Explore the vast range of titles available.

\"Ever wonder how meteorologists predict the weather? Learn how to build a weather station of your very own with readily available tools and supplies. Then, following step-by-step directions, you can design and conduct experiments that will have you predicting the weather too\"--P. [4] of cover.

In this study, we present RainNet, a deep convolutional neural network for radar-based precipitation nowcasting. Its design was inspired by the U-Net and SegNet families of deep learning models, which were originally designed for binary segmentation tasks. RainNet was trained to predict continuous precipitation intensities at a lead time of 5 min, using several years of quality-controlled weather radar composites provided by the German Weather Service (DWD). That data set covers Germany with a spatial domain of 900km×900 km and has a resolution of 1 km in space and 5 min in time. Independent verification experiments were carried out on 11 summer precipitation events from 2016 to 2017. In order to achieve a lead time of 1 h, a recursive approach was implemented by using RainNet predictions at 5 min lead times as model inputs for longer lead times. In the verification experiments, trivial Eulerian persistence and a conventional model based on optical flow served as benchmarks. The latter is available in the rainymotion library and had previously been shown to outperform DWD's operational nowcasting model for the same set of verification events.RainNet significantly outperforms the benchmark models at all lead times up to 60 min for the routine verification metrics mean absolute error (MAE) and the critical success index (CSI) at intensity thresholds of 0.125, 1, and 5 mmh-1. However, rainymotion turned out to be superior in predicting the exceedance of higher intensity thresholds (here 10 and 15 mmh-1). The limited ability of RainNet to predict heavy rainfall intensities is an undesirable property which we attribute to a high level of spatial smoothing introduced by the model. At a lead time of 5 min, an analysis of power spectral density confirmed a significant loss of spectral power at length scales of 16 km and below. Obviously, RainNet had learned an optimal level of smoothing to produce a nowcast at 5 min lead time. In that sense, the loss of spectral power at small scales is informative, too, as it reflects the limits of predictability as a function of spatial scale. Beyond the lead time of 5 min, however, the increasing level of smoothing is a mere artifact – an analogue to numerical diffusion – that is not a property of RainNet itself but of its recursive application. In the context of early warning, the smoothing is particularly unfavorable since pronounced features of intense precipitation tend to get lost over longer lead times. Hence, we propose several options to address this issue in prospective research, including an adjustment of the loss function for model training, model training for longer lead times, and the prediction of threshold exceedance in terms of a binary segmentation task. Furthermore, we suggest additional input data that could help to better identify situations with imminent precipitation dynamics. The model code, pretrained weights, and training data are provided in open repositories as an input for such future studies.

The Tibetan Plateau (TP) plays a critical role in influencing regional and global climate, via both thermal and dynamical mechanisms. Meanwhile, as the largest high-elevation part of the cryosphere outside the polar regions, with vast areas of mountain glaciers, permafrost and seasonally frozen ground, the TP is characterized as an area sensitive to global climate change. However, meteorological stations are biased and sparsely distributed over the TP, owing to the harsh environmental conditions, high elevations, complex topography and heterogeneous surfaces. Moreover, due to the weak representation of the stations, atmospheric conditions and the local land–atmosphere coupled system over the TP as well as its effects on surrounding regions are poorly quantified. This paper presents a long-term (2005–2016) in situ observational dataset of hourly land–atmosphere interaction observations from an integrated high-elevation and cold-region observation network, composed of six field stations on the TP. These in situ observations contain both meteorological and micrometeorological measurements including gradient meteorology, surface radiation, eddy covariance (EC), soil temperature and soil water content profiles. Meteorological data were monitored by automatic weather stations (AWSs) or planetary boundary layer (PBL) observation systems. Multilayer soil temperature and moisture were recorded to capture vertical hydrothermal variations and the soil freeze–thaw process. In addition, an EC system consisting of an ultrasonic anemometer and an infrared gas analyzer was installed at each station to capture the high-frequency vertical exchanges of energy, momentum, water vapor and carbon dioxide within the atmospheric boundary layer. The release of these continuous and long-term datasets with hourly resolution represents a leap forward in scientific data sharing across the TP, and it has been partially used in the past to assist in understanding key land surface processes. This dataset is described here comprehensively for facilitating a broader multidisciplinary community by enabling the evaluation and development of existing or new remote sensing algorithms as well as geophysical models for climate research and forecasting. The whole datasets are freely available at the Science Data Bank (https://doi.org/10.11922/sciencedb.00103; Ma et al., 2020) and additionally at the National Tibetan Plateau Data Center (https://doi.org/10.11888/Meteoro.tpdc.270910, Ma 2020).

