The most frequent updates on the atmosphere (about every 5 minutes) come from weather radar. WeatherScope has access to the past 24 hours of raw US radar data, while the free weather websites will show you about 45 minutes worth. 24 hours is enough to see a lot of patterns – arcs of rain sweeping in from the Pacific, blossoming into downpours when they hit a ridge.

Here’s a WeatherScope saved view for base reflectivity from three overlaid radars covering north-central California (San Francisco, Sacramento, Beale AFB): sf-radars.wxscript

At least on the Mac, WeatherScope should automatically open .wxscript files. After opening the file, you should see a map of California. In the WeatherScope menu bar, go to “Window : Show Animator” and click the Play button in the Animator window. It will take a while for WeatherScope to download 24 hours worth of animation, so you won’t see much to start with. Once it fills in hopefully you’ll feel it was worth waiting for. Leave WeatherScope running and it will download new images as they become available.

UPDATE: Here’s a much better looking sanfrancisco.wxscript courtesy of James Home.

Alpine Lake and Bon Tempe Lake (CC-by-SA license)

I’ve spent several days trying to figure out why the roads in the MRLC land cover data don’t line up with the OpenStreetMap roads. The creeks (from NHD), elevation (from 10m NED), and roads all seem to line up in a way that matches aerial photographs. I double checked all projections, installed the latest versions of everything, lots of web research, no clue what’s going on.  The roads seen in the landcover data are the right shape, they’re just off in position. But they’re off in position by a variable amount.

I finally noticed that the landcover doesn’t even agree with itself! There are places where the landcover data indicates a road crossing open water and there is obviously no bridge. The dam on Bon Tempe lake shows that the vegetation and water are correctly aligned with OpenStreetMap, but the pink line for the road crossing the dam doesn’t line up with the other landcover pixels.

Is this a known problem with MRLC?

North flank of Mt Tamalpais

I’ve been playing with making topo maps using elevation and land cover data. Enough is in place now that some of the bugs look interesting. My initial take on light and shadow turned out to be a bit aggressive.

At full size this image becomes very abstract and reminds me of camouflage. The stripes from the shaded relief look like some animal patterns, and the blockiness of the land cover data looks like MARPAT.

As a thumbnail it’s more obvious what’s going on:

Yesterday a four-alarm fire destroyed four warehouses and damaged three others on our block in Bayview. The rAnch was unburned, in spite of fire burning on two sides. Water damage was minimal, in spite of the ladder truck overhead and firefighters tearing out the two windows in the great room in order to hose down the building in back. Here’s a view through one of those windows:

Unbelievable. Our guardian angel(s) (not to mention the SF Fire Dept) were working overtime. Mauricia’s painting hanging next to the window escaped with nothing more than some spray. Her studio in the next room was completely dry. The wood shop downstairs looks like it will be ok after it dries out.

We started some cleanup last night, in a surreal atmosphere of mist and smoke and floodlights as the fire trucks continued to pour water on the smoldering wreckage next door. Nobody was hurt and no residences burned.

Nothing in life is so exhilarating as to be shot at without result.
— Mr. Winston Churchill

Sand from the base of Eureka Dunes

Desert sand has sharper edges than water-tumbled beach sand.

More exotic sand photos at sandcollectors.org.

This is the most pronounced cloud shadow I’ve ever seen. The cloud in question appears to be the bright white streak near the horizon, which is almost but not exactly lined up with the shadow. But why
the strange alignment of the original cloud? It could be a contrail, but it’s very thick and fluffy and seems to start very suddenly.

The shadow stretched halfway across the sky.

MoFrames is a collection of images that do just that. The image on the right was constructed from a video of two swing dancers, with the background removed and each video frame shifted slightly forward and overlaid using a piece of software called Recreating Movement. The MoFrames site contains more examples including a page of videos that play with time, space, and motion in unusual ways.

(via kottke)

See also Golan Levin’s list of slit-scan video art and research.

One technique I haven’t seen before is the use of tone mapping to prepare each frame. Tone mapping is usually used to compress high dynamic range images so that they can be displayed on a standard computer monitor. You can’t show the the absolute brightnesses in an HDR image, but through tone mapping you can preserve the relative brightnesses.

For timelapse purposes, what tone mapping does is hide some of the variation in lighting conditions between frames. This means that you don’t get unpleasant flickering in the video when a dim, cloudy day is followed by a bright sunny day. Shadows still appear and disappear across frames, but this is a clever way to preserve the same overall brightness.