Shortcuts aren't the problem. Recognizing the common shortcuts,and allowing incorporation of others would cure that.
The problem arises when someone's not astute enough to convey their meaning in clear, grammatically correct form. I've a lot less trouble with with 'BRB' than '1337 5p34k', and less with '1337 5p34k' than with stupid grammatical errors, lazy spelling errors and syntactical miscues because someone is too lazy (or, submit lame excuse here) to have learned how to write a coherent passage.
I spend a lot of time with math and science. I have to be able to express myself to my peers, the public and to students. If I were to lapse into the mindset of the "changing, evolving language" and argue that errors were a form of expression, I'd not, generally, be able to do that. Just as we encourage our engineers to be precise in how they represent their designs and analyses, they have to be able to document what they're doing, or what they've done, or their work won't be usable.
There have been a couple of mainstream articles on the topic of the evolution of English (or whatever we have now) toward an IM-variant. The experts I've seen quoted tend to agree that the language is morphing toward a more streamlined version,and that it's not necessarily bad, although, for us old guys, it's a little disconcerting.
I have, similary, watched the change in the Russian language over the last 20 years. When I studied in Leningrad in the early '70's, I learned a variant of the language... almost a dialect... that was barely understood in Moscow. This year, I had the opportunity to speak to a new faculty member who's from Petersburg... Leningrad... and I found he was considerably less formal and had a lot more foreign word investment in his conversant mode than I'd expected. Didn't make him harder to understand, necessarily, but certainly different.
It's not just the American version of English that is changing, but, I suspect, almost every viable, employed language.
When my daughter was in grade school, middle school, junior-high and high school, I also offered to help with homework, at least when I wasn't travelling. She declined, usually, because I didn't "do it like the teacher showed us" how to complete their work. That said, several times, I found *I* couldn't understand what the teacher had purportedly taught, and in at least a couple of instances, I found the hand-outs from the teacher were patently incorrect.
We used the computer as a reward for some period of time. We still keep the 2 primary computers in the family room, so parents can keep an eye on what the kids are doing/surfing, and so we can answer questions before they make a wrong assumption on homework before it's committed to paper and a battle ensues to make 'em correct their mistake. I think that's helped. What's also helped is having the knowledge and tools to provide some unobtrusive filters and redirection on the network side when I've suspected the kids were wandering into the unseemly side of the 'Net, or when their behavior suggested they were heading down a pathway we felt was dangerous.
I've got one in college, doing well even if I think she's got more potential than her grades show (simply, I think she'd have better grades if she saw less than about 10 movies a week); one in high school doing well, and almost a poster child for 'Net-based school research, and one in elementary school who is learning how to use and leverage the 'Net in a manner his siblings may never understand.
Managed properly, I think having the computers around is a good thing. I'm just not sure I've learned how to manage them properly just yet!
I've gotta say: If this idea had come along when my daughter was discovering IM, it would have helped her spelling/grammar. She got through high school with good grades, but has discovered in college that her writing was subject to more critique and perhaps she wasn't as good as she had thought. Or, more precisely, she's decided a lot of the professors are idiots because they've criticized her grammar, spelling and composition style.
Note: Solectek advertises 72 Mb/s; as I read their ad, they're not exactly 802.16, but either close, or upgradable when the "real" chipsets come available. Several other vendors (Redline Communications...) claim to have 802.16 compatible hardware, but they've been "waiting to ship" for almost a year, and say "sometime next year" when asked. I suspect they're sitting there with a product but the lack of a standard until relatively recently, and the lack of FCC info, have led to a product that's non-compliant in the service for now.
Actually, my evaluation of the Perfigo hardware in a set of controlled, and "in the wild" tests at TAMU sound they were reliable and worked well. I suspect the note on the alkaid.org site, saying the on-campus implementation was flawed, was closer to the truth.
Trying to add a little bit of sanity to this line...
Producing the data are expensive. Often, as well, the aerial imagery companies will retain ownership of the images (often not photos anymore) or ramp the costs of the service and imagery 'way up beyond what the city or state can afford. There's historical precedent to this, back when most of us didn't care or want those data...
He's asking for the whole database. Likely, if it's a reasonably designed GIS database, there's data of a tax/ownership nature that shouldn't be released electronically... if at all. There are some things about my taxes I don't see a reason for you to know, and if they're included therein (and they might be in a "reasonable" but not necessarily in a "good" design) then request was out of line.
In Texas, all GIS data derived with public funds but not restricted by contractual obligations are released as public data, or available from the various agencies upon request. (http://www.tnris.state.tx.us/)
This may change with restrictions and recommendations from the Feds bout reducing access to critical infrastructure data. For a variety of reasons, I can go either way on this. although I'm currently the "data wants to be free" guy in that duscussion.
That said, some of the GIS data we have in Texas on critial infrastructure and critical industries DOES come pretty close to qualifying for "due diligence" on the part of a terrorist. They'd get all the needed to mine the bridge, or do maximum damage to the chemical plant. Should we make it easy?
Finally, on the costs associated with requesting "free" data from state agencies: I've seen the numbers and have gotten the patient explanations on why they're so high. Let's say a CD-RW disk is $.25. Then you have to have a GIS analyst retrieve the data and place it in the burn directory. If it were something like, "Send me the whole database" this is relatively easy. Then you have to have someone burn the CD. Or CDs. The agency, at least in Texas, is required by State law to recover costs using a formula that incorporates the direct and indirect costs of the individuals doing the work, on a per-hour basis, shipping, and a depreciation allowance for the equipment, again prorated. A little bit here, a little bit there, eventually the CD costs $75, which was what TNRIS charged last time I went there rather than downloading the data directly...
What you get from YOUR GPS is an autonomous position estimate.
