About me

I am a senior weather forecaster--now with the Alaksa Aviation Weather Unit in Anchroage, a former teacher of meteorolgy and life-time weather nut. I have returned to meteorology professionally after some years away.

I am continuing to study to update and upgrade my meteorology background. I spend many hours developing software to download, decode, process, and display weather data. The programs I develop are run exclusively on Apple Macintosh computers. Nearly all of the downloading, decoding, gridding, contouring, display, and computational routines based on Python programs making use of many Python packages.

About this site

While some of the analyses you will find on this site are similar to ones you might find on other sites, I make an effort to include analyses based on my own techniques that give a somewhat different slant from what you will find elsewhere. Some of these analyses I have found useful (eg. surface geostrophic vorticity, temperature advection vectors, msl geostrophic wind, isallobaric wind, hemispheric total kinetic energy, hovmoeller diagrams of kinetic energy, APT satellite images received real-time at my houseetc.); others are purely experimental (eg. total super-cooled liquid water to aid iscing forecsting, icing analysis sounding plots, Rossby diagrams for soundings, departure of thickness from mean, etc.). I have also recently added meteorograms based on 5-min metar observations.

Needless to say, this site is continually evolving.

The analyses must be used with the usual cautions: power failures, internet disruptions, computer gremlins, and failures of my programs can render the analyses and plots outdated or erroneous. I have done only rudimentary error checking on the raw data, so garbage in sadly results in garbage out--user beware! Please be sure to check the times on the current weather plots and analyses to see that they are indeed current.

About the analyses

On my older web page, before the switch to Alaska-centric analyses I did much work on my own analyses for the CONUS. Below describes what I did: After downloading and decoding metars, West Texas Mesonet data, CMAN observations, buoy and ship observations, all of these surface data are put in a common format and stored for each hour. Simple error checking is performed to eliminate the wildest errors from the basic data. For the land observations, the station pressure is computed from altimeter setting and station elevation; for ship and buoy reports over oceans the station pressure is set equal to the msl pressure. Using the station pressures, the thermodynamic variables (potential temperature, equivalent potential temperature, and mixing ratio) are found from temperature and dew point reports. Wind components relative to the grid (100 km square in polar stereographic) are computed. Departure of height (D value) and temperature from the standard atmosphere are computed from the station pressure, station elevation and temperature

The thermodynamic variables, wind components, and departures of height and temperature are interpolated to the grid using a four-pass Barnes analysis with constant weight, rather than a two-pass scheme with a variable weight. This follows the suggestion of Barnes (1994) to improve estimates of gradients, laplacians, etc. The weight parameter is intentionally chosen to produce reasonably smooth results in spite of considerable variation in station density. Differential quantities are found from simple centered finite differences on the grid. The analyses are essentially worthless over Mexico (mostly greyed out) with only a few observations available and often suspect over the Gulf of Mexico with sparse and sometimes unrelable data. Again, note should be made that only rudimentary error checking is done. The analyses are contoured and shaded as appropriate and uploaded to this web page each hour. Presently no prog grids are used for initializing the surface analyses.

Surface geostrophic wind components are found from a form of the geostrophic wind equation in a sigma coordinate system without any reduction of pressures. The horizontal pressure gradient force is transformed from pressure coordinates using the D system. The result is two terms, one involving the terrain-following gradient of D value (departure of height from standard atmosphere at a given pressure) and one involving the gradient of station pressure multiplied by the departure of temperature from standard atmosphere. Over relatively smooth terrain where station elevation varies smoothly, the analyses are quite good. In mountainous terrain there are aliasing errors due to a few stations at extreme elevations. The use of the D system makes the first term the dominant one and elimnates the problem of having the p gradient force come out as the difference between two large oppositely signed terms.

Isallobaric winds (computed from the 3 hour change of geostrophic winds), difluence of the surface wind, water vapor flux vectors, and water vapor flux divergence are also depicted.

Some of the surface maps include shading for mixing ratios greater than 10 g/kg and greater than 15 g/kg (denser shading). Surface winds interpolated to grid are shown as vectors (scale lower left). On the msl p/thetae map surface observations are shown at stations with standard surface wind and weather symbols. On some of the other charts present weather reports are shown as a colored dot ("WX Color") using green for rain and rain showers, blue for snow and snow showers, purple for freezing rain. Intensity is depicted by the brightness of the color. Yellow is used for fog and drizzle while a red dot represents a thunderstorm report. All available station reports of signficant weather are used in these plots. Note, hownever, that on the plots without analyses and on the msl p/thetae analysis the stations shown represent only a small sampling of the stations used in the analyses. Radar composites (from the National Weather Service) are overlain on some of the analyses. Animations of laplacian of thetae, geostrophic voriticity, isallbaric wind, and advection of potential temperature are available as well.

