Lightning Density vs Elevation Plots - General Discussion

    The three plots of total strikes, total area and strike density versus elevation form a natural grouping.  This trio of plots has been done for all 72 quads in the study region.  A key ingredient in generating these plots is the NLDN dataset of strikes, augmented with an elevation assignment for each strike.   Details of how this assignment was made is discussed elsewhere in these pages under Data Source Discussion  for the Lightning Strike Maps.   Once the elevation for each lightning strike is determined, the variation with lightning density with terrain can be investigated.

Tucson Sample:
   The first image below shows a sample for the Tucson quad.  The blue plot is the total number of strikes for July and August added together and then averaged for the entire five years (1995-1999) of the study.   The plots were totalled in elevation bins of 100 m.  The count for a given range is plotted against the lower bound of the range, so for instance, the point plotted on the 300 m mark on the horizontal axis gives the total strikes in the 300 m to 399 m range.

The red plot in the above image shows the total area in square km at each 100 m elevation bin.  The shape of these areas is usually torturously serpentine, but the area was calculated essentially by a trivial numerical integration.  The highest resolution elevation data was used.  This gives an elevation value for every rectangle in a 3 arcsec latitude by 3 arcsec longitude grid (3 arcsec correpsonds to roughly 100 m distance).  The whole rectangle is treated as being at the same elevation and it's area is added to the total.  The decreasing width of these rectangles going from south to north is taken into account.  Again, the 300 m data point shows the total area of terrain with elevation of 300 m to 399 m.
     The numbers on the total strikes axis are just the average total counts per year.  The numbers on the area axis are in square km.  However the two scales are given the same numerical values in all the plots so that places where the blue and red curve cross correspond to 1 strike per year per sq. km.  As the blue curve gets higher than the red curve, the strike density increases.  Note in the sample above, the blue data point is just below the red data point for the very first bin (300 m), they are about equal in the next bin and then the blue curve eventually doubles and triples the height of the red curve (although they both decrease at the higher elevations.
    The strike density is just the value on the blue plot divided by the value on the red plot.  These densities have been plotted in a separate set of plots.  The sample below is also for the Tucson quad.  Again, the strike densities are for July and August added together, but then averaged over the five years.

Even though the actual number of strikes is quite small in the top several bins, the amount of area is very small as well and it can be seen that the strike density clearly increases with altitude.   In this and many other quads there also appears to a slight tendency for the strike density to peak just below the very highest one or two elevation bins.  One factor that complicates such an observation is that the density values become significantly more volitale toward the very top of the scale.  As the area gets very small, random fluctuations in the number of strikes can have a larger percent effect.  The top bin might only have 0.4, or 0.2 or even 0.1 square km.  An area that size might only get one or two strikes, or none, in a season.  A change of one strike changes the strike density per square km per season by a huge percent.
   The strike denisty plot above has a lot of fluctuations that have the appearance of random noise.  However, it can seen by looking at the data year by year that many of the peaks and valleys occur in the same places on an annual basis, suggesting that they are tied to topographic features of some kind.  The image below shows the same strike density data for Tucson as the previous plot, except that the contributions from each of the five years is plotted separately.

Note the peristence of peaks at 800 m, 1100 m, 1600 m, and 1900 m.  There is a signal here that appears in even a single year most seasons.  The erratic nature of the values for the highest elevation bins can also be seen.  Year to year changes in the overall pattern can also be seen.  Note that 1999 was the most active lightning year in the low to mid elevation range of 600 m to 1300 m (which are the largest regions in terms of total area) and yet 1999 was eclipsed by both 1996 and 1998 at the highest elevations.  1996 in particular was a very intense year for storms on the highest peaks.
    For another example consider the Mesa quad shown below.  This is also a plot showing the year by year pattern.

There are recurring peaks again (this time at 1100 m, 1300 m, 1600 m and 1900 m).  The highest bin shows the greatest volatility.  In this case, 1999 stands out even more as a different pattern; being the most active year in the lower deserts but only an average year in the higher mountians.

Results from Selected Quads:
   Over the 72 quad region, a variety of different regimes are seen.  One feature to keep in mind is the volatility of the last one or two bins.  The next image shows an extreme case, but one which is seen several times in the plots.  This is a plot for the Williams quad:

 The last bin contains only 0.1 sq. km. of terrain.  This is basically the very tip of the highest peak in the quad.  The area is so small that no strikes happened to occur at all in that area in the five years of the analysis.  Just one strike would have boosted the yearly average to 2.0 strikes per sq. km. per year.
    Four different categories of lightning density plot will be seen below.  Each one from a different part of the 72 quad study.  Southern Arizona and Southern New Mexico, with their mix of lower desert and some large mountains show the greatest variation in lightning density.  The following two plots are for the Mesa quad in Arizona and the Roswell quad in New Mexico:

 In both cases the lightning density steadily increase by five or six fold between the lower and higher portions of the quad.  Southern Arizona and Southern New Mexico also have the highest values as well, with densities as high as 5, 6 or 7 strikes per sq. km. per year.  In Utah, the lightning also tends to increase with elevation, but not as dramatically and the numbers themselves never get very high.  The following two plots are for the Delta quad and the Ogden quad, both in Utah:

 Lightning density still doubles or triples with elevation within a quad, but the values on the most active peaks are only comparable with the least active lower elevations of a Southern New Mexico quad.  The very highest elevations in Colorado are not always anymore active (or even as active) as the middle and lower elevations.  The following two quads are from Western Colorado; the Durango quad and the Leadville quad.  Both of these contain very large and very high mountainous regions.

 In both cases, the curve is fairly flat.  There seems to be a tendency for the most lightning to occur at the middle elevations.  Neither of these quads contain any of the high plains east of the front range.  The next two quads do straddle the front range, and a very different pattern occurs.  The two plots are for Denver and Trinidad:

The next graphic highlights some of these general trends.  In this plot, the data for all 72 quads has been combined into just the four state regions.  The four resulting plots are compared in the figure below:
 Utah has a slight uphill trend.  Colorado has a slight downhill trend.  New Mexico shows a steady increase in lightning up to about 2500 m (8000 ft) but then the density tapers off above that.  The New Mexico plot is effected by the high plains environment.  On the east side of the New Mexico 'front ranges', lightning tends to start early at the highest peaks, but the later storms that move out over the mid-elevations of the foothills are more prolific lightning producers.  Arizona shows the best overall correlation of lightning with increased elevation.  Even here though, there is a big dip in the curve around 1500 m - 2000 m.  The rectangular regions being used here are artifical and can mix together different sorts of terrain in a hit or miss fashion.  To some extent the exact shape of these curves is an accident of where quad boudaries happen to fall.  On the Arizona plot, there is a sharp rise in lightning from sea level, on the Gulf of California, up to about 1200 m (4000 ft).  As one moves up to 5000 ft and 6000 ft in the central basins south of the Mogollon rim, the lightning continues to increase.  But on a statewide basis, moving just above 4000 ft  begins to introduce the broad flat valleys of Southeast Arizona and then a little higher brings in the large flat parts of the Little Colorado River plateau in Northeast Arizona.  These relatively low lightning areas bring the curve down until about 2000 m (6500 ft) at which point the signal from the mountainous parts of the state re-asserts itself.