What I get from my array of GPS receivers is a set of observables, including the pseudorange from the code phase measurements, and carrier phase information. When I get done processing, I have a set of baselines, a set of potential delays in the troposphere and ionosphere, doppler shifts for the satellites in view, timing deltas, and information on signal multipath propagation.
When I report something out, I can give horizontal positions with respect to a known coordinate system to about 1 cm (2sigma RMS) and vertical estimates (referenced to the WGS84 ellipsoid) on the order of 2 cm (2sigma RMS). Given a good gravity model, I can estimate geoid height differences to about 3 cm.
Given code phase observables from a satellite-borne receiver, I can make reasonable estimates of the satellite's ephemerides and state vector, provided 3 sets with a temporal spread of about 20 minutes. Given code and carrier data I can derive the state vector in less time and with better accuracy.
I can estimate the amount of precipitable water in a vertical column of atmosphere above a GPS receiver site.
I can make estimates of the ionosphere's total electron count.
There's more.
For the record, I've been doing all of the above, for several years.
To some of us, it's an incredible deal. It's gonna save me a lot of work and development time. I've done a fair bit of what they've released, but it's usually hacks, technically correct but neither reusable nor pretty.
Now I get to use their work, integrate it with mine, feed results back into the effort, and every one of us looking at TEC, wet delay (GPS-derived precipitable water vapor), physical geodesy (think resolving residuals on a survey baseline), satellite geodesy (Where's the GPS Satellite now?), navigation, RF propagation, geez, all sorts of disciplines, now benefit.
Actually, that's Gig 'em, Aggies! The really awesome engineers in Austin graduated from about 90 miles to the east and migrated toward better parties, higher pay, and really cool projects like GPSTK!
When you place the heart at rest, you remove the load it sees, and use an artificial pump to do the work.
You don't stop the heart. Even today, with cardioplegic solutions significantly advanced, supplemented with NAD-compounds and amino acids, stopping the heart bears the significant risk that you can't get it restarted again.
Cardioplegia for cardiac surgery involves infusing a potassium-rich solution into the coronary arteries, which stops the heart in diastole. Further, the solution is cold, and the heart is bathed in an iced saline slush to cool it further and diminish its metabolic requirements. At this point, the heart is *NOT* getting a blood rich perfusion (barring the use of blood-based cadioplegia, which I'm still not sure is as good an idea as some others think) medium.
If you were to start reperfusing the arrested heart with blood, with a normal electrolyte composition, the extra potassium would be washed out, the heart would rewarm, and if it has sufficient energy stores, and a sufficiently normal physiology, it would begin to contract again.
So: To put the heart at rest, you unload it, keep the blood chemistry as normal as possible, maintain good nutrition status (parenteral alementation), and see if the heart muscle recovers.
Well, having actually spent some time doing partial and total artificial heart research, including about 6 years with the earlier LVAD (pulsatile) technology, I'm of two minds about this.
1. I don't believe we'll see increased atherosclerotic plaque deposition due to non-pulsatile flow. I'm currently subscribing to the theory that plaque is related in chronic bacterial infection of the vascular intima. 2. There was some evidence, but poorly followud up in the past, that renal function was on the short list of critical elements requiring pulsatile flow. One reason for inadequate continuing studies was that the problems with on-pump anticoagulation, infection and anesthesia tended to introduce enough variables to make isolation of the pulse issue too obscure. 3. There have been a number of reports in the past, some as long as 15 years ago, of surgeons using the Biomedicus BioPump, a similar design, for extracorporeal bridging support to transplantation, or similar to the anecdote below, to place the heart at rest to allow its recovery. I'm aware of many cases (I participated in at least 10) of multiple-day attempts, and at least 2 or 3 trials of several weeks. Realize that the patients were already moribund, so supporting them in bridging, awaiting a suitable donor, was their last and sole chance for survival. And, no, most of them didn't survive to transplant.
I'm intrigued. I'm out of the business now, but I'm convinced that we're overdue for some real breakthrough to make implantable artificial devices for continuous perfusion a viable alternative to transplantation with the limited pool of donors.
Just to add a little information, wince 1 MAY 2000, when Selective Availability was disabled, autonomous GPS using the L1 Coarse Acquisition code has had a typical error of 6m about 90% of the time. This is more than sufficient to track these guys both in terms of spatial positioning, and in terms of speed.
Use of EGNOS, correctly compared with WAAS, will enhance the position accuracy to about 1m error, but velocity accuracies won't be enhanced very much.
It'd help if the FUD about GPS had not been included in the ESA article.
Surface weather observations can be well delivered via an internet model, and aggregated. When it comes to assimilation and prediction, your statement becomes much more problemmatic.
In general, models today are gridded parcel physics computations or computational flow dymanics models. These models are best performed with low-latency computational systems as NUMA clusters.
Grid models tend to require processing of discreet elements or simple processes of nearly trivial complexity with no particular interest in whether the work gets done, when it gets done or if the data are reported back. Because of the interdependency of the bounding conditioned enforced in a rigidly gridded model, or the nature of a CFD model, losing a data parcel product would skew the overall results.
Looking at an example, SETI@home takes a chunk of data and performs an FFT, returning an integer result for each little piece of the chunk. These integers are repackaged as an array and eventually returned to the server. Later (much later) analysis allows the server to determine, from this pre-processing if the parcel of data originally analyzed had something of potential interest for reanalysis in a more rigorous fashion.
In weather prediction, the pieces have to come together in a timely manner, as a more distinct data product for integration into the product (output) grid. A missing piece adversely affects subsequent analysis or results in a hole in the data. This is bad.
It should also be noted that trying to get back floating point results from a diverse set of systems with different OS, processor and RAM configurations could provide some interesting results.
