Deep learning models for screening of high myopia using optical coherence tomography

This study aimed to validate and evaluate deep learning (DL) models for screening of high myopia using spectral-domain optical coherence tomography (OCT). This retrospective cross-sectional study included 690 eyes in 492 patients with OCT images and axial length measurement. Eyes were divided into three groups based on axial length: a “normal group,” a “high myopia group,” and an “other retinal disease” group. The researchers trained and validated three DL models to classify the three groups based on horizontal and vertical OCT images of the 600 eyes. For evaluation, OCT images of 90 eyes were used. Diagnostic agreements of human doctors and DL models were analyzed. The area under the receiver operating characteristic curve of the three DL models was evaluated. Absolute agreement of retina specialists was 99.11% (range: 97.78–100%). Absolute agreement of the DL models with multiple-column model was 100.0% (ResNet 50), 90.0% (Inception V3), and 72.22% (VGG 16). Areas under the receiver operating characteristic curves of the DL models with multiple-column model were 0.99 (ResNet 50), 0.97 (Inception V3), and 0.86 (VGG 16). The DL model based on ResNet 50 showed comparable diagnostic performance with retinal specialists. The DL model using OCT images demonstrated reliable diagnostic performance to identify high myopia.

Impacts of the photo-driven post-depositional processing on snow nitrate and its isotopes at Summit, Greenland: a model-based study

Atmospheric information embedded in ice-core nitrate is disturbed by post-depositional processing. Here we used a layered snow photochemical column model to explicitly investigate the effects of post-depositional processing on snow nitrate and its isotopes (δ15N and Δ17O) at Summit, Greenland, where post-depositional processing was thought to be minimal due to the high snow accumulation rate. We found significant redistribution of nitrate in the upper snowpack through photolysis, and up to 21 % of nitrate was lost and/or redistributed after deposition. The model indicates post-depositional processing can reproduce much of the observed δ15N seasonality, while seasonal variations in δ15N of primary nitrate are needed to reconcile the timing of the lowest seasonal δ15N. In contrast, post-depositional processing can only induce less than 2.1 ‰ seasonal Δ17O change, much smaller than the observation (9 ‰) that is ultimately determined by seasonal differences in nitrate formation pathway. Despite significant redistribution of snow nitrate in the photic zone and the associated effects on δ15N seasonality, the net annual effect of post-depositional processing is relatively small, suggesting preservation of atmospheric signals at the annual scale under the present Summit conditions. But at longer timescales when large changes in snow accumulation rate occur this post-depositional processing could become a major driver of the δ15N variability in ice-core nitrate.

A Simple Lagrangian Parcel Model for the Initiation of Summer-time Mesoscale Convective Systems over the Central United States

Mesoscale convective systems (MCSs) account for more than 50% of summer-time precipitation over the central United States (US) and have a significant impact on local weather and hydrologic cycle. It is hypothesized that the inadequate treatment of MCSs is responsible for the longstanding warm and dry bias over the central US in coarse-resolution general circulation model (GCM) simulations. In particular, a better understanding of MCS initiation is still lacking. Here a single-column Lagrangian parcel model is first developed to simulate the basic features of a rising parcel. This simple model demonstrates the collective effects of boundary layer moistening and dynamical lifting in triggering convective initiation and reproduces successfully its early afternoon peak with surface equivalent potential temperature as a controlling factor. It also predicts that convection is harder to trigger in the future climate under global warming, consistent with the results from convection-permitting regional climate simulations. Then a multi-column model that includes an array of single-column models aligned in the east-west direction and incorporates idealized cold pool interaction mechanisms is developed. The multi-column model captures readily the cold pool induced upscale growth feature in MCS genesis from initially scattered convection that is organized into a mesoscale cluster in a few hours. It also highlights the crucial role of lifting effects due to cold pool collision and spreading, subsidence effect, and gust front propagation speed in controlling the final size of mesoscale clusters and cold pool regions. This simple model should be useful for understanding fundamental mechanisms of MCS initiation and providing guidance for improving MCS simulations in GCMs.

Impacts of the photo-driven post-depositional processing on snow nitrate and its isotopes at Summit, Greenland: a model-based study

Atmospheric information embedded in ice-core nitrate is disturbed by post-depositional processing. Here we used a layered snow photochemical column model to explicitly investigate the effects of post-depositional processing on snow nitrate and its isotopes (δ15N and Δ17O) at Summit, Greenland where post-depositional processing was thought to be minimal due to the high snow accumulation rate. We found significant redistribution of nitrate in the upper snowpack through photolysis and up to 21 % of nitrate was lost and/or redistributed after deposition. The model indicates post-depositional processing can reproduce much of the observed δ15N seasonality, while seasonal variations in δ15N of primary nitrate is needed to reconcile the timing of the lowest seasonal δ15N. In contrast, post-depositional processing can only induce less than 2.1 ‰ seasonal Δ17O change, much smaller than the observation (9 ‰) that is ultimately determined by seasonal differences in nitrate formation pathway. Despite significant redistribution of snow nitrate in the photic zone and the associated effects on δ15N seasonality, the net annual effect of post-depositional processing is relatively small, suggesting preservation of atmospheric signals at the annual scale under the present Summit conditions. But at longer timescales when large changes in snow accumulation rate occurs this post-depositional processing could become a major driver of the δ15N variability in ice core nitrate.