Machine learning approaches are valuable methods in hyperspectral remote sensing, especially for the classification of land cover or for the regression of physical parameters. While the recording of ...hyperspectral data has become affordable with innovative technologies, the acquisition of reference data (ground truth) has remained expensive and time-consuming. There is a need for methodological approaches that can handle datasets with significantly more hyperspectral input data than reference data. We introduce the Supervised Self-organizing Maps (SuSi) framework, which can perform unsupervised, supervised and semi-supervised classification as well as regression on high-dimensional data. The methodology of the SuSi framework is presented and compared to other frameworks. Its different parts are evaluated on two hyperspectral datasets. The results of the evaluations can be summarized in four major findings: (1) The supervised and semi-Supervised Self-organizing Maps (SOM) outperform random forest in the regression of soil moisture. (2) In the classification of land cover, the supervised and semi-supervised SOM reveal great potential. (3) The unsupervised SOM is a valuable tool to understand the data. (4) The SuSi framework is versatile, flexible, and easy to use. The SuSi framework is provided as an open-source Python package on GitHub.
The upper troposphere/lower stratosphere (UTLS) region plays an important role in the climate system. Changes in the structure and chemical composition of this region result in particularly large ...changes in radiative forcings of the atmosphere. Quantifying the processes that control UTLS composition (e.g., stratosphere‐troposphere exchange) therefore represents a crucial task. We assess the influence of uncertainties in the atmospheric mixing strength on global UTLS distributions of greenhouse gases (water vapor, ozone, methane, and nitrous oxide) and associated radiative effects. The study is based on multiannual simulations with the Chemical Lagrangian Model of the Stratosphere (CLaMS) driven by ERA‐Interim meteorological data and on a state‐of‐the‐art radiance code. Mixing, the irreversible part of transport, is controlled by the local horizontal strain and vertical shear of the atmospheric flow. We find that simulated radiative effects of water vapor and ozone, both characterized by steep gradients in the UTLS, are particularly sensitive to uncertainties of the atmospheric mixing strength. Globally averaged radiative effects are about 0.72 and 0.17 W/m2for water vapor and ozone, respectively. For ozone, the largest impact of mixing uncertainties is observed in the extra‐tropical lower stratosphere.
Key Points
Assessment of radiative effects of mixing uncertainties
Sensitivity of water vapor and ozone to mixing uncertainties
Importance of quantitative process representation in UTLS
In this work absolute values of gravity wave (GW) momentum flux are derived from global temperature measurements by the satellite instruments High Resolution Dynamics Limb Sounder (HIRDLS) and ...Sounding of the Atmosphere using Broadband Emission Radiometry (SABER). Momentum fluxes in the stratosphere are derived for both instruments and for SABER in the whole mesosphere. The large‐scale atmospheric background state is removed by a two‐dimensional Fourier decomposition in longitude and time, covering even planetary‐scale waves with periods as short as 1–2 days. Therefore, it is possible to provide global distributions of GW momentum flux from observations for the first time in the mesosphere. Seasonal as well as longer‐term variations of the global momentum flux distribution are discussed. GWs likely contribute significantly to the equatorward tilt of the polar night jet and to the poleward tilt of the summertime mesospheric jet. Our results suggest that GWs can undergo large latitudinal shifts while propagating upward. In particular, GWs generated by deep convection in the subtropical monsoon regions probably contribute significantly to the mesospheric summertime wind reversal at mid‐ and high latitudes. Variations in the GW longitudinal distribution caused by those convectively generated GWs are still observed in the mesosphere and could be important for the generation of the quasi two‐day wave. Indications for quasi‐biennial oscillation (QBO) induced variations of GW momentum flux are found in the subtropics. Also variations at time scales of about one 11‐year solar cycle are observed and might indicate a negative correlation between solar flux and GW momentum flux.