Two analyses from three hours before the latest map time are availabie (on the CONUS scale): one of surface geostrophic vorticity; one of laplacian of thetae. Each shows the severe storm reports during the period between the latest map time and 3 hours ago. These maps allow assessment of the geostrophic vorticity and laplacian of thetae as short-term severe weather predictors. The reports are prelminary and are taken from the NWS Storm Prediction Center web page every hour.

Many of the surface analyses are now shown on maps covering the US and extreme southern Canada--the same area and grid depicted on the upper air analyses--as well as on more detailed "Regional" maps over an area surrounding Rapid City.

Rawinsonde observations are downloaded, decoded, and processed to combine wind and thermodynamic data every 12 hours. These are put in a standard format and then processed by sounding plot routines and constant pressure map plot and analysis routines. As with the surface data only very basic error checking is done.

The RAOB Plot window allows selection from a set of U.S. and Canadian soundings. On each sounding page one is taken initially to a plot of theta, thetae, mixing ratio, RH, and RH with respect to ice as functions of ln of pressure. Additionally, one may choose a plot of temperature, dew point, wind speed, RH, and RH with respect to ice as functions of ln p. A hodgraph of the smoothed winds (components smoothed with an exponential function of pressure) is available. Plots of the environment virtual temperature and the virtual temperatures of parcels lifted from the lowest 100 mb mean as well as the one having the highest SMOOTHED thetae in the low levels (parcel of best lift) are shown together with sundry stability measures (LI, CAPE, Deep Shear, etc.) for the lowest 100 mb mean parcel and winds and shear for the lowest 100 mb mean and the surface to 200 mb pressure weighted mean. Lastely a new experimental hodograph showing the water vapor transport vectors computed from smoothed mixing ratios and smoothed winds is shown together with the vapor transport relative to a deep convective storm moving with the surface to 6 km mean wind and relative to the Bunkers right storm motion. Where winds are plotted with conventional symbols they are the observed winds. Plese note that I have intentionally chosen NOT to use SKEW T-ln p plots because I want to plot additional variables on the same scale as that used for T on the T vs. ln pressure diagrams (eg. wind speed, rh, rhi, and the other thermo variables). This is not reasonable to do when the temperature scale is skewed.

The Upper Air Analysis page has plots of hourly VAD winds from NWS doppler radar sites. It shows standard wind symbols for the 600 m, 1km, 2km, and 3km levels above ground. It also shows a vector plot of the winds at 1, 2 and 3 km so that one may quickly assess shears and estimate temperature advections via the thermal wind relationship.

The upper air page includes constant pressure map analyses. One set is a standard plot of station data showing temperature, dew point, wind and D value. The analyses are of D value--that is the departure of geopotential height from that in the standard atmosphere for the same pressure level. Contours of D will, of course, parallel the more commonly shown contours of geopotential height. Relative humidity is shaded for RH>60% and RH>80%, the latter denser shading on the lower tropospheric analyses while those above have shading for wind speed. Note should be made that the relative humidities are with respect to ice for subfreezing temperatures. Another set of analyses shows similar plots with D value contours and RH, but this time with relative vorticity coomputed from the observed winds added.

The upper air analyses are now initalized with a first guess from the 6 hour GFS model FCST inteprolated to my grid. The anslyses are then modified with subsequent passes using observed data through the Barnes scheme. As in the surface analyses, most of Mexico is greyed out due to lack of data. Use of the GFS first guess has resulted in substantial improvement of the analyses over Canada and around the edges. Profiler winds are included in the analyses.

I am also using the GFS model analyes to do hemispheric maps. Following interpolation from the GFS native lat/long grid to my Polar Stereographic grid I show maps of the 500mb contours together with departure of the 1000 to 500 mb mean virtual temperature from long-term mean values. Also shown are analyses designed to separate long and short wave features through use of a heavy space smoothing (using and exponential distance weight) to get smoothed heights and winds and then subtracting these from the basic fields to get departures from the smoothed fields. I moved to downloading the GFS hemispehric data at 25 mb intervals for these analyses and for the computation of the surface to 100 mb mass weighted mean wind and surface to 100 mb total kinetic energy. Plots of these are now available.

The "Statistical Forecasts" page shows the temperature and probability of precipitation as well as the conditional probablilty of snow given precipitation for stations in the Rapid City CWA (as well as Fort Worth, Texas) from the GFS ensemble MOS output. These have the advantage over other displays of showing not just the means and standard deviations, but also every one of the members of the ensemble so that one may quickly see how diverse the results are.

I have done a reorganization of the pages (spring 2013) to include a "Model Data" section and a "Case Studies" section. The former are gradually growing while the latter is presently in the "hope to add soon" category.

AGAIN PLEASE note that while I make every effort to keep this site operational, there are many potential problems that can result in the programs failing to post up-to-the-minute maps or producing maps from inadequate data.