Typically, today, numerical weather prediction takes place in a serial fashion on computers designed for "normal" desktop use, over a small area, or with large grid sizes, or for speed and/or finer-grained resolution, on a cluster or formal supercomputer where message-passing allows the more normal method of loop-unrolling to be utilized, and where, rather than parcing out large data chunks and hoping for a result, smaller pieces are distributed to the nodes and they pass their results directly back into the central node.
It might be an interesting exercise to play with Globus message passing over a geographically distributed array of similar processors. Sounds like another research project...
Er... Level II data's available today. If you're interested in a single site, that's easy. If you're interested in a bunch, the bandwidth gets a bit hairy. My typical 5-min volume scan is 14MB/site...
While I happen to support the Fair Weather policy proposal, and at first blush really don't like the approach Accuweather is taking in the lead for this, realistically, they're just the front-men for the attack. It's the commercial weather industry that seeks to restrict access to the NOAA and NWS data, generally, rather than just one company.
A boycott against Accuweather won't accomplish much. It's an attractive idea, but they're not the only players in the industry.
Folks like AWS would prefer to tell you their school-nets are sufficient for climatologic monitoring. Their normal locations are not consistent with what we consider good practice, located atop buildings to keep prying little fingers from destroying the equipment easily, but they claim they have sufficient evidence that there's no significant bias in their readings, and that they should be paid by states to create mesonets using schools as their sites.
And then charge the states for the data, and charge the schools for the weather stations and periodic maintenance.
There are benefits to doing something like this. The concept of contractors instead of dedicated employees allows a governmental entity to cut apparent costs, and allows a private sector industry to thrive. It also tends to drive up costs, overall, because the private sector position will usually be at a higher FTE and overhead cost than the government sector will budget. But, that's OK: It's hidden.
On the positive side, there are economies of scale in having a company or companies do instrument installation and maintenance on a large scale. These may, in the long-run, make such deals work better.
So. Go boycott one firm that's not likely to get a lot of money from this group anyway, and that has a long history of providing good products to its customers. I don't think it'll have a significant impact. Did we accomplish anything?
CWOP lacks sufficient metadata to provide adequate information on instrument types, calibration, resolution or accuracy, never mind instrument placement in compliance with conventions, standards and norms.
It's a very valuable tool (I use it on my site) but one MUST understand the limitations. As it stands now, it's something we use professionally, but cannot rely on for climatological monitoring.
IF we can start getting the appropriate metadata, perhaps it will be more useful. The QC evaluations performed on CWOP by Patty Miller and Mike Barth at NOAA's Forecast Systems Laboratory provide an invaluable improvement to its usability for "real" work. They do not, however, mitigate the problems associated with siting and resolution.
The concept of a distributed network of sites, maintained by NOAA/NWS and volunteers, is embodied in the COOP program. NWS has a plan to replace all the old sensors with new sensors, enhance some of the instrumentation to extend beyond temperature and rain gauges (the current norm), and collect data electronically at 1 hr (and eventually 5 min) intervals. Coupled with professional guidance on siting and installation (and, yes, this really IS important for consistent data collection) and appropriate metadata, plus a competent QC program, we'll see some marked improvements in the amount and quality of data we recover.
There *are* sources of data beyond radar images...
The reason a lot of commercial enterprises pay for commercial weather data is to get the specialized products and service they want. A large corporate flight department might contract with Universal Wx for individualized forecast services for its pilots on-demand. A farmer might contract for evapotranspiration data for his region to support irrigation requirements. The list goes on.
NWS doesn't have a lot of folks to do the heavy lifting in the Weather Forecast Offices, nor, for that matter, in the Region HQs. It's become a much more lean organization than it used to be.
And, NO, I don't work for NWS. I have a fair bit of interaction, with the guys in the trenches, but I'm not in the Government's employ.
The Fair Weather policy strikes me as one of the most rational pieces of information to come out of NWS HQ in a long time. It reflects a better understanding of the capabilities of the Internet to disseminate timely information, than I've seen expressed in Silver Spring, MD, in a long time.
The Commercial Weather enterprises will survive. They probably will maintain their customer bases, assuming some new upstart doesn't come up and snag customers the old fashioned way: providing better services at lower costs. What Accuweather and its corporate bretheren would prefer is to keep their club small, and make sure we all have to pay for the data NWS has to collect anyway.
-- Could a PC (any OS) with a browser or other machine independant program be used to control all the Linux machines ala Distributed computing? --
Well, yes. We have a virtual network teaching laboratory. Save a couple of boxes for testing, all are PC's running Linux. Mostly RedHat, one Slack and a pair of GenToo.
Our interfaces to the lab (we have developed 2, and are moving one to distributed network management and configuration) are a Java implementation in one case, and a web-based PHP and C package with a PostGreSQL backend, in the other.
Students are presented with a CLI on the network hardware assigned for their exercise. In some cases, said hardware is Linux running the Zebra routing interface.
We teach network design and configuration, incorporating some testing capabilities to evaluate how well the config they did really works. We also teach an attack/defend network security class, mostly Linux, although we offer some shill Windows boxes and we recently reincorporated some Solaris systems. All this occurs in a sandbox environment; no one may directly attack from outside the sandbox.
We also use the linux systems for testing network hardware from the "real" world. We use a variety of systems to generate "real" traffic at reasonable bandwidths, allowing us to look at, for instance, at QoS capabilities. We use some other, proprietary software, running on Linux, for custom traffic generation at setable bandwidths.
While we tend to use "RedHat", we also tend to customize it significantly. We don't tend to make our customizations into RPMs; we do remove problemmatic RPMs and replace them with either custom tarballs or in some cases (Apache comes to mind) the latest tarballs. We recently refined our PXE-boot system loads to automate most of the install to a short process requiring little operator intervention.
So: Can we control networked hardware using Linux? I'd say the answer is, "YES!"
