Inland lake temperature initialization via coupled cycling with atmospheric data assimilation

Abstract

Abstract. Application of lake models coupled within earth-system prediction models, especially for predictions from days to weeks, requires accurate initialization of lake temperatures. Commonly used methods to initialize lake temperatures include interpolation of global sea-surface temperature (SST) analyses to inland lakes, daily satellite-based observations, or model-based reanalyses. However, each of these methods have limitations in capturing the temporal characteristics of lake temperatures (e.g., effects of anomalously warm or cold weather) for all lakes within a geographic region and/or during extended cloudy periods. An alternative lake-initialization method was developed which uses two-way-coupled cycling of a small-lake model within an hourly data assimilation system of a weather prediction model. The lake model simulated lake temperatures were compared with other estimates from satellite and in situ observations and interpolated-SST data for a multi-month period in 2021. The lake cycling initialization, now applied to two operational US NOAA weather models, was found to decrease errors in lake surface temperature from as much as 5–10 K vs. interpolated-SST data to about 1–2 K compared to available in situ and satellite observations.

A database of in situ water temperatures for large inland lakes across the coterminous United States

“…Accurately representing spatial and temporal variability of lake surface water temperatures in numerical weather prediction (NWP) systems has been shown (particularly for Earth’s largest lakes) to improve short- and long-term forecasts of regional precipitation, air temperature, and surface wind velocity 1 – 4 . Thus, realistic representation of lake conditions is crucial for the development of the next generation of climate and weather forecast models 5 , 6 . The database we introduce here was developed to support this advancement by providing in situ validation data for a broader project sponsored by the National Oceanic and Atmospheric Administration (NOAA) through its Joint Technology Transfer Initiative (JTTI).…”

Lake ice simulation using a 3D unstructured grid model

“…Most lake ice models are adapted from sea ice models with minor modifications in parameterizations (e.g., White et al 2012;Bai et al 2020;Li et al 2021). Aside from the large lakes (e.g., the Great Lakes), a popular efficient approach for modeling lake ice is based on one-dimensional models (Mironov et al 2010;Smirnova et al 2016;Xiao et al 2016;Benjamin et al 2022). These one-dimensional lake models appear to be sufficient for modeling small lakes but have limitations for modeling ice and hydrodynamics in large lakes.…”