This study uses ship-based weather radar observations collected from research vessel Investigator to evaluate the Australian weather radar network calibration monitoring technique that uses spaceborne radar observations from the NASA Global Precipitation Mission (GPM). Quantitative operational applications such as rainfall and hail nowcasting require a calibration accuracy of ±1 dB for radars of the Australian network covering capital cities. Seven ground-based radars along the western coast of Australia and the ship-based OceanPOL radar are first calibrated independently using GPM radar overpasses over a 3-month period. The calibration difference between the OceanPOL radar (used as a moving reference for the second step of the study) and each of the seven operational radars is then estimated using collocated, gridded, radar observations to quantify the accuracy of the GPM technique. For all seven radars the calibration difference with the ship radar lies within ±0.5 dB, therefore fulfilling the 1 dB requirement. This result validates the concept of using the GPM spaceborne radar observations to calibrate national weather radar networks (provided that the spaceborne radar maintains a high calibration accuracy). The analysis of the day-to-day and hourly variability of calibration differences between the OceanPOL and Darwin (Berrimah) radars also demonstrates that quantitative comparisons of gridded radar observations can accurately track daily and hourly calibration differences between pairs of operational radars with overlapping coverage (daily and hourly standard deviations of ∼ 0.3 and ∼ 1 dB, respectively).

The Chemistry-Aerosol Mediterranean Experiment (ChArMEx; http://charmex.lsce.ipsl.fr) is a collaborative research program federating international activities to investigate Mediterranean regional chemistry-climate interactions. A special observing period (SOP-1a) including intensive airborne measurements was performed in the framework of the Aerosol Direct Radiative Impact on the regional climate in the MEDiterranean region (ADRIMED) project during the Mediterranean dry season over the western and central Mediterranean basins, with a focus on aerosol-radiation measurements and their modeling. The SOP-1a took place from 11 June to 5 July 2013. Airborne measurements were made by both the ATR-42 and F-20 French research aircraft operated from Sardinia (Italy) and instrumented for in situ and remote-sensing measurements, respectively, and by sounding and drifting balloons, launched in Minorca. The experimental setup also involved several ground-based measurement sites on islands including two ground-based reference stations in Corsica and Lampedusa and secondary monitoring sites in Minorca and Sicily. Additional measurements including lidar profiling were also performed on alert during aircraft operations at EARLINET/ACTRIS stations at Granada and Barcelona in Spain, and in southern Italy. Remote-sensing aerosol products from satellites (MSG/SEVIRI, MODIS) and from the AERONET/PHOTONS network were also used. Dedicated meso-scale and regional modeling experiments were performed in relation to this observational effort. We provide here an overview of the different surface and aircraft observations deployed during the ChArMEx/ADRIMED period and of associated modeling studies together with an analysis of the synoptic conditions that determined the aerosol emission and transport. Meteorological conditions observed during this campaign (moderate temperatures and southern flows) were not favorable to producing high levels of atmospheric pollutants or intense biomass burning events in the region. However, numerous mineral dust plumes were observed during the campaign, with the main sources located in Morocco, Algeria and Tunisia, leading to aerosol optical depth (AOD) values ranging between 0.2 and 0.6 (at 440 nm) over the western and central Mediterranean basins. One important point of this experiment concerns the direct observations of aerosol extinction onboard the ATR-42, using the CAPS system, showing local maxima reaching up to 150 M m−1 within the dust plume. Non-negligible aerosol extinction (about 50 M m−1) has also been observed within the marine boundary layer (MBL). By combining the ATR-42 extinction coefficient observations with absorption and scattering measurements, we performed a complete optical closure revealing excellent agreement with estimated optical properties. This additional information on extinction properties has allowed calculation of the dust single scattering albedo (SSA) with a high level of confidence over the western Mediterranean. Our results show a moderate variability from 0.90 to 1.00 (at 530 nm) for all flights studied compared to that reported in the literature on this optical parameter. Our results underline also a relatively low difference in SSA with values derived near dust sources. In parallel, active remote-sensing observations from the surface and onboard the F-20 aircraft suggest a complex vertical structure of particles and distinct aerosol layers with sea spray and pollution located within the MBL, and mineral dust and/or aged North American smoke particles located above (up to 6–7 km in altitude). Aircraft and balloon-borne observations allow one to investigate the vertical structure of the aerosol size distribution showing particles characterized by a large size (> 10 µm in diameter) within dust plumes. In most of cases, a coarse mode characterized by an effective diameter ranging between 5 and 10 µm, has been detected above the MBL. In terms of shortwave (SW) direct forcing, in situ surface and aircraft observations have been merged and used as inputs in 1-D radiative transfer codes for calculating the aerosol direct radiative forcing (DRF). Results show significant surface SW instantaneous forcing (up to −90 W m−2 at noon). Aircraft observations provide also original estimates of the vertical structure of SW and LW radiative heating revealing significant instantaneous values of about 5° K per day in the solar spectrum (for a solar angle of 30°) within the dust layer. Associated 3-D modeling studies from regional climate (RCM) and chemistry transport (CTM) models indicate a relatively good agreement for simulated AOD compared with observations from the AERONET/PHOTONS network and satellite data, especially for long-range dust transport. Calculations of the 3-D SW (clear-sky) surface DRF indicate an average of about −10 to −20 W m−2 (for the whole period) over the Mediterranean Sea together with maxima (−50 W m−2) over northern Africa. The top of the atmosphere (TOA) DRF is shown to be highly variable within the domain, due to moderate absorbing properties of dust and changes in the surface albedo. Indeed, 3-D simulations indicate negative forcing over the Mediterranean Sea and Europe and positive forcing over northern Africa. Finally, a multi-year simulation, performed for the 2003 to 2009 period and including an ocean–atmosphere (O–A) coupling, underlines the impact of the aerosol direct radiative forcing on the sea surface temperature, O–A fluxes and the hydrological cycle over the Mediterranean.