Pne has different levels of accuracy in the various positioning systems by using different means int terms of code sequence length and clocking speed to change the error budget. You also achieve a better code- and carrier-phase solution by mitigating atmospheric (ionoshpheric and tropospheric) influences (linear water vapor delay and ionospheric scintillation which manifests as diminished signal strength and non-linear delay) by using both the L1 and L2 signals.
Selective availability, a long-period dithering algorithm applied to clocks and code-phase signals, doesn't play into this.
And, from a security and engineering perspective, I'm pretty comfortable with the concept that Selective Availability would not have been discontinued unless another, more geographically agile/specific method of denying accuracies had not been developed. I don't know what it is or might be, and I don't have a need-to-know about it. But I'm pretty confident that we didn't give up a means of denial that doesn't have a larger-scale effect on safety of navigation like SA had.
Lots of folks have implemented dual-system receivers that incorporate GLONAS and GPS. Selective Availability was never incorporated into GLONAS. I've done some real-time surveying with GPS-GLONAS dual systems that were impressive in their ability to maintain lock even in areas where there was a lot of physical masking of the sky, because they had so many (14-16, for me) satellites in view at the same time.
Oh, and for the spatially challenged, GPS does NOT incorporate triangulation, but rather, trilateration.
OK. I really hate to be pedantic about this but let's talk.
1. When you talk about the 100m accuracies, 36m accuracies, 1m accuracies, 5 decimeter accuracies, 1 cm accuracies, etc. in general, you're talking about achievable reproducability with regard to "truth." More on this shortly.
2. Depending on how you account for the various error terms, the CURRENT system and standard allow for autonomous L1, C/A code-phase accuracies of about 29m.
3. Since Selective Availability was discontinued, I and others in the "game" have demonstrated routine L1 C/A code-phase autonomous accuracies of 6-10m, with most of the autonomous accuracies leaning toward 6,
4. The guys who determine error budgets are pessimists (God, I love those guys). They assume that all error terms are scalar additive values. So, they assume that there's no way an error term can be included in the error budget to your advantage. Thus, an error budget that is 6x larger than what I'm seeing regularly.
With a lot of work, long term observations, and use of a variety of hybrid GPS and conventional surve techniques, as well as using a number of sites with methods of ascertaining their relative positions to precise levels, without using GPS, we have been able to cross-correlate the international network of survey monumentation (more true in the CONUS than some other countries) to GPS positioning. Geodesists use network adjustment statistics (least-squares method) to minimize errors and use stable geometric networks to determine the coordinates in 2D or 3D space of an unknown point of interest. Densification of this network leads to a network that allows land surveyors to use either GPS or conventional monumentation with a degree of confidence.
When we refer to a 36m error budget, as the DoD spec does, it's a worst case scenario, and is consistent with the current signal specifications. As we enhance the GPS system with additional signals and methods, we should be able to refine the spec to reflect "reality" over time.
The 100m error budget mentioned refers to a 27-36m error budget for autonomous L1, C/A code phase systems, intentionally degraded using a pseudorandom algorithm called Selective Availability. This was a method of denying precise positioning to a potential enemy of the US. Over time, the integration of techniques, methods and systems to augment GPS autonomous positioning, as well as new techniques (I'm making a few assumptions now) that selective area degradation (denial of quality of service in selected regions/areas) as well as in-theatre jamming led to a US decision to terminate Selective Availability last May. With that, improved C/A positioning was obtained, with an almost instantaneous accuracy of 6m according to hose who were watching and looking for changes.
In short, the collective errors associated with troposphere (water vapor), ionospheric scintillation (solar influence) and refraction, those associated with signal multipath, Rayleigh propagation, and clock errors associated with both the user-receiver and those vaguaries on the spacecraft themselves... not to mention relativity, conspire to degrade the signal. The pessimists call it 36m. The engineers call it, currently, about 6m in my area (Texas; YMMV if you're close to the poles or near the South Atlantic Anomoly). And for those with some older receivers, you have to measure it and ascertain reality for yourselves.
There's no way you're going to reliably obtain a 1m autonomous GPS positional accuracy sans augmentation (WAAS, DGPS, etc). If you think you can, let's talk. I want to confirm your procedures and see if it's publishable.
Can we do better using carrier phase techniques? Sure. I can routinely use 2 receivers (or more, or the US NGS-housed CORS system) and get horizontal accuracies in a least-squares network adjustment on the order of 1cm (2 sigma). But that's not what you were implying.
Transdermal induction-based power supplies were first successfully introduced about 1985; before that they were bigger, bulkier, damaged the skin, had poor efficiency, and were just plain painful. Still, the amount of power transferrable across the intact skin, without damaging same, isn't likely to be enough to provide much juice for a continuously operating motor. Even at low power settings, the continuous portion of that equation is a battery-draining issue.
Further, batteries have to be non-venting and encased in a bio-neutral container. Last time I looked, that pretty well limited them to nickle-cadmium cells, or sealed, depleted electrolyte lead-acid cells. The lead-acid cells are too heavy for these purposes. The number of NiCd's one can package in an unobtrusive manner, and manage to parallel in a fashion that will provide good current-density, is still small. Further, despite advances in NiCd chemistry, there are still issues with "memory" (dendritic formation within the cells) and cycle life. I anticipate more battery work will ensue, especially if the physiology of this device in humans is borne out (most of the testing has been in calves; the issues of cell morphology and fragility are not identical!), then a device like this will spur significant research in associated fields... because the money AND the good will are there.
Slashdot Mirror
Shortcuts aren't the problem. Recognizing the common shortcuts,and allowing incorporation of others would cure that.
The problem arises when someone's not astute enough to convey their meaning in clear, grammatically correct form. I've a lot less trouble with with 'BRB' than '1337 5p34k', and less with '1337 5p34k' than with stupid grammatical errors, lazy spelling errors and syntactical miscues because someone is too lazy (or, submit lame excuse here) to have learned how to write a coherent passage.
