In this paper, we introduce a testbed for evaluating and comparing climate modeling systems at cloud resolving scales using hindcasts of the June 2012 North American derecho. The testbed is applied to two models: the regionally-refined Simple Cloud-Resolving E3SM Atmosphere Model (SCREAM) at horizontal resolutions ranging from 6.5 to 1.625 km and the Weather Research and Forecasting (WRF) model with 4 km grid spacing. We find the simulation results to be highly sensitive to the initial conditions, initialization time, and model configurations, with initial conditions from the Rapid Refresh (RAP) producing the best simulation. Significant improvement is identified in the SCREAM simulations as horizontal grid spacing is refined. While a propagation delay of approximately 2 hours is found in both models, SCREAM at 1.625 km simulates the observed bow echo structure of the derecho well and predicts strong surface gusts that exceed 30 m/s. In comparison, WRF hardly produces surface wind over 25 m/s, and the derecho wind gust in WRF is 42-46% lower than in SCREAM. Moreover, WRF has a lower bias in simulating cold clouds but overestimates the precipitation intensity. Both models well reproduce the observed outgoing longwave radiation spatial patterns (Pearson correlation > 0.88) while they simulate larger areas of composite radar reflectivity > 40 dBZ by up to 4 times and underestimate the precipitating area by ~ 70\% in WRF and 47\% in SCREAM compared to observations.

Atmospheric rivers (ARs) are long and narrow filaments of vapor transport responsible for most poleward moisture transport outside of the tropics. Many AR detection algorithms have been developed to automatically identify ARs in climate data. The diversity of these algorithms has introduced appreciable uncertainties in quantitative measures of AR properties and thereby impedes the construction of a unified and internally consistent climatology of ARs. This paper compares eight global AR detection algorithms from the perspective of AR life cycles following the propagation of ARs from origin to termination in the MERRA2 reanalysis over the period 1980-2017. Uncertainties related to lifecycle characteristics, including number, lifetime, intensity, and frequency distribution are discussed. Notably, the number of AR events per year in the Northern Hemisphere can vary by a factor of 5 with different algorithms. Although all algorithms show that the maximum origin (termination) frequency locates over the northwestern (northeastern) Pacific, significant disagreements occur in regional distribution. Spreads are large in AR lifetime and intensity. The number of landfalling AR events produced by the algorithms can vary from 16 to 78 events per cool season, i.e. by almost a factor of five, although the agreement improves for stronger ARs. By examining the AR’s connection with the Madden-Julian Oscillation and El Niño Southern Oscillation, we find that the overall responses of ARs (such as changes in AR frequency, origin, and landfalling activity) to low-frequency climate variabilities are consistent among algorithms.

Drought-busting events in the Colorado River Basin, such as the “Miracle May” of 2015 that greatly alleviated an unprecedented water shortage, have been observed for more than a century. But while such events are much prayed for in times of drought, they have not been well researched or even characterized. In this study, conducted in collaboration with water managers from across the basin, we propose a definition for “miracle events” that reflects real-world, actionable relevance. The resulting characterization offers a framework by which to quantify the frequency and strength of extreme dry-to-wet springtime transitions. While limited by uncertainties in model simulations and the myriad hydrological futures these simulations seek to project, and thus requiring cautious interpretation, this study finds that such transitions may become less frequent and less powerful under climate warming.

Cyclonic low-pressure systems (LPS) produce abundant rainfall in South Asia, where they are traditionally categorized as monsoon lows, monsoon depressions, and more intense cyclonic storms. The India Meteorological Department (IMD) has tracked monsoon depressions for over a century, finding a large decline in their number in recent decades, but their methods have changed over time and do not include monsoon lows. This study presents a fast, objective algorithm for identifying monsoon LPS in high-resolution datasets. Variables and thresholds used in the algorithm are selected to best match a subjectively analyzed LPS dataset while minimizing disagreement between four atmospheric reanalyses in a training period. The streamfunction of the 850 hPa horizontal wind is found to be the best variable for tracking LPS; it is less noisy than vorticity and represents the complete non-divergent wind, even when flow is not geostrophic. Using this algorithm, LPS statistics are computed for five reanalyses, and none show a detectable trend in monsoon depression counts since 1979. Both the Japanese 55-year Reanalysis (JRA-55) and the IMD dataset show a step-like reduction in depression counts when they began using geostationary satellite data, in 1979 and 1982 respectively; the 1958-2018 linear trend in JRA-55, however, is smaller than in the IMD dataset and its error bar includes zero. There are more LPS in seasons with above-average monsoon rainfall and also in La Nin ̵̃a years, but few other large-scale modes of interannual climate variability are found to modulate LPS counts, lifetimes, or track length consistently across all reanalyses.

