|
|
 | | From: | Glenn Woodell | | Subject: | Response from iWindsurf on nw enhancements | | Date: | Fri, 21 Jan 2005 22:28:28 -0500 |
|
|
 | Just got the following from Davis St. John, the IT guy at iWindsurf.
Glenn
Thanks for the nice compilation, Glenn. Sorry for the slow response
but we've been squashing bugs and fine-tuning since launching the new
codebase.
This is pretty constructive feedback in general - very helpful. We've
also been getting a lot of direct feedback through our "contact us"
interface, which mirrors much of this input. We're aware of many of
the points below and are working to mitigate them - some truly need to
be "fixed", while some are simply things that hopefully just take a
bit of getting used to.
Here's a response to some of the most popular topics. We haven't had
time to respond to all the public commentary, but we have read most of
it. Please feel free to post these responses back to rec.ws and the
other forums from which these were culled. But please don't include
any of our personal addresses. The best way to contact us is through
http://www.iwindsurf.com/support
- What new features? If you haven't seen this page yet, check it out:
http://www.iwindsurf.com/whatsnew (or
http://www.ikitesurf.com/whatsnew ). Yes, many of the new features
were inspired by www.SailFlow.com - some of the "new" things on
iWindsurf are almost a year old now - so thank your friendly sailboat
sailor next time you see him!
- Regional boundary changes. Some of the maps have been reworked
based on feedback from users, but in all cases the same information is
still available. You may have to look around a bit though. In some
cases the same bookmark will pull up a very different map. For
example, the south San Francisco Bay region
(http://www.iwindsurf.com/windandwhere.iws?regionID=126) is more
zoomed in than it used to be. However, we've added a full Bay Area
map: http://www.iwindsurf.com/windandwhere.iws?regionID=163 In fact,
any time you're looking at a map, you can click the "zoom out" link to
take you up a zoom level. Also, you can add two or more regions to
your "myIwindsurf" page now: http://my.iwindsurf.com , which is
much-improved IMO.
- Site labels on maps. We've received lots of complaints over the
years that our maps were so cluttered with labels that you couldn't
see the wind data. We've also received complaints that it's difficult
to orient yourself in an unfamiliar region. The truth is, if you're in
an unfamiliar region, having the names of the sites doesn't help much
- if you have no idea where that site is. So, we're doing two things
to address this, neither of which was totally completed when we
launched the new codebase: (1) We're creating new maps for all regions
- maps with more geographic reference points. (2) The new system
allows labels to be optionally (and dynamicaly) placed on the maps,
which we're doing region-by-region. We're also working on a better
font treatment for the labels so they aren't too distracting. Many
regions still look naked, but check out the Los Angeles region to get
an idea of what we're shooting for:
http://www.iwindsurf.com/windandwhere.iws?regionID=129 The red
asterisk is also an issue we're aware of, and are working to improve
it.
- Computer model data. While we don't believe they'll ever be
*better* than our professional meterologist's forecasts, the Computer
tables are definitely getting better. In addition to the expert
system post-processing of publicly available models, like ETA 12km and
GFS 40km from NWS/NOAA, that we've been doing for several years, we
are for the first time running our own numerical models down to 2km
resolution. Last year we installed a 14-processor "cluster" - a mini
super computer on which we run our own heavily tweaked version of a
regional atmospheric modelling system that's known for it's accuracy
with respect to meso-scale wind forecasting. This project is funded
primarily by non-iWindsurf contracts, but iWindsurf gets the benefit.
In regions where we're running this model we've seen significant
improvements over the previously best public model out there:
ETA 12km. We're running it in the Bay Area this winter, and will
start it up in the northeast this Spring. We hope to continue
expanding the cluster so that we can run it in more regions at better
resolutions than other publicly available models.
- New graphs. For the first time, and after many many requests, the
graphs plot the datapoints accurately, according to the timestamp on
the observations rather than binning the obs into four "bars" per hour
like we used to do. The bars were a legacy of the Call of the Wind
system, which was much cruder and was based on the windtalkers that
only called in a max of 4 times per hour, and also had a 3-5 minute
uncertainty because of the path the data took to get to us. These
days the windtalker data gets to us more directly (still only 4x per
hour max), and the newer sensors (NextGens) are accurate to within a
couple seconds and update as often as 12 times an hour. The new line
graph takes a little getting used to, but most people agree they're
better. As for the time range across the x-axis, we're probably going
to go back to the old daylight hours (6am to 8pm) scale as a default,
since most sensors only report in that range anyway, with the option
to "show 24 hour scale" for those who want to know what the wind's
doing at night.
- Knots vs. MPH. This is another legacy issues that's all about
inertia. The wind reports have always been in MPH (because that's what
the windtalkers report in), while the forecasts have always been in
KNOTS (because that's what our meteorologists think in). But it's not
a technical issue. If you look at www.sailflow.com you'll see that
everything is in KNOTS. But on iWindsurf and iKitesurf, wind
observations are in MPH and forecast data is in KNOTS. We realize a
sudden change to the measurement used for displaying the data on the
snapshots, graphs, wind alerts, custom homepage, could undoubtedly
cause considerable frustration for some users. For now, we plan to
leave the wind observations in MPH and the forecasts in KNOTS on the
iWindsurf and iKitesurf websites. In the future we may add a
configuration option to allow users to select which measurement they
would like each displayed in.
- Meteogram tables. Several folks seem to think we've gotten rid of
the "day-at-a-glance" regional forecast tables that everyone's used to
seeing, but we have not. Those are still there - just click the "Wind
Forecast" link from virtually any region. The new 7-day-for-one-site
feature we call the "7-Day Meteogram" and it's actually been on the
iWindsurf site for a couple months. I think perhaps the new layout
has caused some navigational confusion, but I swear: everything we
used to have is still there!