I spend a lot of time with math and science. I have to be able to express myself to my peers, the public and to students. If I were to lapse into the mindset of the "changing, evolving language" and argue that errors were a form of expression, I'd not, generally, be able to do that. Just as we encourage our engineers to be precise in how they represent their designs and analyses, they have to be able to document what they're doing, or what they've done, or their work won't be usable.
There have been a couple of mainstream articles on the topic of the evolution of English (or whatever we have now) toward an IM-variant. The experts I've seen quoted tend to agree that the language is morphing toward a more streamlined version,and that it's not necessarily bad, although, for us old guys, it's a little disconcerting.
I have, similary, watched the change in the Russian language over the last 20 years. When I studied in Leningrad in the early '70's, I learned a variant of the language... almost a dialect... that was barely understood in Moscow. This year, I had the opportunity to speak to a new faculty member who's from Petersburg... Leningrad... and I found he was considerably less formal and had a lot more foreign word investment in his conversant mode than I'd expected. Didn't make him harder to understand, necessarily, but certainly different.
It's not just the American version of English that is changing, but, I suspect, almost every viable, employed language.
Ah gess y'all ain't frum round heah? (Texian dialect)
When my daughter was in grade school, middle school, junior-high and high school, I also offered to help with homework, at least when I wasn't travelling. She declined, usually, because I didn't "do it like the teacher showed us" how to complete their work. That said, several times, I found *I* couldn't understand what the teacher had purportedly taught, and in at least a couple of instances, I found the hand-outs from the teacher were patently incorrect.
We used the computer as a reward for some period of time. We still keep the 2 primary computers in the family room, so parents can keep an eye on what the kids are doing/surfing, and so we can answer questions before they make a wrong assumption on homework before it's committed to paper and a battle ensues to make 'em correct their mistake. I think that's helped. What's also helped is having the knowledge and tools to provide some unobtrusive filters and redirection on the network side when I've suspected the kids were wandering into the unseemly side of the 'Net, or when their behavior suggested they were heading down a pathway we felt was dangerous.
I've got one in college, doing well even if I think she's got more potential than her grades show (simply, I think she'd have better grades if she saw less than about 10 movies a week); one in high school doing well, and almost a poster child for 'Net-based school research, and one in elementary school who is learning how to use and leverage the 'Net in a manner his siblings may never understand.
Managed properly, I think having the computers around is a good thing. I'm just not sure I've learned how to manage them properly just yet!
I've gotta say: If this idea had come along when my daughter was discovering IM, it would have helped her spelling/grammar. She got through high school with good grades, but has discovered in college that her writing was subject to more critique and perhaps she wasn't as good as she had thought. Or, more precisely, she's decided a lot of the professors are idiots because they've criticized her grammar, spelling and composition style.
$4k using Solectek's version of the current technology. 802.16 is technically NLOS; it uses a 256-carrier OFDM constellation. It sorta lives for multipath. Check out http://www.solectek.com/products/bridges-and-route rs/prod-sm5kBH-feat.html for a look at their PtP solution.
Note: Solectek advertises 72 Mb/s; as I read their ad, they're not exactly 802.16, but either close, or upgradable when the "real" chipsets come available. Several other vendors (Redline Communications...) claim to have 802.16 compatible hardware, but they've been "waiting to ship" for almost a year, and say "sometime next year" when asked. I suspect they're sitting there with a product but the lack of a standard until relatively recently, and the lack of FCC info, have led to a product that's non-compliant in the service for now.
Actually, my evaluation of the Perfigo hardware in a set of controlled, and "in the wild" tests at TAMU sound they were reliable and worked well. I suspect the note on the alkaid.org site, saying the on-campus implementation was flawed, was closer to the truth.
Think "Marketing" and remember that hyperbole is their mainstay, and restraint by facts or physics is antithetical...
Trying to add a little bit of sanity to this line...
Producing the data are expensive. Often, as well, the aerial imagery companies will retain ownership of the images (often not photos anymore) or ramp the costs of the service and imagery 'way up beyond what the city or state can afford. There's historical precedent to this, back when most of us didn't care or want those data...
He's asking for the whole database. Likely, if it's a reasonably designed GIS database, there's data of a tax/ownership nature that shouldn't be released electronically... if at all. There are some things about my taxes I don't see a reason for you to know, and if they're included therein (and they might be in a "reasonable" but not necessarily in a "good" design) then request was out of line.
In Texas, all GIS data derived with public funds but not restricted by contractual obligations are released as public data, or available from the various agencies upon request. (http://www.tnris.state.tx.us/)
This may change with restrictions and recommendations from the Feds bout reducing access to critical infrastructure data. For a variety of reasons, I can go either way on this. although I'm currently the "data wants to be free" guy in that duscussion.
That said, some of the GIS data we have in Texas on critial infrastructure and critical industries DOES come pretty close to qualifying for "due diligence" on the part of a terrorist. They'd get all the needed to mine the bridge, or do maximum damage to the chemical plant. Should we make it easy?
Finally, on the costs associated with requesting "free" data from state agencies: I've seen the numbers and have gotten the patient explanations on why they're so high. Let's say a CD-RW disk is $.25. Then you have to have a GIS analyst retrieve the data and place it in the burn directory. If it were something like, "Send me the whole database" this is relatively easy. Then you have to have someone burn the CD. Or CDs. The agency, at least in Texas, is required by State law to recover costs using a formula that incorporates the direct and indirect costs of the individuals doing the work, on a per-hour basis, shipping, and a depreciation allowance for the equipment, again prorated. A little bit here, a little bit there, eventually the CD costs $75, which was what TNRIS charged last time I went there rather than downloading the data directly...
