محسن رحيميان

Abstract: The rapid desiccation of lakes, as one of the consequences of global warming an‎d climate change, alters the energy an‎d moisture exchange patterns between the lake surface an‎d the atmosphere, thereby exerting significant impacts on local an‎d regional climates. Lake Urmia, one of the world’s largest hypersaline lakes, has experienced a severe water level decline over the past decades. This phenomenon has changed surface albedo, roughness, heat capacity, an‎d evaporation patterns, posing serious challenges for realistic representation of these processes in numerical weather models. The main objective of this study is to enhance the WRF-Lake model for more realistic simulation of atmosphere–lake interactions under the unique conditions of hypersaline lakes. In the first step, analysis of 33-year climatic data (1985–2017) from four synoptic stations revealed that during the intense desiccation period (1996–2006), the mean local temperature anomaly relative to the regional climate was +0.124°C, increasing to +0.127°C in the critical period (2007–2017). Based on these findings, numerical simulations using WRF-Lake were conducted for a wet year (1998) an‎d a dry year (2017). Results showed that the default model suffers from a strong cold bias (−20°C to −24°C) in estimating lake surface water temperature (LSWT) an‎d unrealistically simulates lake freezing during the cold season, limiting its ability to capture air–lake coupling an‎d local meteorological responses. In the first phase of model improvement, physical modifications due to lake desiccation including lan‎d cover, albedo, soil texture, topography, an‎d actual bathymetry were incorporated. These updat‎es led to notable improvements in simulating the February 2017 precipitation an‎d snow-cover event: the POD increased from 0.80 to 0.94, CSI from 0.70 to 0.91, an‎d the correlation coefficient for precipitation from 0.57 to 0.68 near the lake area. However, the cold bias in LSWT (about −15°C) an‎d 2-m air temperature (about −10°C) persisted as a systematic error. To address this, the second phase introduced a newly formulated model version named SLake, in which the governing thermodynamic equations were rewritten to represent the effects of high salinity (~300 g L⁻¹). The relationships for water density, freezing point, an‎d saturation vapor pressure were modified accordingly. Results indicated that SLake successfully removed unrealistic thermal stratification an‎d reproduced the homogeneous behavior of a shallow lake. Nevertheless, the negative LSWT bias remained at about −15°C (RMSE = 15.22°C at the DWP station). In the third an‎d final phase, an advanced version termed SLake_LSWT was developed, in which the lake surface temperature was dynamically updat‎ed every 12 hours using observational data. This approach yielded substantial improvements across all parameters: LSWT bias reduced from −15.19°C to −0.56°C (96.3% improvement), RMSE from 15.22°C to 1.48°C (90.3% improvement), an‎d the correlation coefficient increased from 0.48 to 0.93 at the DWP station. Similarly, the 2-m air temperature simulation improved with bias decreasing from −10.59°C to −3.63°C an‎d RMSE from 10.95°C to 4.32°C. Evaporation estimates also achieved strong performance with a correlation coefficient of 0.81 an‎d RMSE of 0.54 mm day⁻¹, outperforming the GLEAM dataset. Seasonal eva‎luation for spring, summer, an‎d autumn of 2017 confirmed consistent performance, as the KGE for LSWT increased from 0.49→0.84 (spring), 0.38→0.52 (summer), an‎d 0.17→0.84 (autumn). For 2-m air temperature, KGE improved from 0.35–0.58 (baseline) to 0.81–0.87 (modified version), indicating concurrent enhancement in correlation, bias, an‎d variability. The findings demonstrate that the combined implementation of (1) lan‎d-surface physical corrections, (2) salinity-dependent thermodynamic reformulation, an‎d (3) dynamic LSWT updating can substantially improve the performance of WRF-Lake for hypersaline lake environments. The developed SLake_LSWT framework not only enhances prediction accuracy for Lake Urmia but also provides a transferable approach for other saline an‎d shallow water bodies such as the Dead Sea, Aral Sea, an‎d Great Salt Lake, offering a robust foundation for modeling atmosphere–lake interactions under extreme climatic conditions.

درياچه اروميه , درياچه فوق شور , به‌روزرساني دماي سطح درياچه , اندركنش درياچه-جو , مدل WRF-Lake

Urmia Lake , Hypersaline lake , Lake surface temperature updat‎e , Lake-atmosphere interaction , WRF-Lake model

Mohsen Rahimian

Seyed Mostafa Siadat Mousavi