Tim Hultberg

A Global Assessment of NASA AIRS v6 and EUMETSAT IASI v6 Precipitable Water Vapor using Ground-based GPS SuomiNet Stations†

Satellite remote sensing of Precipitable Water Vapor (PWV) is essential for monitoring moisture in real-time for weather applications, as well as tracking the long-term changes in PWV for climate change trend detection. This study assesses the accuracies of the current satellite observing system, specifically the National Aeronautics and Space Administration (NASA) Atmospheric Infrared Sounder (AIRS) v6 PWV product and the European Organization for the Exploitation of Meteorological Satellite Studies (EUMETSAT) Infrared Atmospheric Sounding Interferometer (IASI) v6 PWV product, using Ground-Based SuomiNet Global Positioning System (GPS) network as truth. Elevation-corrected collocated matchups to each SuomiNet GPS station in North America and around the world was created and results were broken down by station, ARM-region, climate zone, and latitude zone. The greatest difference, exceeding 5%, between IASI and AIRS retrievals occurred in the tropics. Generally, IASI and AIRS fall within a 5% error in the PWV range of 20-40 mm (a mean bias less than 2 mm), with a wet bias for extremely low PWV values (less than 5 mm) and a dry bias for extremely high PWV values (greater than 50 mm). The operational IR satellite products are able to capture the mean PWV but degrade in the extreme dry and wet regimes.

FLOPC++ An Algebraic Modeling Language Embedded in C++

FLOPC++ is an open source algebraic modeling language implemented as a C++ class library. It allows linear optimization problems to be modeled in a declarative style, similar to algebraic modeling languages, such as GAMS and AMPL, within a C++ program. The project is part of COmputational INfrastructure for Operations Research (COIN-OR) and uses its Open Solver Interface (OSI) to achieve solver independence.

Identification and intercomparison of surface‐based inversions over Antarctica from IASI, ERA‐Interim, and Concordiasi dropsonde data

Surface based temperature inversions (SBIs) occur frequently over Antarctica and play an important role in climate and weather. Antarctic SBIs are examined during Austral spring, 2010 using measurements from dropsondes, ERA-Interim Atmospheric Reanalysis Model, and the recently released version 6 of the Infrared Atmospheric Sounding Interferometer (IASI) level 2 product. A SBI detection algorithm is applied to temperature profiles from these datasets. The results will be used to determine if satellite and reanalysis products can accurately characterize SBIs and if so, then they may be used to study SBIs outside of the spring 2010 study period. From the dropsonde data, SBIs occurred in 20% of profiles over sea ice and 54% of profiles over land. IASI and ERA-Interim surface air temperatures are found to be significantly warmer than dropsonde observations at high plateau regions, while IASI surface air temperature is colder over sea ice. IASI and ERA-Interim have a cold bias at nearly all levels above the surface when compared to the dropsonde. SBIs are characterized by their frequency, depth, and intensity. It is found that SBIs are more prevalent, deeper, and more intense over the continent than over sea ice, especially at higher surface elevations. Using IASI and ERA-Interim data the detection algorithm has a high probability of detection of SBIs but is found to severely overestimate the depth and underestimate the intensity for both data sets. These over- and underestimations are primarily due to the existence of extremely shallow inversion layers that neither satellite nor reanalysis products can resolve.

Stochastic Extensions to FlopC

We extend the open-source modelling language FlopC++, which is part of the COIN-OR project, to support multi-stage stochastic programs with recourse. We connect the stochastic version of FlopC++ to the existing COIN class stochastic modelling interface (SMI) to provide a direct interface to specialized solution algorithms. The new stochastic version of FlopC++ can be used to specify scenario-based problems and distribution-based problems with independent random variables. A data-driven scenario tree generation method transforms a given scenario fan, a collection of different data paths with specified probabilities, into a scenario tree. We illustrate our extensions by means of a two-stage mixed integer strategic supply chain design problem and a multi-stage investment model.