On 13 August 2019, a severe convective precipitation event affected the Shanghai region. At 850 hPa, a low-level shear line influenced Shanghai with surface convergence, while at 700 hPa, an inversion layer separated warm, moist lower air from colder, drier air aloft, favoring convection. Observations also revealed vertical wind shear, facilitating additional convective growth. Observations from local automatic weather stations (AWSs) and wind profiler radars (WPRs) indicate that five minutes before rainfall began, ground heat and northerly winds collided, triggering the precipitation. Both the S-band Qingpu SA radar and a novel single-polarization X-band Array weather radar system (Array Weather Radar, AWR) with three phased-array radar frontends and one radar backend captured this event. Compared with the relatively coarse spatiotemporal resolution of the Qingpu SA radar, the AWR provides high-resolution wind-field data, enabling the derivation of horizontal divergence and vertical vorticity. A detailed analysis of reflectivity, divergence, and vorticity in the AWR’s overlapping detection areas shows that, during the development and mature stages of the cell’s lifecycle, the volume of echoes with Z > 25 dBZ consistently increases, whereas echoes with Z > 45 dBZ grow in an oscillatory pattern, reaching five peaks. Moreover, at the altitudes where Z > 45 dBZ appears, regions of cyclonic vorticity emerge.

As an attractive technique for measuring water vapor, the Global Navigation Satellite System (GNSS) faces additional challenges in dynamic applications such as in the open sea. We present a new method of retrieving precipitable water vapor (PWV) based on GNSS multi‐antenna precise point positioning (PPP), which uses GNSS data from multiple antennas and incorporates the constraints of known baseline vector and common tropospheric delay. The 4‐day shipborne dynamic experiment along the China coast demonstrates that the baseline vector constraint shortens the convergence time of positioning and atmospheric parameters, and also slightly improves their accuracies. The common tropospheric delay constraint helps to provide compromised, more robust, and sometimes more accurate PWV estimates. An evaluation with radiosonde‐derived PWVs shows that the combination of the two constraints achieves the best accuracy reaching 4.2 mm. This method helps to expand GNSS meteorology to the vast ocean and benefits satellite altimetry and weather forecasting. Plain Language Summary Water vapor plays an important role in atmospheric processes ranging from global climate change to mesoscale and micro‐scale weather systems. The Global Navigation Satellite System (GNSS) is an attractive technique to measure water vapor, which has been extensively investigated for ground‐based static stations. In dynamic applications such as in the open sea, it is faced with additional challenges which may potentially degrade the performance. This study presents a new method for retrieving precipitable water vapor (PWV) based on GNSS multi‐antenna precise point positioning (PPP). It uses GNSS data from multiple antennas closely spaced and mounted on the same platform, and incorporates additional constraints of known baseline vector and common tropospheric delay. We have conducted a 4‐day shipborne dynamic experiment along the China coast and demonstrated that this method can shorten the convergence time and provide more robust and sometimes more accurate PWVs compared to conventional PPP. An accuracy level of 4.2 mm is achieved with this method for the ocean‐based dynamic experiment. This study helps to expand GNSS meteorology to challenging environments such as in the open sea, which will benefit weather forecasting and nowcasting. Key Points A new method of retrieving precipitable water vapor (PWV) is presented based on Global Navigation Satellite System (GNSS) multi‐antenna precise point positioning (PPP) This method can shorten the convergence time and provide more robust and more accurate GNSS‐based PWV The improved performance is helpful to expand GNSS meteorology to the vast ocean and benefits satellite altimetry and weather forecasting

