Search Results

You are looking at 1 - 3 of 3 items for

  • Author or Editor: Dan Lu x
  • Refine by Access: All Content x
Clear All Modify Search
Pengyuan Li, Gang Fu, Chungu Lu, Dan Fu, and Shuai Wang


In this paper, a dense sea fog event that occurred over the Yellow Sea (YS) on 9 March 2005 is investigated using the Weather Research and Forecasting Model version 3.1.1 (WRF v3.1.1). It is shown that the WRF can reasonably reproduce the main features of this fog case with a newly implemented planetary boundary layer (PBL) scheme developed by Mellor–Yamada–Nakanishi–Niino (MYNN). The low-level jet (LLJ) associated with this fog episode played an important role in triggering the turbulence. During the fog formation, sea fog extended vertically with the aid of turbulence. The mechanical production term resulting from wind shear contributed to the generation of the turbulence. WRF simulation results showed that the fog layer was thicker in the northeastern part of the YS than that in the southwestern part due to the intensity of the inversion layer and the LLJ. The topography test in which the mountain region in Fujian Province was removed showed that the roles of topography were to prevent the moisture from extending to land, to intensify the inversion layer, and to enhance the intensity of LLJ, as well as to elevate its altitude.

Full access
John Turner, Hua Lu, John King, Gareth J. Marshall, Tony Phillips, Dan Bannister, and Steve Colwell


We present the first Antarctic-wide analysis of extreme near-surface air temperatures based on data collected up to the end of 2019 as part of the synoptic meteorological observing programs. We consider temperatures at 17 stations on the Antarctic continent and nearby sub-Antarctic islands. We examine the frequency distributions of temperatures and the highest and lowest individual temperatures observed. The variability and trends in the number of extreme temperatures were examined via the mean daily temperatures computed from the 0000, 0600, 1200, and 1800 UTC observations, with the thresholds for extreme warm and cold days taken as the 5th and 95th percentiles. The five stations examined from the Antarctic Peninsula region all experienced a statistically significant increase (p < 0.01) in the number of extreme high temperatures in the late-twentieth-century part of their records, although the number of extremes decreased in subsequent years. For the period after 1979 we investigate the synoptic background to the extreme events using ECMWF interim reanalysis (ERA-Interim) fields. The majority of record high temperatures were recorded after the passage of air masses over high orography, with the air being warmed by the foehn effect. At some stations in coastal East Antarctica the highest temperatures were recorded after air with a high potential temperature descended from the Antarctic plateau, resulting in an air mass 5°–7°C warmer than the maritime air. Record low temperatures at the Antarctic Peninsula stations were observed during winters with positive sea ice anomalies over the Bellingshausen and Weddell Seas.

Open access
Dan Lu, Goutam Konapala, Scott L. Painter, Shih-Chieh Kao, and Sudershan Gangrade


Hydrologic predictions at rural watersheds are important but also challenging due to data shortage. Long short-term memory (LSTM) networks are a promising machine learning approach and have demonstrated good performance in streamflow predictions. However, due to its data-hungry nature, most LSTM applications focus on well-monitored catchments with abundant and high-quality observations. In this work, we investigate predictive capabilities of LSTM in poorly monitored watersheds with short observation records. To address three main challenges of LSTM applications in data-scarce locations, i.e., overfitting, uncertainty quantification (UQ), and out-of-distribution prediction, we evaluate different regularization techniques to prevent overfitting, apply a Bayesian LSTM for UQ, and introduce a physics-informed hybrid LSTM to enhance out-of-distribution prediction. Through case studies in two diverse sets of catchments with and without snow influence, we demonstrate that 1) when hydrologic variability in the prediction period is similar to the calibration period, LSTM models can reasonably predict daily streamflow with Nash–Sutcliffe efficiency above 0.8, even with only 2 years of calibration data; 2) when the hydrologic variability in the prediction and calibration periods is dramatically different, LSTM alone does not predict well, but the hybrid model can improve the out-of-distribution prediction with acceptable generalization accuracy; 3) L2 norm penalty and dropout can mitigate overfitting, and Bayesian and hybrid LSTM have no overfitting; and 4) Bayesian LSTM provides useful uncertainty information to improve prediction understanding and credibility. These insights have vital implications for streamflow simulation in watersheds where data quality and availability are a critical issue.

Restricted access