Key Points
Gravity wave momentum flux global distributions in both strato‐ and mesosphere
Interaction of gravity waves with zonal wind jets studied
Discussion of variations on different timescales: seasonal, QBO, solar cycle
It is known that atmospheric dynamics in the tropical stratosphere have an influence on higher altitudes and latitudes as well as on surface weather and climate. In the tropics, the dynamics are ...governed by an interplay of the quasi-biennial oscillation (QBO) and semiannual oscillation (SAO) of the zonal wind. The QBO is dominant in the lower and middle stratosphere, and the SAO in the upper stratosphere/lower mesosphere. For both QBO and SAO the driving by atmospheric waves plays an important role. In particular, the role of gravity waves is still not well understood. In our study we use observations of the High Resolution Dynamics Limb Sounder (HIRDLS) satellite instrument to derive gravity wave momentum fluxes and gravity wave drag in order to investigate the interaction of gravity waves with the SAO. These observations are compared with the ERA-Interim reanalysis. Usually, QBO westward winds are much stronger than QBO eastward winds. Therefore, mainly gravity waves with westward-directed phase speeds are filtered out through critical-level filtering already below the stratopause region. Accordingly, HIRDLS observations show that gravity waves contribute to the SAO momentum budget mainly during eastward wind shear, and not much during westward wind shear. These findings confirm theoretical expectations and are qualitatively in good agreement with ERA-Interim and other modeling studies. In ERA-Interim most of the westward SAO driving is due to planetary waves, likely of extratropical origin. Still, we find in both observations and ERA-Interim that sometimes westward-propagating gravity waves may contribute to the westward driving of the SAO. Four characteristic cases of atmospheric background conditions are identified. The forcings of the SAO in these cases are discussed in detail, supported by gravity wave spectra observed by HIRDLS. In particular, we find that the gravity wave forcing of the SAO cannot be explained by critical-level filtering alone; gravity wave saturation without critical levels being reached is also important.
Land cover and its change are crucial for many environmental applications. This study focuses on the land cover classification and change detection with multitemporal and multispectral Sentinel-2 ...satellite data. To address the challenging land cover change detection task, we rely on two different deep learning architectures and selected pre-processing steps. For example, we define an excluded class and deal with temporal water shoreline changes in the pre-processing. We employ a fully convolutional neural network (FCN), and we combine the FCN with long short-term memory (LSTM) networks. The FCN can only handle monotemporal input data, while the FCN combined with LSTM can use sequential information (multitemporal). Besides, we provided fixed and variable sequences as training sequences for the combined FCN and LSTM approach. The former refers to using six defined satellite images, while the latter consists of image sequences from an extended training pool of ten images. Further, we propose measures for the robustness concerning the selection of Sentinel-2 image data as evaluation metrics. We can distinguish between actual land cover changes and misclassifications of the deep learning approaches with these metrics. According to the provided metrics, both multitemporal LSTM approaches outperform the monotemporal FCN approach, about 3 to 5 percentage points (p.p.). The LSTM approach trained on the variable sequences detects 3 p.p. more land cover changes than the LSTM approach trained on the fixed sequences. Besides, applying our selected pre-processing improves the water classification and avoids reducing the dataset effectively by 17.6%. The presented LSTM approaches can be modified to provide applicability for a variable number of image sequences since we published the code of the deep learning models. The Sentinel-2 data and the ground truth are also freely available.
Enhanced tropospheric trace gases such as CO, CH4 and H2O and reduced stratospheric O3 were measured in situ in the lowermost stratosphere over northern Europe on 26 September 2012 during the TACTS ...aircraft campaign. The measurements indicate that these air masses clearly differ from the stratospheric background. The calculation of 40-day backward trajectories with the trajectory module of the CLaMS model shows that these air masses are affected by the Asian monsoon anticyclone. Some air masses originate from the boundary layer in Southeast Asia/West Pacific and are rapidly lifted (1–2 days) within a typhoon up to the outer edge of the Asian monsoon anticyclone. Afterwards, the air parcels are entrained by the anticyclonic circulation of the Asian monsoon. The subsequent long-range transport (8–14 days) of enhanced water vapour and pollutants to the lowermost stratosphere in northern Europe is driven by eastward transport of tropospheric air from the Asian monsoon anticyclone caused by an eddy shedding event. We found that the combination of rapid uplift by a typhoon and eastward eddy shedding from the Asian monsoon anticyclone is a novel fast transport pathway that may carry boundary emissions from Southeast Asia/West Pacific within approximately 5 weeks to the lowermost stratosphere in northern Europe.
