 |
|
|
|
|
|
| Earth Projects |
|
| Massively Parallel Imagery Assimilation Using the 3D Multiscale Multicomponent Modeling Framework (MMMF) (PI: Palaniappan, K,
University of Missouri-Columbia
) | |
| This proposal will increase by several orders of magnitude the use of EOS data in numerical weather cloud resolving models at the
synoptic and global scales by extracting EOS satellite image-derived atmospheric parameters using high-throughput parallel stream
processing algorithms implemented on inexpensive graphics processing units (GPUs). Rapid image processing will be coupled to a
massively parallel Grid-based process to assimilate image derived data into the Multiscale Multicomponent Modeling Framework
(MMMF) that combines for the first time cloud resolving models within a global circulation model for hurricane studies. The objective
of this proposal involves two components. One is to develop new computational image analysis and computer vision techniques for
satellite wind and microphysics extraction using highly parallelized implementations on inexpensive GPUs (graphics processing units)
for rapid processing of large volumes of satellite imagery. The second is to assimilate atmospheric fields such as wind and microphysics
parameters into the MMMF for hurricane studies, on NASA Grid-clusters, using a multiscale approach that embeds the fine resolution
Goddard Cumulus Ensemble (GCE) for cloud scale parameters within the coarse resolution Goddard finite volume Global Circulation
Model (fvGCM) for global coupling of dynamic quantities. Approximately 13,104 GCE models are embedded, one each per 2x2.5
degree grid element of the fvGCM and code parallelization using the Message Passing Interface (MPI) shows extremely encouraging
speedups and scalability. A massively parallel version of the 3D GCE model using MPI for the 10,240 processor Columbia
supercomputer at NASA Ames based on the current 512 CPU implementation when operational will be used in this project. A testbed
system that integrates experimental atmospheric wind and cloudiness extraction from multiple satellite imagery data with multiscale
modeling using the MMMF will be evaluated using hurricane data from the recent intense hurricane seasons. We propose new
approaches for semi-fluid cloud tracking, estimating cloud structure from multiview multispectral imagery, robust tensor-based optical
flow, adaptive data thinning and spline-based 3D scene flow motion estimation. The proposed novel computational approaches using
both inexpensive GPUs and large scale Grid computing based clusters will be utilized in this project for satellite image analysis and
data assimilation to evaluate impact on hurricane prediction and seasonal forecasts using the MMMF running on multi-resolution
spatial grids.
| |
| Enhancing GrADS, GDS and Greta to Support Multi-Member, Multi-Model and Multi-Format Earth System Science Data (PI: Doty, Brian,
Institute of Global Environment & Society, Inc
) | |
| The Grid Analysis and Display System (GrADS) is an interactive desktop tool used for the analysis and display of data. GrADS
handles both gridded and in situ data in a variety of formats. The GrADS Data Server (GDS) is an OPeNDAP server that provides
remote data access, subsetting, and server-side analysis for all the data formats and data types that GrADS supports. Greta is a
web-based search engine that collects metadata from any number of GrADS Data Servers and creates a searchable master catalogue
for those servers. GrADS, GDS, and Greta were developed and are maintained at the Center for Ocean-Land-Atmosphere Studies
(COLA). These open source tools provide a broad range of capabilities, including data discovery, metadata search, distributed data
access and analysis, and end-user display. The flexible interfaces to these tools provide rich opportunities for future software
integration. Together, these tools form a core cyberinfrastructure for scientific investigation of the Earth's weather and climate. The
initial implementation of GrADS was supported by the NASA Applied Information Systems Research (AISR) program in the early
1990s. Since then, GrADS has become the data analysis and display tool of choice in the meteorological, oceanographic and land
surface modeling, simulation and analysis research communities. GrADS and GDS are used worldwide by thousands of researchers
and educators for distributed access, analysis, and display of Earth science data. Earth system modelers at NASA Goddard
(Laboratory for Atmospheres and Global Modeling and Assimilation Office) have taken great advantage of these tools, recognizing in
2004 the ¿huge impact on our science¿ that GrADS has made. We now propose to the NASA AISR program to enhance GrADS, the
GDS, and Greta to enable these software components to support improved interactive analysis, batch data processing, and visualization
of data sets of critical importance to NASA¿s Earth science mission. The proposed enhancements will increase the software's access
and analysis capabilities, increase their scope to include Graphical Information System (GIS) applications, and enhance the user
interface to improve the integration of data search and discovery. The proposed software enhancements will directly serve the goals of
the AISR program, particularly increasing the productivity of NASA Science Mission Directorate research activities and fostering
interdisciplinary collaborations across the Earth sciences. The proposed work directly addresses the area of interest of the AISR
program that includes enhancements to persistent software tools that assist productivity of users of NASA¿s high-end computing
resources, specifically those who are engaged in Earth system modeling.