Synoptic-scale cyclonic vortices produce abundant rainfall in South Asia, where these low pressure systems (LPS) are traditionally categorized as monsoon lows, monsoon depressions, and more intense cyclonic storms. The India Meteorological Department has tracked monsoon depressions for over a century, finding a large decline in the number of those storms in recent decades; their tracking methods, however, seem to have changed over time and do not include monsoon lows, which can produce intense rainfall despite their weak winds. This study presents a fast and objective tracking algorithm that can identify monsoon LPS in high-resolution datasets with a variety of grid structures. A sensitivity analysis has been performed to select a set of atmospheric variables and their corresponding thresholds for optimal tracking of LPS. Approximately 250 combinations of variables and thresholds are used to identify LPS over roughly a decade (the training period) in each of four atmospheric reanalyses, and these combinations are ranked using a skill score that compares the reanalyses with each other and with a preexisting track dataset that was compiled by subjective identification of LPS. This procedure finds the streamfunction of the 850 hPa horizontal wind to be the best variable for tracking LPS. The streamfunction is smoother than the vorticity field and represents the complete non-divergent component of the wind even when the flow is not geostrophic, unlike the geopotential height or sea level pressure. Using this tracking algorithm, LPS statistics are then computed in five reanalysis products that each span at least 40 years, with a primary goal being to determine whether the large decrease in monsoon depressions seen in the India Meteorological Department track dataset since the 1970s can be found in any reanalysis. This trend assessment is particularly relevant for the ERA5 reanalysis, which extends back to 1950 and which contains explicit climate forcings. In addition to secular trends, this study assesses the decadal variation of LPS, as well as interannual changes in LPS activity that are associated with the El Niño-Southern Oscillation and the Indian Ocean Dipole.

Substantial progress on machine learning (ML) models and graphical processing units (GPUs) has stimulated emerging research in applications of ML to earth science. As snow is a vital component of the global hydroclimate system, precise snowpack prediction is of considerable value for science and society. In this work, we have trained three different ML models (LSTM, CNN and Attention) to predict daily snow water equivalent (SWE) with both dynamic and static features in the Western Contiguous United States from Snow Telemetry (SNOTEL) observations. Dynamic features include precipitation, minimum and maximum temperature, minimum and maximum relative humidity, specific humidity, solar radiation and wind velocity. Static features are latitude, longitude, elevation, diurnal anisotropic heating (DAH) index and topographic radiative aspect (TRASP) index. This choice of features allows us to produce high-resolution maps of regional SWE for a given set of input meteorological conditions. The importance and the sensitivity of input variables will be tested by several explainable AI methods including feature permutation and integrated gradient. The ML-based dataset is further up-sampled and compared with the 4km gridded SWE dataset from the National Snow & Ice Data Center (NSIDC), which is from a physical-based model. Future SWE estimates are also produced under climate conditions projected by CMIP class models, along with associated uncertainty estimates based on our sensitivity analysis. The ML models are demonstrated to be a fast and accurate method of producing high-resolution SWE estimates with minimal computing power.

Future changes in tropical cyclone properties are an important component of climate change impacts and risk for many tropical and mid-latitude countries. In this study we assess the performance of a multi-model ensemble of climate models, at resolutions ranging from 250km to 25km. We use a common experimental design including both atmosphere-only and coupled simulations run over the period 1950-2050, with two tracking algorithms applied uniformly across the models. There are overall improvements in tropical cyclone frequency, spatial distribution and intensity in models at 25 km resolution, with several of them able to represent very intense storms. Projected tropical cyclone activity by 2050 generally declines in the South Indian Ocean, while changes in other ocean basins are more uncertain and sensitive to both tracking algorithm and imposed forcings. Coupled models with smaller biases suggest a slight increase in average TC 10m wind speeds by 2050.