Engineering topics
Publications and source records attributed to Crawford, James.
Clouds directly affect tropospheric photochemistry through modification of solar radiation that determines photolysis frequencies. This effect is an important component of global tropospheric chemistry-climate interaction, and its understanding is thus essential for predicting the feedback of climate change on tropospheric chemistry.
Radiative effect of clouds is one of the major factors that affect tropospheric OH. Large differences in cloud distributions among current (chemistry-climate or chemical transport) models could contribute significantly to the wide model spread of tropospheric OH, which was reported by the ACCMIP activity (Voulgarakis et al., ACP 2013). CCCM, a 3-D cloud data product developed at NASA Langley and merged from multiple A-Train satellite observations, provides unprecedentedly strong constraints on the vertical distribution of clouds and therefore simulated effects of clouds on key tropospheric oxidants.
The first field campaign of DISCOVER-AQ (Deriving Information on Surface conditions from COlumn and VERtically resolved observations relevant to Air Quality) took place in July 2011 over Baltimore-Washington Corridor (BWC). A suite of airborne remote sensing and in-situ sensors was deployed along with ground networks for mapping vertical and horizontal distribution of aerosols. Previous researches were based on a single lidar station because of the lack of regional coverage. This study uses the unique airborne HSRL (High Spectral Resolution Lidar) data to baseline PM2.5 (particulate matter of aerodynamic diameter less than 2.5 μm) estimates and applies to regional air quality with satellite AOD (Aerosol Optical Depth) retrievals over BWC (∼6500 sq. km). The linear approximation takes into account aerosols aloft above AML (Aerosol Mixing Layer) by normalizing AOD with haze layer height (i.e., AOD/HLH). The estimated PM2.5 mass concentrations by HSRL AOD/HLH are shown within 2 RMSE (Root Mean Square Error ∼9.6 μg/cu. m) with correlation ∼0.88 with the observed over BWC. Similar statistics are shown when applying HLH data from a single location over the distance of 100 km. In other words, a single lidar is feasible to cover the range of 100 km with expected uncertainties. The employment of MPLNET-AERONET (MicroPulse Lidar NETwork - AErosol RObotic NETwork) measurements at NASA GSFC produces similar statistics of PM2.5 estimates as those derived by HSRL. The synergy of active and passive remote sensing aerosol measurements provides the foundation for satellite application of air quality on a daily basis. For the optimal range of 10 km, the MODIS-estimated PM2.5 values are found satisfactory at 27 (out of 36) sunphotometer locations with mean RMSE of 1.6-3.3 μg/cu. m relative to PM2.5 estimated by sunphotometers. The remaining 6 of 8 marginal sites are found in the coastal zone, for which associated large RMSE values ∼4.5-7.8 μg/cu. m are most likely due to overestimated AOD because of water-contaminated pixels.
The first deployment of the Earth Venture -1 DISCOVER-AQ (Deriving Information on Surface conditions from Column and Vertically Resolved Observations Relevant to Air Quality) project was conducted during July 2011 in the Baltimore-Washington region. Two aircraft (a P-3B for in-situ sampling and a King Air for remote sensing) were used along with an extensive array of surface-based in-situ and remote sensing instrumentation. Fourteen flight days were accomplished by both aircraft and over 250 profiles of trace gases and aerosols were performed by the P-3B over surface air quality monitoring stations, which were specially outfitted with sunphotometers and Pandora UV/Vis spectrometers. The King Air flew with the High Spectral Resolution Lidar for aerosols and the ACAM UV/Vis spectrometer for trace gases. This suite of observations allows linkage of surface air quality with the vertical distributions of gases and aerosols, with remotely-sensed column amounts observed from the surface and from the King Air, and with satellite observations from Aura (OMI and TES), GOME-2, MODIS and GOES. The DISCOVER-AQ data will allow determination of under what conditions satellite retrievals are indicative of surface air quality, and they will be useful in planning new satellites. In addition to an overview of the project, a preliminary comparison of tropospheric column NO2 densities from the integration of in-situ P-3B observations, from the Pandoras and ACAM, and from the new Goddard OMI NO2 algorithm will be presented.
JavaGenes is a general-purpose, evolutionary software system written in Java. It implements several versions of a genetic algorithm, simulated annealing, stochastic hill climbing, and other search techniques. This software has been used to evolve molecules, atomic force field parameters, digital circuits, Earth Observing Satellite schedules, and antennas. This version differs from version 0.7.28 in that it includes the molecule evolution code and other improvements. Except for the antenna code, JaveGenes is available for NASA Open Source distribution.