There will be a quiz next hour.
What you get from YOUR GPS is an autonomous position estimate.
What I get from my array of GPS receivers is a set of observables, including the pseudorange from the code phase measurements, and carrier phase information. When I get done processing, I have a set of baselines, a set of potential delays in the troposphere and ionosphere, doppler shifts for the satellites in view, timing deltas, and information on signal multipath propagation.
When I report something out, I can give horizontal positions with respect to a known coordinate system to about 1 cm (2sigma RMS) and vertical estimates (referenced to the WGS84 ellipsoid) on the order of 2 cm (2sigma RMS). Given a good gravity model, I can estimate geoid height differences to about 3 cm.
Given code phase observables from a satellite-borne receiver, I can make reasonable estimates of the satellite's ephemerides and state vector, provided 3 sets with a temporal spread of about 20 minutes. Given code and carrier data I can derive the state vector in less time and with better accuracy.
I can estimate the amount of precipitable water in a vertical column of atmosphere above a GPS receiver site.
I can make estimates of the ionosphere's total electron count.
There's more.
For the record, I've been doing all of the above, for several years.
To some of us, it's an incredible deal. It's gonna save me a lot of work and development time. I've done a fair bit of what they've released, but it's usually hacks, technically correct but neither reusable nor pretty.
Now I get to use their work, integrate it with mine, feed results back into the effort, and every one of us looking at TEC, wet delay (GPS-derived precipitable water vapor), physical geodesy (think resolving residuals on a survey baseline), satellite geodesy (Where's the GPS Satellite now?), navigation, RF propagation, geez, all sorts of disciplines, now benefit.
This is just 'way kewl!
Actually, that's Gig 'em, Aggies! The really awesome engineers in Austin graduated from about 90 miles to the east and migrated toward better parties, higher pay, and really cool projects like GPSTK!
When you place the heart at rest, you remove the load it sees, and use an artificial pump to do the work.
You don't stop the heart. Even today, with cardioplegic solutions significantly advanced, supplemented with NAD-compounds and amino acids, stopping the heart bears the significant risk that you can't get it restarted again.
Cardioplegia for cardiac surgery involves infusing a potassium-rich solution into the coronary arteries, which stops the heart in diastole. Further, the solution is cold, and the heart is bathed in an iced saline slush to cool it further and diminish its metabolic requirements. At this point, the heart is *NOT* getting a blood rich perfusion (barring the use of blood-based cadioplegia, which I'm still not sure is as good an idea as some others think) medium.
If you were to start reperfusing the arrested heart with blood, with a normal electrolyte composition, the extra potassium would be washed out, the heart would rewarm, and if it has sufficient energy stores, and a sufficiently normal physiology, it would begin to contract again.
So: To put the heart at rest, you unload it, keep the blood chemistry as normal as possible, maintain good nutrition status (parenteral alementation), and see if the heart muscle recovers.
Well, having actually spent some time doing partial and total artificial heart research, including about 6 years with the earlier LVAD (pulsatile) technology, I'm of two minds about this.
1. I don't believe we'll see increased atherosclerotic plaque deposition due to non-pulsatile flow. I'm currently subscribing to the theory that plaque is related in chronic bacterial infection of the vascular intima.
2. There was some evidence, but poorly followud up in the past, that renal function was on the short list of critical elements requiring pulsatile flow. One reason for inadequate continuing studies was that the problems with on-pump anticoagulation, infection and anesthesia tended to introduce enough variables to make isolation of the pulse issue too obscure.
3. There have been a number of reports in the past, some as long as 15 years ago, of surgeons using the Biomedicus BioPump, a similar design, for extracorporeal bridging support to transplantation, or similar to the anecdote below, to place the heart at rest to allow its recovery. I'm aware of many cases (I participated in at least 10) of multiple-day attempts, and at least 2 or 3 trials of several weeks. Realize that the patients were already moribund, so supporting them in bridging, awaiting a suitable donor, was their last and sole chance for survival. And, no, most of them didn't survive to transplant.
I'm intrigued. I'm out of the business now, but I'm convinced that we're overdue for some real breakthrough to make implantable artificial devices for continuous perfusion a viable alternative to transplantation with the limited pool of donors.
Just to add a little information, wince 1 MAY 2000, when Selective Availability was disabled, autonomous GPS using the L1 Coarse Acquisition code has had a typical error of 6m about 90% of the time. This is more than sufficient to track these guys both in terms of spatial positioning, and in terms of speed.
Use of EGNOS, correctly compared with WAAS, will enhance the position accuracy to about 1m error, but velocity accuracies won't be enhanced very much.
It'd help if the FUD about GPS had not been included in the ESA article.
Surface weather observations can be well delivered via an internet model, and aggregated. When it comes to assimilation and prediction, your statement becomes much more problemmatic.
In general, models today are gridded parcel physics computations or computational flow dymanics models. These models are best performed with low-latency computational systems as NUMA clusters.
Grid models tend to require processing of discreet elements or simple processes of nearly trivial complexity with no particular interest in whether the work gets done, when it gets done or if the data are reported back. Because of the interdependency of the bounding conditioned enforced in a rigidly gridded model, or the nature of a CFD model, losing a data parcel product would skew the overall results.
Looking at an example, SETI@home takes a chunk of data and performs an FFT, returning an integer result for each little piece of the chunk. These integers are repackaged as an array and eventually returned to the server. Later (much later) analysis allows the server to determine, from this pre-processing if the parcel of data originally analyzed had something of potential interest for reanalysis in a more rigorous fashion.
In weather prediction, the pieces have to come together in a timely manner, as a more distinct data product for integration into the product (output) grid. A missing piece adversely affects subsequent analysis or results in a hole in the data. This is bad.
