How do PBL schemes in WRF describe summer and winter conditions at a high arctic site?

We compare 4 planetary boundary layer (PBL) schemes of Weather Research and Forecasting model for high Artic conditions, documented during summer and winter campaigns at Station Nord, Greenland. During March 2012, 22 radiosonds were launched at 00 and 12 UTC. During July-August 2011, 25 radiosondes were launched at 00, 06, 12, and 18 UTC.

The chosen PBL schemes are 3 TKE schemes: MYJ, MYNN and QNSE and non-local YSU. Comparison is performed between data from radiosoundings and corresponding in time model results up to different height from 100 m to 8000 m. Sensitivity of model to vertical and spatial resolution is examined with MYJ through 4 configurations combining 26 or 42 vertical levels and 4 km or 1.33 km horizontal grid step.

Sensitivity to the resolution tests showed that increasing horizontal resolution from 4 km to 1.33 km did not improve model performance. Increasing the number of vertical layers lead to closer to observed profiles and slightly improved statistics by layers. Sensitivity to the lead time (24 h or 48 h) is examined with MYJ at 1.33 km grid step and 42 vertical layers.

Quality of forecast for day 1 and day 2 is similar for the summer. Temperature and wind speed (WS) biases for the winter are with 1 K and 1 ms-1 larger for 48 h compared to 24 h lead time. The lack of diurnal variability during both campaigns is correctly simulated by all PBL schemes. The performed tests show that TKE schemes outperform YSU and as a whole MYNN gives the highest scores.