Numerous Earth-observing satellites provide high-resolution and high-volume data that facilitate scientific inference on physical and environmental processes. Most remote sensing data products used for scientific investigations are often subject to multiple stages of processing before they are widely used, and the scientific utility of these data products critically depends on a comprehensive assessment of the sources of uncertainty encountered in these stages of processing. One key stage involves the use of a retrieval algorithm to infer a geophysical quantity of interest from a satellite’s observed intensity of radiation.
The retrieval is an inverse problem that has been implemented mathematically and computationally in numerous ways for different satellite missions. Several of the presentations in this mini-symposium will each highlight an individual Earth-observing satellite and its retrieval methodology, emphasizing important contributions to uncertainty in retrieval data products. Methodological developments that interrogate the joint distribution of true geophysical states, retrieved states, and observed satellite spectra will be introduced. The presentations will span multiple Earth science applications, including weather and climate, the carbon cycle, air quality, atmospheric chemistry, and ecosystem health.
14:00
Uncertainty quantification for NASA’s Microwave Limb Sounder
Maggie Johnson | Jet Propulsion Laboratory, California Institute of Technology | United States
Show details
Authors:
Maggie Johnson | Jet Propulsion Laboratory, California Institute of Technology | United States
Nathaniel Livesey | Jet Propulsion Laboratory, California Institute of Technology | United States
Amy Braverman | Jet Propulsion Laboratory, California Institute of Technology | United States
NASA's Microwave Limb Sounder (MLS) has been collecting data on the chemistry and dynamics of the upper troposphere, stratosphere, and mesosphere since its launch aboard EOS-Aura in July 2004. MLS scans the "forward" limb, and ground data processing software retrieves vertical profiles of temperature, water vapor, and other constituents, along the Aura orbit track. Sets of individual retrievals are performed simultaneously in 15-degree "chunks" along the orbit using all along-track sequences of soundings belonging to the chunk and in a surrounding along-track buffer region. The current MLS retrieval algorithm uses standard optimal estimation methodology as one component of an uncertainty characterization procedure for individual retrievals. However, these uncertainty estimates do not account for potential dependencies in uncertainty across space (geographically) or due to the state of the atmosphere. In this talk, we describe our approach to quantifying not only the individual per-sounding uncertainties but also across-sounding, spatially-dependent uncertainties. The latter are essential for estimating uncertainties in quantities derived from multiple individual retrievals, such as long-term trends, spatial gradients, etc.
14:30
Objective frequentist uncertainty quantification for atmospheric CO2 retrieval
Pratik Patil | Carnegie Mellon University | United States
Show details
Authors:
Pratik Patil | Carnegie Mellon University | United States
Mikael Kuusela | Carnegie Mellon University | United States
Jonathan Hobbs | Jet Propulsion Laboratory, California Institute of Technology | United States
The steadily increasing level of atmospheric CO2 since the beginning of the Industrial Revolution is affecting the global climate and threatening the long-term sustainability of Earth's ecosystem. In order to better understand the geographic distribution of CO2 and to track its sources and sinks on regional scales through time, NASA operates a dedicated satellite called Orbiting Carbon Observatory-2 (OCO-2) to monitor CO2 from space. OCO-2 makes radiance measurements of the sunlight reflected off the Earth's surface in different spectral bands, which are then inverted to obtain CO2 estimates. In this work, we first analyze the current operational retrieval algorithm, which uses prior knowledge in the form of distributions over various state variables to regularize the underlying ill-posed inverse problem, and find evidence of estimation bias and under-coverage in the uncertainty estimates both at individual locations and over a spatial region. To alleviate these issues, we then propose a method that only uses actual physical constraints on the state variables to regularize the problem and constructs well-calibrated confidence intervals for functionals of the CO2 profile based on convex programming. Furthermore, we study the influence of individual state variables on the length of the intervals and identify key elements that can greatly reduce the uncertainty given additional deterministic or probabilistic constraints.
15:00
Applications of spatial-statistical models to UQ for Earth observing missions: ECOSTRESS and SBG
Kerry Cawse-Nicholson | Jet Propulsion Laboratory, California Institute of Technology | United States
Show details
Authors:
Kerry Cawse-Nicholson | Jet Propulsion Laboratory, California Institute of Technology | United States
Amy Braverman | Jet Propulsion Laboratory, California Institute of Technology | United States
Emily Kang | University of Cincinnati | United States
Miaoqi Li | University of Cincinnati | United States
Water management, agricultural irrigation, drought forecasting, and a myriad of science investigations rely on remotely sensed evapotranspiration (ET) data. However, most ET products are provided without any measures of uncertainty on the retrieved estimates. ECOSTRESS, a thermal instrument mounted on the International Space Station (ISS), produces ET products using two different algorithms. We developed a spatial-statistical model to generate ensembles of statistical realizations of ET input fields in a way that preserves both spatial and inter-variable dependence. The ensemble members are then input to the ET models to generate ensembles of retrieved fields, which can be compared to the original ET field to quantify uncertainty. We find that: 1) the spatial-statistical model results in more efficient uncertainty estimates than a naive white-noise model; and, 2) the non-linear ET models result in retrieval errors at a pixel level that are highly variable and non-Guassian. This means that traditional precision and accuracy are inadequate metrics of uncertainty in ET, which will have impacts on uncertainty reporting, and in mission formulation. Surface Biology and Geology (SBG) is a future satellite mission, which also has goals to measure – amongst other things – global hydrological cycles and water resources. Because the mission is still in design phase, uncertainty quantification can play a part in evaluating different instrument architectures.
15:30
Accelerated MCMC for remote sensing of atmospheric CO2
Otto Lamminpää | Finnish Meteorological Institute | Finland
Show details
Author:
Otto Lamminpää | Finnish Meteorological Institute | Finland
Proper monitoring and prediction of climate change and global warming requires global, satellite-based measurements of atmospheric greenhouse gas concentrations with an unprecedented accuracy. This is an ill-posed and non-linear statistical inverse problem, to which Markov Chain Monte Carlo methods offer a rigorous and reliable means of uncertainty quantification and accuracy assessment. In this work, we implement adaptive MCMC together with dimension reduction to NASA’s Orbiting Carbon Observatory 2 satellite’s atmospheric carbon dioxide concentration measurements.