This paper is concerned with the problem of allocating a unit capacity resource to multiple users within a pre-defined time period. The resource is indivisible, so that at most one user can use it at each time instance. However, different users may use it at different times. The users have independent, se@sh preferences for when and for how long they are allocated this resource. Thus, they value different resource access durations differently, and they value different time slots differently. We seek an optimal allocation schedule for this resource. This problem arises in many institutional settings where, e.g., different departments, agencies, or personal, compete for a single resource. We are particularly motivated by the problem of scheduling NASA's Deep Space Satellite Network (DSN) among different users within NASA. Access to DSN is needed for transmitting data from various space missions to Earth. Each mission has different needs for DSN time, depending on satellite and planetary orbits. Typically, the DSN is over-subscribed, in that not all missions will be allocated as much time as they want. This leads to various inefficiencies - missions spend much time and resource lobbying for their time, often exaggerating their needs. NASA, on the other hand, would like to make optimal use of this resource, ensuring that the good for NASA is maximized. This raises the thorny problem of how to measure the utility to NASA of each allocation. In the typical case, it is difficult for the central agency, NASA in our case, to assess the value of each interval to each user - this is really only known to the users who understand their needs. Thus, our problem is more precisely formulated as follows: find an allocation schedule for the resource that maximizes the sum of users preferences, when the preference values are private information of the users. We bypass this problem by making the assumptions that one can assign money to customers. This assumption is reasonable; a committee is usually in charge of deciding the priority of each mission competing for access to the DSN within a time period while scheduling. Instead, we can assume that the committee assigns a budget to each mission.This paper is concerned with the problem of allocating a unit capacity resource to multiple users within a pre-defined time period. The resource is indivisible, so that at most one user can use it at each time instance. However, different users may use it at different times. The users have independent, se@sh preferences for when and for how long they are allocated this resource. Thus, they value different resource access durations differently, and they value different time slots differently. We seek an optimal allocation schedule for this resource. This problem arises in many institutional settings where, e.g., different departments, agencies, or personal, compete for a single resource. We are particularly motivated by the problem of scheduling NASA's Deep Space Satellite Network (DSN) among different users within NASA. Access to DSN is needed for transmitting data from various space missions to Earth. Each mission has different needs for DSN time, depending on satellite and planetary orbits. Typically, the DSN is over-subscribed, in that not all missions will be allocated as much time as they want. This leads to various inefficiencies - missions spend much time and resource lobbying for their time, often exaggerating their needs. NASA, on the other hand, would like to make optimal use of this resource, ensuring that the good for NASA is maximized. This raises the thorny problem of how to measure the utility to NASA of each allocation. In the typical case, it is difficult for the central agency, NASA in our case, to assess the value of each interval to each user - this is really only known to the users who understand their needs. Thus, our problem is more precisely formulated as follows: find an allocation schedule for the resource that maximizes the sum ofsers preferences, when the preference values are private information of the users. We bypass this problem by making the assumptions that one can assign money to customers. This assumption is reasonable; a committee is usually in charge of deciding the priority of each mission competing for access to the DSN within a time period while scheduling. Instead, we can assume that the committee assigns a budget to each mission.
This paper is concerned with the problem of allocating a unit capacity resource to multiple users within a pre-defined time period. The resource is indivisible, so that at most one user can use it at each time instance. However, different users may use it at different times. The users have independent, selfish preferences for when and for how long they are allocated this resource. Thus, they value different resource access durations differently, and they value different time slots differently. We seek an optimal allocation schedule for this resource.
Scheduling observations by coordinated fleets of Earth Observing Satellites (EOS) involves large search spaces, complex constraints and poorly understood bottlenecks, conditions where evolutionary and related algorithms are often effective. However, there are many such algorithms and the best one to use is not clear. Here we compare multiple variants of the genetic algorithm: stochastic hill climbing, simulated annealing, squeaky wheel optimization and iterated sampling on ten realistically-sized EOS scheduling problems. Schedules are represented by a permutation (non-temperal ordering) of the observation requests. A simple deterministic scheduler assigns times and resources to each observation request in the order indicated by the permutation, discarding those that violate the constraints created by previously scheduled observations. Simulated annealing performs best. Random mutation outperform a more 'intelligent' mutator. Furthermore, the best mutator, by a small margin, was a novel approach we call temperature dependent random sampling that makes large changes in the early stages of evolution and smaller changes towards the end of search.
We hypothesize that evolutionary algorithms can effectively schedule coordinated fleets of Earth observing satellites. The constraints are complex and the bottlenecks are not well understood, a condition where evolutionary algorithms are often effective. This is, in part, because evolutionary algorithms require only that one can represent solutions, modify solutions, and evaluate solution fitness. To test the hypothesis we have developed a representative set of problems, produced optimization software (in Java) to solve them, and run experiments comparing techniques. This paper presents initial results of a comparison of several evolutionary and other optimization techniques; namely the genetic algorithm, simulated annealing, squeaky wheel optimization, and stochastic hill climbing. We also compare separate satellite vs. integrated scheduling of a two satellite constellation. While the results are not definitive, tests to date suggest that simulated annealing is the best search technique and integrated scheduling is superior.
Earth observing satellite (EOS) scheduling is a complex real-world domain representative of a broad class of over-subscription scheduling problems. Over-subscription problems are those where requests for a facility exceed its capacity. These problems arise in a wide variety of NASA and terrestrial domains and are .XI important class of scheduling problems because such facilities often represent large capital investments. We have run experiments comparing multiple variants of the genetic algorithm, hill climbing, simulated annealing, squeaky wheel optimization and iterated sampling on two variants of a realistically-sized model of the EOS scheduling problem. These are implemented as permutation-based methods; methods that search in the space of priority orderings of observation requests and evaluate each permutation by using it to drive a greedy scheduler. Simulated annealing performs best and random mutation operators outperform our squeaky (more intelligent) operator. Furthermore, taking smaller steps towards the end of the search improves performance.
We describe work in progress concerning multi-instrument, multi-satellite scheduling. Most, although not all, Earth observing instruments currently in orbit are unique. In the relatively near future, however, we expect to see fleets of Earth observing spacecraft, many carrying nearly identical instruments. This presents a substantially new scheduling challenge. Inspired by successful commercial applications of evolutionary algorithms in scheduling domains, this paper presents work in progress regarding the use of evolutionary algorithms to solve a set of Earth observing related model problems. Both the model problems and the software are described. Since the larger problems will require substantial computation and evolutionary algorithms are embarrassingly parallel, we discuss our parallelization techniques using dedicated and cycle-scavenged workstations.