven simple snow model serve as benchmarks and allow us to reconcile the trends from one prominent snow cover record with other recent studies. In particular, artificial increasing snow cover trends in the National Oceanic and Atmospheric Administration’s Snow Cover Extent Climate Data Record (CDR) during the onset season are related to changes in snow detection sensitivity. This artificial drift primarily affects September, October, and November snow cover but is detectible through February. Revised trends produced by merging the last decade’s CDR estimates with the offline model datasets reveal decreasing Northern Hemisphere trends in all months but January. This approach shows that offline snow models produce useful benchmarks that can expose biases in observational snow cover datasets with other cross-validation.

With “better glasses” on, satellites detect more snow cover now than 40 years ago, but this optical illusion should be corrected.