Global model data from the European Centre for Medium-Range Weather Forecasts (ECMWF) are analyzed for resolved gravity waves (GWs). Based on fitted 3-D wave vectors of individual waves and using the ...ECMWF global scale background fields, backward ray tracing from 25 km altitude is performed. Different sources such as orography, convection and winter storms are identified. It is found that due to oblique propagation waves spread widely from narrow source regions. Gravity waves which originate from regions of strong convection are frequently excited around the tropopause and have in the ECMWF model low phase and group velocities as well as very long horizontal wavelengths compared to other models and to measurements. While the total amount of momentum flux for convective GWs changes little over season, GWs generated by storms and mountain waves show large day-to-day variability, which has a strong influence also on total hemispheric fluxes; from one day to the next the total hemispheric flux may increase by a factor of 3. Implications of these results for using the ECMWF model in predicting, analyzing and interpreting global GW distributions as well as implications for seamless climate prediction are discussed.
The Asian summer monsoon provides an important pathway of tropospheric source gases and pollution into the lower stratosphere. This transport is characterized by deep convection and steady upwelling, ...combined with confinement inside a large-scale anticyclonic circulation in the upper troposphere and lower stratosphere (UTLS). In this paper, we show that a barrier to horizontal transport along the 380 K isentrope in the monsoon anticyclone can be determined from a local maximum in the gradient of potential vorticity (PV), following methods developed for the polar vortex (e.g., Nash et al., 1996). The monsoon anticyclone is dynamically highly variable and the maximum in the PV gradient is weak, such that additional constraints are needed (e.g., time averaging). Nevertheless, PV contours in the monsoon anticyclone agree well with contours of trace gas mixing ratios (CO, O3) and mean age from model simulations with a Lagrangian chemistry transport model (CLaMS) and satellite observations from the Microwave Limb Sounder (MLS) instrument. Hence, the PV-based transport barrier reflects the separation between air inside the core of the anticyclone and the background atmosphere well. For the summer season 2011 we find an average PV value of 3.6 PVU for the transport barrier in the anticyclone on the 380 K isentrope.
We analyze the variability of mean age of air (AoA) and of the local effects of the stratospheric residual circulation and eddy mixing on AoA within the framework of the isentropic zonal mean ...continuity equation. AoA for the period 1988–2013 has been simulated with the Lagrangian chemistry transport model CLaMS driven by ERA‐Interim winds and diabatic heating rates. Model simulated AoA in the lower stratosphere shows good agreement with both in situ observations and satellite observations from Michelson Interferometer for Passive Atmospheric Sounding, even regarding interannual variability and changes during the last decade. The interannual variability throughout the lower stratosphere is largely affected by the quasi‐biennial‐oscillation‐induced circulation and mixing anomalies, with year‐to‐year AoA changes of about 0.5 years. The decadal 2002–2012 change shows decreasing AoA in the lowest stratosphere, below about 450 K. Above, AoA increases in the Northern Hemisphere and decreases in the Southern Hemisphere. Mixing appears to be crucial for understanding AoA variability, with local AoA changes resulting from a close balance between residual circulation and mixing effects. Locally, mixing increases AoA at low latitudes (40°S–40°N) and decreases AoA at higher latitudes. Strongest mixing occurs below about 500 K, consistent with the separation between shallow and deep circulation branches. The effect of mixing integrated along the air parcel path, however, significantly increases AoA globally, except in the polar lower stratosphere. Changes of local effects of residual circulation and mixing during the last decade are supportive of a strengthening shallow circulation branch in the lowest stratosphere and a southward shifting circulation pattern above.
Key Points
Good agreement of simulated age and trend with MIPAS observations
Local residual circulation and mixing effects on mean age are opposite
QBO strongly affects mean age interannual variability globally
It is an outstanding issue to what degree trends in stratospheric mean age of air reflect changes in the (slow) residual circulation and how they are affected by (fast) eddy mixing. We present a ...method to quantify the effects of mixing and residual circulation on mean age trends, based on simulations with the Chemical Lagrangian Model of the Stratosphere (CLaMS) driven by ERA‐Interim reanalysis and on the integrated tracer continuity equation. During 1990–2013, mean age decreases throughout most of the stratosphere, qualitatively consistent with results based on climate model simulations. During 2002–2012, age changes show a clear hemispheric asymmetry in agreement with satellite observations. We find that changes in the residual circulation transit time cannot explain the mean age trends, and including the integrated effect of mixing is crucial. Above about 550 K (about 22 km), trends in the mixing effect on mean age appear to be coupled to residual circulation changes.
Key Points
Reconcile long‐term mean age decrease and observed decadal variability
Mixing effect is crucial to explain mean age trends
Mixing effect on mean age is partly coupled to the residual circulation