This post is the first in a set for the 12 calendar months. Graphs are sixteen-year logs of the monthly mean anomaly values of nine climate variables for Manilla, NSW, with fitted trend lines. I have explained the method in notes at the foot of the page.
Raw anomaly values for March
Extreme values for March anomalies in this period were:
Daily Minimum Temperature anomaly -2.9°: March 2008;
Morning Dew Point Anomaly +3.5°: March 2000;
Morning Dew Point Anomaly -3.2°: March 2008;
Rainfall anomaly +61 mm: March 2007;
Percent cloudy mornings anomaly +37%: March 2011.
Trend lines for March
Daily maximum temperature showed minima about -1 deg in 2001 and 2014, and a maximum about zero in 2007.
Daily minimum showed a minimum two years later, about 2003, then rose in parallel with daily maximum, but ended very high.
Subsoil temperature did not agree, and varied less. It had maxima in 2002 and 2012. It may show a lag of five years behind daily maximum.
All moisture indicators fell in parallel from 1999 to 2005. Rainfall continued to fall to a minimum in 2010 before rising sharply.
From 2005, dew point remained steady while other indicators rose.
The composite “Moisture Index” was high in the first year (1999) and in the last year (2014) and lowest in 2006.
Each data point is an anomaly value that is the difference beween the mean value for a month and the normal value for that calendar month based on the decade beginning March 1999.
Raw anomaly values vary a lot from month to month, and different variables often do not move in the same sense.
(Raw values for variables in a given month are in a report for that month. Look for the report for a given month in the “Archive” for the month following it.)
Four of the anomalies of variables are grouped as indicators of the anomaly of sensible heat at the site: daily maximum air temperature, daily minimum air temperature, daily mean air temperature (mean of maximum and minimum) and subsoil temperature (at 750 mm).
The anomalies of five more variables are grouped as moisture indicators relating to latent heat rather than sensible heat. They are: rainfall total (mm), percent cloudy mornings (>4 octas), early morning dew point, daily temperature range (minus), and a composite measure called “Moisture Index”. For plotting, the observed anomaly values of percent cloudy mornings have been divided by ten and the observed rainfall anomalies have been divided by twenty. In the same way, the moisture index is calculated as:
MI = ((Rf anom/20)+(%Cloudy anom/10)+(DP anom)+(-(TempRange anom)))/4
Changes in raw anomaly values are very large from year to year and show no clear pattern. To reveal a pattern calls for trend lines to be fitted.
When I fit linear trend lines, they have almost no meaning. They have R-squared values around 0.01! That is, linear trend lines “explain” hardly any of the variation. When I fit trend lines that are parabolic, cubic, or quartic the R-squared value goes up, until it is around 0.3 for quartic trends. (Quartic trends “explain” about 30% of the variation.) Beyond quartic functions, there are not enough data points to justify fitting the trend line.
Quartic trend lines can identify up to three local extreme points, whether maxima or minima, if they exist in the data.