| |
| Improving Remote Sensed Data Products Using Bayesian Methodology for the Analysis of Computer Model Output (contd) (PI: Morris, Robin,
Universities Space Research Association
) | |
| The NASA Science Directorate supports an array of Earth Observing satellites which provide global coverage, whose observations are
used to produce a wide array of data products, from sea surface temperature, to polar ice coverage, to plant type and growth rates.
These are used in a wide variety of further scientific studies, and also as inputs to important policy decisions, especially those
concerning the impact of human activity on the biosphere. Many of the studies of human impact concern changes over time. To
accurately characterize changes it is vitally important that the uncertainties in the estimates of the quantities being observed are
known, so that the uncertainty in the estimated changes can be accurately determined -- making scientific or policy decisions based on
estimates with large, or worse, unknown or poorly determined errors, is poor science and poor policy. Many of the data products (eg
Leaf Area Index (LAI), Photosynthetically Active Radiation (fPAR) from MODIS) are produced by inverting a Radiative Transfer
Model (RTM), which simulates the upwelling radiation at the top of the atmosphere (and so observed by the satellite) as a function of
the biospheric parameters (e.g. land cover type; available water; leaf chemistry; etc.). These RTMs are implemented as complex
computer codes, and the analysis and inversion of these codes is a challenging task. In the last several years the area of Bayesian
Methods for the Analysis of Computer Model Output has made great progress, and is coming to a point where its wider application
will show significant utility in application domains. These methods are well developed in the statistics literature, but are almost
unknown in the Geoscience/Remote Sensing domain (apart from Kriging). Applied to an RTM, these methods will allow the
determination of: a) the uncertainty in the RTM output; b) the main effects, ie, which of the inputs is mainly responsible for the output
uncertainty; c) validation using field data; d) rapid approximation of the RTM for use when computing the inverse; e) a direct model
for the inverse incorporating uncertainty. Advances in these areas will improve the accuracy and utility of the data products. The RTM
in operational use for estimating LAI and fPAR is the MOD15 algorithm. We will initially apply the methodology to the LCM
(leaf-canopy model) RTM, a similar model to which we have more direct access. The methods and code developed will be readily
applicable to MOD15 and other RTMs -- treated as a "black boxes" by the methodology. We will use a number of Bayesian
nonparametric methods to approximate the LCM, including Gaussian Processes, and Dirichlet Process Mixtures, to study the
robustness of the analysis to different models. The models will be built using runs of the LCM. First we will perform a sensitivity
analysis to determine which inputs have most effect on the variability of the output, and so determine where more information is
required about the distributions of the inputs. We will use the approximate model to enable practical inversion of the LCM, to
determine the distribution over LAI given measured reflectances. We will incorporate field data, and build a model of the bias between
the LCM and the field data. This will result in a fully calibrated inversion procedure. A successful conclusion of the work in this
proposal would demonstrate the utility of the methodology in an important application domain, with significant scientific and
policymaking consequences, and would significantly advance the state-of-the-practice in remote sensing science.
| |
| Predicting the Perfect Storm: Feature Identification and State Estimation for Nonlinear Systems (PI: Turmon, Michael,
Jet Propulsion Laboratory
) | |
| We propose to develop new methods for representing and propagating statistical distributions that will encode multiple competing
hypotheses about weather system state. We allow the distributions representing states to undergo nonlinear evolution as time unfolds,
using known data to learn and track configurations of weather features. Furthermore, we propose to provide a conditional forecasting
capability, enabled by operating not just at the conventional level of grid cells but using interpretable features like fronts. By
conditioning on certain outcomes of particular concern (e.g., hurricane landing points), state trajectories likely to produce these
outcomes can be derived, and their relative likelihood computed to aid risk assessment. This conditioning capability is far more
efficient than current ensemble methods, especially in evaluating low-probability (but high-cost) outcomes. We will show how to link
the dense, gridded data NASA gathers into dynamic feature models comprehensible to humans, a capability NASA does not have. This
line of work could revolutionize atmospheric and oceanic forecasting capabilities by addressing directly the predictability of distinct
flow features, such as hurricanes, fronts and eddies, versus the approach of whole-field forecasting currently used in numerical
weather prediction. We offer a conditional forecast technology far more controllable and efficient than current ensemble-based
methods, and present plans to test it by identifying a low-probability formation that would be overlooked by conventional prediction
methods.