It should also be noted that trying to get back floating point results from a diverse set of systems with different OS, processor and RAM configurations could provide some interesting results.
Typically, today, numerical weather prediction takes place in a serial fashion on computers designed for "normal" desktop use, over a small area, or with large grid sizes, or for speed and/or finer-grained resolution, on a cluster or formal supercomputer where message-passing allows the more normal method of loop-unrolling to be utilized, and where, rather than parcing out large data chunks and hoping for a result, smaller pieces are distributed to the nodes and they pass their results directly back into the central node.
It might be an interesting exercise to play with Globus message passing over a geographically distributed array of similar processors. Sounds like another research project...
Er... Level II data's available today. If you're interested in a single site, that's easy. If you're interested in a bunch, the bandwidth gets a bit hairy. My typical 5-min volume scan is 14MB/site...
While I happen to support the Fair Weather policy proposal, and at first blush really don't like the approach Accuweather is taking in the lead for this, realistically, they're just the front-men for the attack. It's the commercial weather industry that seeks to restrict access to the NOAA and NWS data, generally, rather than just one company.
A boycott against Accuweather won't accomplish much. It's an attractive idea, but they're not the only players in the industry.
Folks like AWS would prefer to tell you their school-nets are sufficient for climatologic monitoring. Their normal locations are not consistent with what we consider good practice, located atop buildings to keep prying little fingers from destroying the equipment easily, but they claim they have sufficient evidence that there's no significant bias in their readings, and that they should be paid by states to create mesonets using schools as their sites.
And then charge the states for the data, and charge the schools for the weather stations and periodic maintenance.
There are benefits to doing something like this. The concept of contractors instead of dedicated employees allows a governmental entity to cut apparent costs, and allows a private sector industry to thrive. It also tends to drive up costs, overall, because the private sector position will usually be at a higher FTE and overhead cost than the government sector will budget. But, that's OK: It's hidden.
On the positive side, there are economies of scale in having a company or companies do instrument installation and maintenance on a large scale. These may, in the long-run, make such deals work better.
So. Go boycott one firm that's not likely to get a lot of money from this group anyway, and that has a long history of providing good products to its customers. I don't think it'll have a significant impact. Did we accomplish anything?
CWOP lacks sufficient metadata to provide adequate information on instrument types, calibration, resolution or accuracy, never mind instrument placement in compliance with conventions, standards and norms.
It's a very valuable tool (I use it on my site) but one MUST understand the limitations. As it stands now, it's something we use professionally, but cannot rely on for climatological monitoring.
IF we can start getting the appropriate metadata, perhaps it will be more useful. The QC evaluations performed on CWOP by Patty Miller and Mike Barth at NOAA's Forecast Systems Laboratory provide an invaluable improvement to its usability for "real" work. They do not, however, mitigate the problems associated with siting and resolution.
The concept of a distributed network of sites, maintained by NOAA/NWS and volunteers, is embodied in the COOP program. NWS has a plan to replace all the old sensors with new sensors, enhance some of the instrumentation to extend beyond temperature and rain gauges (the current norm), and collect data electronically at 1 hr (and eventually 5 min) intervals. Coupled with professional guidance on siting and installation (and, yes, this really IS important for consistent data collection) and appropriate metadata, plus a competent QC program, we'll see some marked improvements in the amount and quality of data we recover.
There *are* sources of data beyond radar images...
The reason a lot of commercial enterprises pay for commercial weather data is to get the specialized products and service they want. A large corporate flight department might contract with Universal Wx for individualized forecast services for its pilots on-demand. A farmer might contract for evapotranspiration data for his region to support irrigation requirements. The list goes on.
NWS doesn't have a lot of folks to do the heavy lifting in the Weather Forecast Offices, nor, for that matter, in the Region HQs. It's become a much more lean organization than it used to be.
And, NO, I don't work for NWS. I have a fair bit of interaction, with the guys in the trenches, but I'm not in the Government's employ.
The Fair Weather policy strikes me as one of the most rational pieces of information to come out of NWS HQ in a long time. It reflects a better understanding of the capabilities of the Internet to disseminate timely information, than I've seen expressed in Silver Spring, MD, in a long time.
The Commercial Weather enterprises will survive. They probably will maintain their customer bases, assuming some new upstart doesn't come up and snag customers the old fashioned way: providing better services at lower costs. What Accuweather and its corporate bretheren would prefer is to keep their club small, and make sure we all have to pay for the data NWS has to collect anyway.
--
Could a PC (any OS) with a browser or other machine independant program be used to control all the Linux machines ala Distributed computing?
--
Well, yes. We have a virtual network teaching laboratory. Save a couple of boxes for testing, all are PC's running Linux. Mostly RedHat, one Slack and a pair of GenToo.
Our interfaces to the lab (we have developed 2, and are moving one to distributed network management and configuration) are a Java implementation in one case, and a web-based PHP and C package with a PostGreSQL backend, in the other.
Students are presented with a CLI on the network hardware assigned for their exercise. In some cases, said hardware is Linux running the Zebra routing interface.
We teach network design and configuration, incorporating some testing capabilities to evaluate how well the config they did really works. We also teach an attack/defend network security class, mostly Linux, although we offer some shill Windows boxes and we recently reincorporated some Solaris systems. All this occurs in a sandbox environment; no one may directly attack from outside the sandbox.
We also use the linux systems for testing network hardware from the "real" world. We use a variety of systems to generate "real" traffic at reasonable bandwidths, allowing us to look at, for instance, at QoS capabilities. We use some other, proprietary software, running on Linux, for custom traffic generation at setable bandwidths.
While we tend to use "RedHat", we also tend to customize it significantly. We don't tend to make our customizations into RPMs; we do remove problemmatic RPMs and replace them with either custom tarballs or in some cases (Apache comes to mind) the latest tarballs. We recently refined our PXE-boot system loads to automate most of the install to a short process requiring little operator intervention.