The national German weather radar network operates in C-band between 5.6 and 5.65 GHz. In a radar network, individual transmit frequencies have to be chosen such that radar–radar-induced interferences are avoided. In a unique experiment the Hohenpeißenberg research radar and five operational systems from the radar network were used to characterize radar–radar-induced interferences as a function of the radar frequency. The results allow assessment of the possibility of adding additional C-band radars with magnetron transmitters into the existing network. Based on the experiment, at least a 15 MHz separation of the nominal radar frequency is needed to avoid a radar–radar interference. The most efficient mitigation of radar–radar interference is achieved by the “Radar Tango”, which refers to the synchronized scanning of all radar systems in the network. Based on those results, additional C-band radar systems can be added to the German weather radar network if a further improvement of the radar coverage is needed.

This study investigates the impact of very high frequency data assimilation on analysis and forecast accuracy with the local ensemble transform Kalman filter for idealized deep convection. Previous studies showed that assimilating every 30 s data from Phased Array Weather Radar (PAWR) alleviates the problem of strongly non-Gaussian error probability distribution due to rapid nonlinear evolution of deep convection in real-world cases. This study performs perfect model observing system simulation experiments to understand better the impact of assimilating radar reflectivity every 30 s focusing on non-Gaussianity. The idealized experimental settings have unique advantage in verifications for unobserved variables since it was unclear in the previous studies with real-world data. The results show that every 30 s data assimilation contributes to a significant improvement of the analysis accuracy, particularly for vertical velocity associated with strong convection, although the impact on the forecast accuracy is limited. We also find a significant reduction in the non-Gaussianity of first guess ensemble. The impact of assimilation frequency on reducing non-Gaussianity is enhanced when the uncertainty in background wind or stability is included in the initial ensemble perturbation.

This study investigates and compares large-scale moisture and heat budgets over the eastern rainy sea area around Dongsha, the western rainless sea area around Xisha, and the northern coastland of the South China Sea. Ten-year (2011–20) surface, balloon-sounding, satellite measurements, and ERA5 reanalysis are merged into the physically consistent data to study annual and vertical variations of the budgets. It shows that the surface and column-integrated heat and moisture budgets have the smallest annual evolution over the coastland. The latent heat as a key heat contributor in summer is mainly offset by total cold advection and partially offset by net radiative cooling. The horizontal moisture advection below 700 hPa presents moistening over the sea whereas drying over the coastland during rainy months, in which the vertical moisture advection presents moistening up to 250 hPa for all three subregions. The horizontal temperature advection is weak throughout the year over the sea but displays strong top warming and bottom cooling in summer and nearly the opposite in winter over the coastland. The diabatic cooling with a peak at ∼700 hPa in winter is largely due to the enhanced radiative cooling and latent cooling. While the diabatic heating with a peak at ∼500 hPa in summer is largely due to the enhanced latent heating. The earliest atmospheric heating and moistening occur in spring over the coastland, inducing the earliest precipitation increase. The enhanced heating and moistening over Xisha have a 1-month lag relative to Dongsha, resulting in lagging precipitation.