| |
| MTool: An Interactive Parallel Visualization and Computational Analysis Environment for Massive Multidimensional Earth Science Data Sets (PI: Lo, Martin,
Jet Propulsion Laboratory
) | |
| Global climate change is a major concern today. NASA's Strategic Subgoal 3A is a direct response to this concern: Study planet Earth
from space to advance scientific understanding and meet societal needs. However, a key technology challenge to a deeper
understanding of our climate is the massive amounts of data from SMD's many Earth Science missions and programs. A major
impediment is the high dimensionality of the data (up to hundreds of variables) which have spatial and temporal context that must be
maintained in order to understand their relationships to the underlying physical processes. Traditional GIS and visualization tools
typically preserve high-dimensional relationships at the expense of spatio-temporal ones. We propose to build an advanced analysis
and visualization tool called MTool (Multidimensional Tool), to address these issues using a new approach based on the intrinsic
geometry of the physical processes and mathematics behind the Earth science data using dynamical systems theory, computational
topology and discrete differential geometry. It is now well recognized that topology and geometry play significant roles in data analysis.
DARPA has instituted a new "Topological Data Analysis Program" to specifically address this new field. The GIS community is using
topological data structures to enable fast, new search algorithms that cannot be done with traditional Relational Data Base
Management Systems. In 2002, UNESCO even developed an on-line tutorial to train its users on "topological data search algorithms"
because of their power and efficiency. Similarly, computational differential geometry has made great advances in computer graphics,
computer vision, image processing, modeling and simulations. E.g., in 2005, Adobe Photoshop introduced the popular "Healing Brush"
feature which can remove cracks in paintings with the push of a button. This feature used the sophisticated concept of covariant
derivatives on a fiber bundle from differential geometry. This is the type of new technology which we propose for MTool. Some of our
methods include: manifold theory, dynamical systems theory (including invariant manifolds and its time-varying version, Lagrangian
coherent structures), discrete exterior calculus (to perform differential geometric calculations on manifolds), homology theory (to
analyze and 'visualize' high dimensional data), and the powerful level set methods (to gracefully handle topological changes in high
dimensional objects). Our objective is to investigate these new technologies and implement a working proof-of-concept tool, MTool
Version 0.3, in an interactive user-extensible parallel computational environment with visualization and animation capabilities. This is
achieved with the parallel version of Matlab on JPL's 1000-node Dell supercomputer, COSMOS. Co-I Braverman is a member on both
the Atmosphere Infrared Sounder and Multi-angle Imaging Spectro Radiometer projects at JPL. She will enable us to work with
actual users for requirements and for frequent user-reviews (every 4 months) needed in our Evolutionary Fusion development process.
This was successfully used by the PI on the 3-year, $2.5M LTool software development delivered in 2001. Our technology partner is the
award winning Caltech Barr Lab and Applied Geometry Center which pioneered the use of differential geometry and topology in
computer graphics and modeling. MTool is a 3-year project at a total cost of $772,700. MTool will introduce some of the newest and
most promising technologies for high dimension massive data to SMD missions and programs. These methods are useful to both
engineering and scientific users in SMD, from the many levels of data processing to the scientific analysis and visualization of the data.
This will help SMD not only reach Subgoal 3A to study the Earth from space, but NASA's many other goals in planetary science,
astrophysics, and the exploration of space.
| |
| Improving Remote Sensed Data Products Using Bayesian Methodology for the Analysis of Computer Model Output (PI: Morris, Robin,
USRA-RIACS
) | |
| The NASA Science Directorate supports an array of Earth Observing satellites which provide global coverage, and whose observations are used to produce a wide array of data products, from measurements of sea surface temperature, to polar ice coverage, to plant type and growth rates. These data products are used in a wide variety of further scientific studies, and also as inputs to important policy decisions, especially those concerning the impact of human activity on the biosphere.