So: Can we control networked hardware using Linux? I'd say the answer is, "YES!"
Pne has different levels of accuracy in the various positioning systems by using different means int terms of code sequence length and clocking speed to change the error budget. You also achieve a better code- and carrier-phase solution by mitigating atmospheric (ionoshpheric and tropospheric) influences (linear water vapor delay and ionospheric scintillation which manifests as diminished signal strength and non-linear delay) by using both the L1 and L2 signals.
Selective availability, a long-period dithering algorithm applied to clocks and code-phase signals, doesn't play into this.
And, from a security and engineering perspective, I'm pretty comfortable with the concept that Selective Availability would not have been discontinued unless another, more geographically agile/specific method of denying accuracies had not been developed. I don't know what it is or might be, and I don't have a need-to-know about it. But I'm pretty confident that we didn't give up a means of denial that doesn't have a larger-scale effect on safety of navigation like SA had.
Javad/TopconGPS; Ashtech-Magellan; Trimble Navigation; Leica; Novatel.
Lots of folks have implemented dual-system receivers that incorporate GLONAS and GPS. Selective Availability was never incorporated into GLONAS. I've done some real-time surveying with GPS-GLONAS dual systems that were impressive in their ability to maintain lock even in areas where there was a lot of physical masking of the sky, because they had so many (14-16, for me) satellites in view at the same time.
Oh, and for the spatially challenged, GPS does NOT incorporate triangulation, but rather, trilateration.
OK. I really hate to be pedantic about this but let's talk.
1. When you talk about the 100m accuracies, 36m accuracies, 1m accuracies, 5 decimeter accuracies, 1 cm accuracies, etc. in general, you're talking about achievable reproducability with regard to "truth." More on this shortly.
2. Depending on how you account for the various error terms, the CURRENT system and standard allow for autonomous L1, C/A code-phase accuracies of about 29m.
3. Since Selective Availability was discontinued, I and others in the "game" have demonstrated routine L1 C/A code-phase autonomous accuracies of 6-10m, with most of the autonomous accuracies leaning toward 6,
4. The guys who determine error budgets are pessimists (God, I love those guys). They assume that all error terms are scalar additive values. So, they assume that there's no way an error term can be included in the error budget to your advantage. Thus, an error budget that is 6x larger than what I'm seeing regularly.
With a lot of work, long term observations, and use of a variety of hybrid GPS and conventional surve techniques, as well as using a number of sites with methods of ascertaining their relative positions to precise levels, without using GPS, we have been able to cross-correlate the international network of survey monumentation (more true in the CONUS than some other countries) to GPS positioning. Geodesists use network adjustment statistics (least-squares method) to minimize errors and use stable geometric networks to determine the coordinates in 2D or 3D space of an unknown point of interest. Densification of this network leads to a network that allows land surveyors to use either GPS or conventional monumentation with a degree of confidence.
When we refer to a 36m error budget, as the DoD spec does, it's a worst case scenario, and is consistent with the current signal specifications. As we enhance the GPS system with additional signals and methods, we should be able to refine the spec to reflect "reality" over time.
The 100m error budget mentioned refers to a 27-36m error budget for autonomous L1, C/A code phase systems, intentionally degraded using a pseudorandom algorithm called Selective Availability. This was a method of denying precise positioning to a potential enemy of the US. Over time, the integration of techniques, methods and systems to augment GPS autonomous positioning, as well as new techniques (I'm making a few assumptions now) that selective area degradation (denial of quality of service in selected regions/areas) as well as in-theatre jamming led to a US decision to terminate Selective Availability last May. With that, improved C/A positioning was obtained, with an almost instantaneous accuracy of 6m according to hose who were watching and looking for changes.
In short, the collective errors associated with troposphere (water vapor), ionospheric scintillation (solar influence) and refraction, those associated with signal multipath, Rayleigh propagation, and clock errors associated with both the user-receiver and those vaguaries on the spacecraft themselves... not to mention relativity, conspire to degrade the signal. The pessimists call it 36m. The engineers call it, currently, about 6m in my area (Texas; YMMV if you're close to the poles or near the South Atlantic Anomoly). And for those with some older receivers, you have to measure it and ascertain reality for yourselves.
There's no way you're going to reliably obtain a 1m autonomous GPS positional accuracy sans augmentation (WAAS, DGPS, etc). If you think you can, let's talk. I want to confirm your procedures and see if it's publishable.
Can we do better using carrier phase techniques? Sure. I can routinely use 2 receivers (or more, or the US NGS-housed CORS system) and get horizontal accuracies in a least-squares network adjustment on the order of 1cm (2 sigma). But that's not what you were implying.
Transdermal induction-based power supplies were first successfully introduced about 1985; before that they were bigger, bulkier, damaged the skin, had poor efficiency, and were just plain painful. Still, the amount of power transferrable across the intact skin, without damaging same, isn't likely to be enough to provide much juice for a continuously operating motor. Even at low power settings, the continuous portion of that equation is a battery-draining issue.
Further, batteries have to be non-venting and encased in a bio-neutral container. Last time I looked, that pretty well limited them to nickle-cadmium cells, or sealed, depleted electrolyte lead-acid cells. The lead-acid cells are too heavy for these purposes. The number of NiCd's one can package in an unobtrusive manner, and manage to parallel in a fashion that will provide good current-density, is still small. Further, despite advances in NiCd chemistry, there are still issues with "memory" (dendritic formation within the cells) and cycle life. I anticipate more battery work will ensue, especially if the physiology of this device in humans is borne out (most of the testing has been in calves; the issues of cell morphology and fragility are not identical!), then a device like this will spur significant research in associated fields... because the money AND the good will are there.