- FlowViz maps. This is also a brand new feature. Right now it's
forecast data only, but our plan is to take the wind reports for a
given time across an entire region and normalize it into gridded data
that can be fed into the FlowViz applet. Easier said than done,
however, but we're working on it.
- Wind forecasts from other sources. You don't need a dartboard, or a
magic 8-ball, or any of those other weather websites anymore! They're
all just repackaging National Weather Service data anyway.
- Fee hike. We have no plans to raise the rates this year.
Thanks again and good winds.
--
David St. John
WeatherFlow, Inc.
www.weatherflow.com
PS: As for the topic of posting forecast data side by side with
observations, this has nothing to do with the new website features -
we've *never* posted old forecast data - but it's obviously a hot
topic so here's our reply: There are several reasons we don't post
forecast data after the time has past. One of the main reasons being
that forecasts are updated several times a day and at that point we
don't produce a forecast for the times that have already passed.
Keeping it simple, we only display the latest forecast and the tables
only show what is forecasted to happen in the future.
PPS: We do quite a bit of forecast analysis and model validation
(that's how we know the computer model tables are getting better!) We
also tested a very helpful human forecast feeback tool last year that
we hope to expand on and run again to get the qualitative feedback
that's so important to continuing to provide a good forecast product.
|
|
| | From: | Bob A. | | Subject: | Re: Response from iWindsurf on nw enhancements | | Date: | Sat, 22 Jan 2005 00:58:27 -0500 |
|
|
|
>
>
> - Computer model data. While we don't believe they'll ever be
> *better* than our professional meterologist's forecasts, the Computer
> tables are definitely getting better. In addition to the expert
> system post-processing of publicly available models, like ETA 12km and
> GFS 40km from NWS/NOAA, that we've been doing for several years, we
> are for the first time running our own numerical models down to 2km
> resolution. Last year we installed a 14-processor "cluster" - a mini
> super computer on which we run our own heavily tweaked version of a
> regional atmospheric modelling system that's known for it's accuracy
> with respect to meso-scale wind forecasting. This project is funded
> primarily by non-iWindsurf contracts, but iWindsurf gets the benefit.
> In regions where we're running this model we've seen significant
> improvements over the previously best public model out there:
> ETA 12km. We're running it in the Bay Area this winter, and will
> start it up in the northeast this Spring. We hope to continue
> expanding the cluster so that we can run it in more regions at better
> resolutions than other publicly available models.
>
I'm very interested to see how the improved model works for the NE this
spring. If it can forecast accurately the wind that actually couples with
the water surface, and if it can predict the fair weather troughing that
often surprises Buzzard's Bay and/or Cape Cod with moderate to strong SW
wind in the midst of NW synoptic flow, then I'll be stoked and hats off to
you.
>
>
> - Wind forecasts from other sources. You don't need a dartboard, or a
> magic 8-ball, or any of those other weather websites anymore! They're
> all just repackaging National Weather Service data anyway.
It's gonna take a lot to wean me off those real time 1 mb isobar mesoscale
analyses charts from CoolWx, along with the mesoscale streamlines. .I think
weather forecasting is most useful when it alerts the user to "the
possibilities", and assigns "probabilities". Getting it right may sometimes
mean making sure the end user has the tools to make the most of the
possibilities. An updated forcast every hour would be nice (but maybe not
feasible). Access to the most useful real time data by wireless interface
every hour (i.e. 1 mb isobars, streamlines, skew-T coupling) would add
significantly to our arsenal When it's me at the beach in something other
than what was forecast it would be nice to know why, and nice to have the
best information available to decide if waiting or driving further, versus
driving home, is warranted.
>
>
> PPS: We do quite a bit of forecast analysis and model validation
> (that's how we know the computer model tables are getting better!) We
> also tested a very helpful human forecast feeback tool last year that
> we hope to expand on and run again to get the qualitative feedback
> that's so important to continuing to provide a good forecast product.
I'm game.
>
>
>> Thanks again and good winds.
> --
> David St. John
> WeatherFlow, Inc.
> www.weatherflow.com
>
Thanks for always striving for improvement. We're a lot better off today
than ever.
--Bob A.
|
|
| | From: | Paul Braunbehrens | | Subject: | Re: Response from iWindsurf on nw enhancements | | Date: | Fri, 21 Jan 2005 20:12:00 -0800 |
|
|
| However, we've added a full Bay Area
> map: http://www.iwindsurf.com/windandwhere.iws?regionID=163 In fact,
> any time you're looking at a map, you can click the "zoom out" link to
> take you up a zoom level.
Yes, I noticed that, but it's useless because it doesn't show TI,
Richmond, Angel Island, etc. etc. so you don't have an idea what is
going on in the areas, i.e. where the "river" is flowing. There is
plenty of room on that graph to have all those sites.
|
|
| | From: | Cliff Frost | | Subject: | Re: Response from iWindsurf on nw enhancements | | Date: | Sat, 22 Jan 2005 18:59:37 +0000 (UTC) |
|
|
| Exactly!!! This totally sucks. Please, please, fix!!!!!!!
Thanks,
Cliff
Paul Braunbehrens wrote:
> However, we've added a full Bay Area
>> map: http://www.iwindsurf.com/windandwhere.iws?regionID=163 In fact,
>> any time you're looking at a map, you can click the "zoom out" link to
>> take you up a zoom level.
> Yes, I noticed that, but it's useless because it doesn't show TI,
> Richmond, Angel Island, etc. etc. so you don't have an idea what is
> going on in the areas, i.e. where the "river" is flowing. There is
> plenty of room on that graph to have all those sites.
|
|
|