Many of the studies of human impact concern changes over time. To accurately characterize changes it is vitally important that the uncertainties in the estimates of the quantities being observed are known, so that the uncertainty in the estimated changes can be accurately determined -- making scientific or policy decisions based on estimates with large, or worse, unknown or poorly determined errors, is poor science and poor policy.
Many of the data products (e.g. Leaf Area Index, Net Primary Production, Photosynthetically Active Radiation from MODIS (Moderate Resolution Imaging Spectroradiometer)) are produced by inverting a Radiative Transfer Model (RTM), which simulates the upwelling radiation at the top of the atmosphere (and so observed by the satellite) as a function of the biospheric parameters (e.g. land cover type; available water; leaf chemistry; etc.). These RTMs are implemented as complex computer codes, and the analysis and inversion of these codes is a challenging task.
In the last several years the area of Bayesian Methods for the Analysis of Computer Model Output has made great progress, and is coming to a point where its wider application will show significant utility in the application domains. These methods are well developed in the statistics literature, but are almost unknown in the Geoscience/Remote Sensing domain (with the exception of Kriging). Applied to an RTM, these methods will allow the determination of: a) the uncertainty in the RTM output; b) the main effects, that is, which of the inputs is mainly responsible for the output uncertainty; c) validation using field data; d) rapid approximation of the RTM for use when computing the inverse; e) a direct model for the inverse incorporating uncertainty. Advances in these areas will improve the accuracy and utility of the data products.
The RTM in operational use for estimating LAI (leaf area index) and fPAR (fraction of photosynthetically active radiation) is the MOD15 algorithm. We propose initially to apply the methodology to the LCM (leaf-canopy model) RTM, a similar model to which we have more direct access. The methods and code developed will be readily applicable to MOD15 and other RTMs -- the RTM is treated as a "black box" by the methodology.
Initially, a Gaussian Process (GP) model will be built using runs of the LCM. Sensitivity analysis using the GP model approximation will be performed to determine where more information is required about the distributions of the input variables. The fast computations available with the GP model will enable practical inversion of the LCM, to determine the distribution over LAI given measured reflectances. Computing the distribution over LAI for each observation using the forward model as described above may not be sufficiently computationally cheap to implement over large areas. We will build a second GP model that directly models the inverse problem, using training data generated from the full inverse. This will model the relationship between satellite observations and the probability distribution over LAI, and will demonstrate a new generic approach to characterizing the uncertainty in inverse problems.
A successful conclusion of the work in this proposal would demonstrate the utility of the methodology in an important application domain, with significant scientific and policymaking consequences, and would significantly advance the state-of-the-practice in remote sensing science.
| |
| Global cyclone detection and tracking (GLYDER) for climate variability studies: A multisensor data processing approach (PI: Talukder, Ashit,
NASA JPL
) | |
| We will implement an automated data processing and pattern recognition system, termed GLYDER (GLobal cYclone DEtection and tRacking), that classifies and tracks cyclones in multi-sensor satellite data sets. This system will be used to quantify the spatial and temporal variability of historical cyclones and their tracks globally and thereby supports a key goal of NASA's Earth Science Enterprise (ESE) to quantify the Earths natural variability and understand how it is changing. Cyclone detection with high accuracy using only single data modalities is an unsolvable problem, largely due to variations in the types of cyclones, and the presence of other weather patterns that appear similar within the sensor measurements. Incorporating multiple sensor types together provides several cues that can better characterize cyclones.