Validation of the IASI temperature and water vapor profile retrievals by correlative radiosondes

The METOP-A satellite Infrared Atmospheric Sounding Interferometer (IASI) Level 2 products comprise retrievals of vertical profiles of temperature and water vapor. The L2 data were validated through assessment of their error covariances and biases using radiosonde data for the reference. The radiosonde data set includes dedicated launches as well as the ones performed at regular synoptic times at Lindenberg station (Germany). For optimal error estimate the linear statistical Validation Assessment Model (VAM) was used. The model establishes relation between the compared satellite and reference measurements based on their relations to the true atmospheric state. The VAM utilizes IASI averaging kernels and statistical characteristics of the ensembles of the reference data to allow for finite vertical resolution of the retrievals and spatial and temporal non-coincidence. For temperature retrievals expected and assessed errors are in good agreement; error variances/rms of a single FOV retrieval are 1K between 800 - 300 mb with an increase to ~1K in tropopause and ~2K at the surface, possibly due to wrong surface parameters and undetected clouds/haze. Bias against radiosondes oscillates within ±0 5K . between 950 - 100 mb. As for water vapor, its highly variable complex spatial structure does not allow assessment of retrieval errors with the same degree of accuracy as for temperature. Error variances/rms of a single FOV relative humidity retrieval are between 10 - 13% RH in the 800 - 300 mb range.

Local or global? How to choose the training set for principal component compression of hyperspectral satellite measurements: a hybrid approach

Principal Component (PC) compression is the method of choice to achieve band-width reduction for dissemination of hyper spectral (HS) satellite measurements and will become increasingly important with the advent of future HS missions (such as IASI-NG and MTG-IRS) with ever higher data-rates. It is a linear transformation defined by a truncated set of the leading eigenvectors of the covariance of the measurements as well as the mean of the measurements. We discuss the strategy for generation of the eigenvectors, based on the operational experience made with IASI. To compute the covariance and mean, a so-called training set of measurements is needed, which ideally should include all relevant spectral features. For the dissemination of IASI PC scores a global static training set consisting of a large sample of measured spectra covering all seasons and all regions is used. This training set was updated once after the start of the dissemination of IASI PC scores in April 2010 by adding spectra from the 2010 Russian wildfires, in which spectral features not captured by the previous training set were identified. An alternative approach, which has sometimes been proposed, is to compute the eigenvectors on the fly from a local training set, for example consisting of all measurements in the current processing granule. It might naively be thought that this local approach would improve the compression rate by reducing the number of PC scores needed to represent the measurements within each granule. This false belief is apparently confirmed, if the reconstruction scores (root mean square of the reconstruction residuals) is used as the sole criteria for choosing the number of PC scores to retain, which would overlook the fact that the decrease in reconstruction score (for the same number of PCs) is achieved only by the retention of an increased amount of random noise. We demonstrate that the local eigenvectors retain a higher amount of noise and a lower amount of atmospheric signal than global eigenvectors. Local eigenvectors do not increase the compression rate, but increase the amount of atmospheric loss and should be avoided. Only extremely rare situations, resulting in spectra with features which have not been observed previously, can lead to problems for the global approach. To cope with such situations we investigate a hybrid approach, which first apply the global eigenvectors and then apply local compression to the residuals in order to identify and disseminate in addition any directions in the local signal, which are orthogonal to the subspace spanned by the global eigenvectors.

Removal of instrument artefacts by harmonisation of hyperspectral sensor data from multiple detectors

IASI has 4 different detectors, CrIS has 9, IASI-NG will have 16 and MTG-IRS 25600. There is a clear interest to harmonise the sensor data originating from different detectors, if it can be done be removing the parts of the instrument artefacts, which are not common to all detectors. When IASI spectra are analysed in principal component (PC) score space, differences between the four detectors are clearly observed. These differences are caused by different characteristics and different strengths of the ghost effect among the detectors and although they are small when analysed in radiance space, they can have a distinct negative impact on the use of the data. Considering that a large part of the operationally disseminated IASI PC scores are dominated by instrument artefacts, the partial removal of instrument artefacts is also of interest for data compression purposes. The instrument artefacts can be partly removed by projection onto a subspace common to all detectors. We show how the techniques of canonical angles can be used to compute a set of orthogonal vectors capturing only directions which are close to directions found in the signal spaces of all detectors. This principle can also be applied to detectors on-board different satellites, as we demonstrate with the example of IASI-A and IASI-B. The danger of the method is that a single deficient detector, ’blind’ to one or more directions of the atmo- spheric signal, could potentially ’contaminate’ the data from the other detectors. We discuss how to detect and avoid this problem and check it in practice with CrIS data.