Search for: All records

We compare high‐resolution land‐surface temperature (LST) estimates from the GOES‐16/17 (GOES) satellites to ERA‐5 Land (ERA‐5) reanalysis data across nine large US cities. We quantify the offset and find that ERA‐5 generally overestimates LST compared to GOES by 1.63°C. However, this overestimation is less pronounced in urban areas, underscoring the limitations of ERA‐5 in capturing the LST gradient between urban and non‐urban areas. We then examine three quantities: Surface Urban Heat Island Intensity (SUHII), extreme LST events, and LST exposure by population. We find that ERA‐5 does not accurately represent the diurnal variation and magnitude of SUHII in GOES. Furthermore, while ERA‐5 was on average too warm, ERA‐5 underestimates extreme heat by an average of 2.40°C. Our analysis reveals higher population exposure to high LST in the GOES data set across the cities studied. This discrepancy is especially pronounced when estimating the population fraction that are most exposed to heat.

 

Mortality due to extreme temperatures is one of the most worrying impacts of climate change. In this analysis, we use historic mortality and temperature data from 106 cities in the United States to develop a model that predicts deaths attributable to temperature. With this model and projections of future temperature from climate models, we estimate temperature‐related deaths in the United States due to climate change, changing demographics, and adaptation. We find that temperature‐related deaths increase rapidly as the climate warms, but this is mainly due to an expanding and aging population. For global average warming below 3°C above pre‐industrial levels, we find that climate change slightly reduces temperature‐related mortality in the U.S. because the reduction of cold‐related mortality exceeds the increase in heat‐related deaths. Above 3°C warming, whether the increase in heat‐related deaths exceeds the decrease in cold‐related deaths depends on the level of adaptation. Southern U.S. cities are already well adapted to hot temperatures and the reduction of cold‐related mortality drives overall lower mortality. Cities in the Northern U.S. are not well adapted to high temperatures, so the increase in heat‐related mortality exceeds the reduction in cold‐related mortality. Thus, while the total number of climate‐related mortality may not change much, climate change will shift mortality in the U.S. to higher latitudes.

 

The Electric Reliability Council of Texas (ERCOT) manages the electric power across most of Texas. They make short-term assessments of electricity demand on the basis of historical weather over the last two decades, thereby ignoring the effects of climate change and the possibility of weather variability outside the recent historical range. In this paper, we develop an empirical method to predict the impact of weather on energy demand. We use that with a large ensemble of climate model runs to construct a probability distribution of power demand on the ERCOT grid for summer and winter 2021. We find that the most severe weather events will use 100% of available power—if anything goes wrong, as it did during the 2021 winter, there will not be sufficient available power. More quantitatively, we estimate a 5% chance that maximum power demand would be within 4.3 and 7.9 GW of ERCOT’s estimate of best-case available resources during summer and winter 2021, respectively, and a 20% chance it would be within 7.1 and 17 GW. The shortage of power on the ERCOT grid is partially hidden by the fact that ERCOTs seasonal assessments, which are based entirely on historical weather, are too low. Prior to the 2021 winter blackout, ERCOT forecast an extreme peak load of 67 GW. In reality, we estimate hourly peak demand was 82 GW, 22% above ERCOT’s most extreme forecast and about equal to the best-case available power. Given the high stakes, ERCOT should develop probabilistic estimates using modern scientific tools to predict the range of power demand more accurately.

 

Abstract. This study investigates the impact of global warming on heat and humidityextremes by analyzing 6 h output from 28 members of the Max PlanckInstitute Grand Ensemble driven by forcing from a 1 % yr−1 CO2 increase. We find that unforced variability drives large changes in regional exposure to extremes in different ensemble members, and these variations are mostly associated with El Niño–Southern Oscillation (ENSO) variability. However, while the unforced variability in the climate can alter the occurrence of extremes regionally, variability within the ensemble decreases significantly as one looks at larger regions or at a global population perspective. This means that, for metrics of extreme heat and humidity analyzed here, forced variability in the climate is more important than the unforced variability at global scales. Lastly, we found that most heat wave metrics will increase significantly between 1.5 and 2.0 ∘C, and that low gross domestic product (GDP) regions show significantly higher risks of facing extreme heat events compared to high GDP regions. Considering the limited economic adaptability of the population to heat extremes, this reinforces the idea that the most severe impacts of climate change may fall mostly on those least capable of adapting. 

Identifying dust aerosols from passive satellite images is of great interest for many applications. In this study, we developed five different machine-learning (ML) based algorithms, including Logistic Regression, K Nearest Neighbor, Random Forest (RF), Feed Forward Neural Network (FFNN), and Convolutional Neural Network (CNN), to identify dust aerosols in the daytime satellite images from the Visible Infrared Imaging Radiometer Suite (VIIRS) under cloud-free conditions on a global scale. In order to train the ML algorithms, we collocated the state-of-the-art dust detection product from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) with the VIIRS observations along the CALIOP track. The 16 VIIRS M-band observations with the center wavelength ranging from deep blue to thermal infrared, together with solar-viewing geometries and pixel time and locations, are used as the predictor variables. Four different sets of training input data are constructed based on different combinations of VIIRS pixel and predictor variables. The validation and comparison results based on the collocated CALIOP data indicate that the FFNN method based on all available predictor variables is the best performing one among all methods. It has an averaged dust detection accuracy of about 81%, 89%, and 85% over land, ocean and whole globe, respectively, compared with collocated CALIOP. When applied to off-track VIIRS pixels, the FFNN method retrieves geographical distributions of dust that are in good agreement with on-track results as well as CALIOP statistics. For further evaluation, we compared our results based on the ML algorithms to NOAA’s Aerosol Detection Product (ADP), which is a product that classifies dust, smoke, and ash using physical-based methods. The comparison reveals both similarity and differences. Overall, this study demonstrates the great potential of ML methods for dust detection and proves that these methods can be trained on the CALIOP track and then applied to the whole granule of VIIRS granule.