The joint temporal-spatial structure of cyclones has not been exploited in previous efforts and we believe that using such constraints will significantly improve detection accuracy over NCEP analysis, while reducing false alarms. In GLYDER, we will demonstrate the first multimodal cyclone detection scheme from multiple remote sensors with varying spatio-temporal resolution. We will develop and test a software system that will co-register multiple remote sensory data (GOES, QuickScat, AVHRR, SSMI and AMSR-E) that are at different spatial and temporal resolutions and detect and track cyclones from co-registered multiple remote sensors using advanced proven pattern recognition and tracking capabilities. Our goal in GLYDER is to co-register & detect cyclones at a spatial resolution of 10 kms and a temporal resolution of 3 hrs. We will use advanced multiscale wavelet techniques to co-register disparate multiple satellite datasources at different temporal and spatial resolutions. We will then use proven pattern recognition and data mining Maximum Discriminant Function Classifiers (MDF) combined with robust feature tracking algorithms on the remote multisensor co-registered data to detect cyclones and then track them temporally as they evolve over time. Based on previous performance of our detection and tracking algorithms on other datasets, we estimate that GLYDER will enable at least a 50% improvement in automated cyclone detection/classification from remote sensors with a 4X reduction in false alarms compared to current NCEP-based single sensor cyclone detection. We will initially train, test and validate our algorithms on regions that have multiple in-situ measurements and also experience frequent and intense storms and hurricanes (tropical eastern Atlantic and northeast pacific). Finally we will test and demonstrate our system with data from the Southern Oceans.
The GLYDER system will empower scientists with the capability to detect global cyclones in long time series of data. GLYDER also has the potential for use in an operational environment for early detection of tropical storms. Another arena in which GLYDER can play an important role is that of knowledge search and discovery. Current NASA's GCMD satellite data search support only simple searches for satellite measurements in time & space only. To facilitate advanced content-based searches, the NASA PO.DAAC would use GLYDER to mine its data holdings for the global cyclones and then tag the metadata of each granule with properties related to the cyclone for use in systems such as ECHO.
As part of the NASA AISR program, GLYDER will be utilized by:
1) JPL climate scientists to study and quantify the spatial and temporal variability of cyclones and their tracks (a critical need that has been identifies by the Intergovernmental Panel on Climate Change);
2) NASA's Physical Oceanography Distributed Active Archive Center (PO.DAAC) to tag metadata with information pertaining to cyclones and enable content-based searching.
| |
| Monitoring and Diagnosis of Complex Software and Hardware for Earth Observing Missions (PI: Williams, Brian,
MIT
) | |
| The Autonomous Sciencecraft Experiment (ASE) onboard the Earth Observing One (EO-1) mission has demonstrated the potential of autonomous systems to maximize scientific return. As complex software systems are being developed to control space assets and optimize onboard operations, there is a growing need for verification and validation of these systems. However, verification is traditionally performed offline during design and development, and does not guarantee a safeguard from all possible system failures. To complement offline verification techniques and ensure extremely high reliability operations, we propose to develop an onboard model-based fault management system to monitor and diagnose complex software and hardware systems, and track the progress of high-level mission objectives, in the context of the EO-1 mission. Our proposed technology will extend previous hardware diagnosis engines, such as Livingstone, to mixed hardware and software systems in several ways. First, it will monitor embedded software and diagnose software anomalies to enable robust execution and maximum science return. Second, monitoring software state will be used for refining the diagnosis of hardware components. Finally, the proposed engine will be capable of diagnosing in the presence of delayed symptoms, for the general case of mixed hardware and software systems. We will demonstrate the capability of the new model-based fault management by diagnosing ASE software and the progress of the high-level scientific goals. To assure the authenticity of the demonstration, we will integrate this fault management system with the ASE software and test by simulating the actual software anomalies detected during the execution of ASE onboard the EO-1 mission. We additionally propose an optional 12 month extension to flight validate the capability on EO-1. This proposal directly responds to the Applied Information Systems Research (AISR) program objectives of NASA's Research Opportunities in Space and Earth Sciences (ROSES). In particular, our technology is expected to enhance the science productivity of NASA's space flight missions that are sponsored by the Science Mission Directorate (SMD). The proposal will build upon the success of the ASE onboard the EO-1 mission, by providing an onboard capability for monitoring and diagnosing software and hardware systems, based on lessons learned from the ASE. Enhancing the ASE software through the proposed fault management capability will enable extremely high reliability operations, resulting in an increased return of scientific data. This proposal also responds to NASA's Strategic National Objective to �Study the Earth system from space and develop new space-based and related capabilities for this purpose.� The maturation and validation of our proposed technology in the context of EO-1 will demonstrate its potential for long term impact on many future NASA missions that are increasingly relying on complex software and hardware systems.
| |
|
|
|
|
|
