Profile of an Extreme Drought

Rainfall vs Maximum temperature, 2002 droughtAt Manilla, NSW, there was a drought in 2002 that was extreme, but brief. There have been no other extreme droughts at Manilla in the 21st century. The current drought is not as bad (yet).
The first graph shows a profile of the 2002 drought. Low rainfall is at the top, and hot days are on the right. Droughts, with low rainfall and hot days, will be near the top right corner. Normal climate is marked by a rectangle (coloured aqua (aqua)) in the middle.
The climate in these months moved into drought and out of it. January 2001 (Start) had perfectly normal climate with no drought, and so did February 2003 (Finish). Rainfall first became lower than normal after January 2002, and reached a minimum 27 mm below normal in July 2002. Rainfall returned to the normal range by December 2002. Day-time temperature went above the normal range in May 2002, reached a peak 1.3 degrees above normal in September-October 2002, and fell back into the normal range in January 2003. For rainfall lower than normal, the drought lasted ten months: for days hotter than normal, it lasted eight months. In this drought, the time of lowest rainfall came two to three months earlier than the time of hottest days.

(There is more detailed analysis of  the 2002 drought in a post dated September 2004.)

The loop on the graph shows this drought as a simple event with a beginning, a middle, and an end. Droughts are not usually seen to be so simple. This graph is made using two “tricks”: anomalies and smoothing. You must judge whether you trust them to describe the drought as it happened.

Data adjustment

The left graph below shows the original data for mean daily maximum temperature and monthly total rainfall for these months. The axes of the graph are in the same sense as on the other graphs, putting “droughts” in the top right corner. It is hard to see any pattern at all. The clearest trend is shown by the dashed green line. The warmest months (on the right) are mainly summer months, and these months are usually wetter (nearer the bottom). In the same way, winter months (cooler, drier) group at the top left.

As earlier graph, but unsmoothed

This seasonal trend gets in the way of studying the drought. I get rid of it by subtracting the normal value for each calendar month from the actual value. The remaining value is called the anomaly. Anomalies show whether a month was warmer or cooler or wetter or drier than normal.
Anomaly values for each month are plotted on the right graph. This pattern is close to chaos. Anomaly values for any month are totally different to the month before it and the month after it. (Try working back from the “Finish” month.) No drought can be seen on this graph.
To show the increase and decrease of drought, we must smooth out the graph to make it less jumpy – we must quieten the “noise”. One way is to take average values over several months, such as a three-month moving average, or a six-month moving average. I find that averages like these do a very poor job of smoothing this data. To get a clear pattern, as in the graph at the top of the page, I have used a Gaussian smoothing function. This is like a six-month moving average, but does not cut off sharply at exactly six months. My Gaussian smoothing function uses data from six months before a particular month to six months after it.
Graphs that clearly show climate patterns can seldom be drawn without smoothing. This means they can never be up to date. If, like mine, the smoothing function needs data from six months ahead, then the latest smoothed data will always be six months old.
The price of seeing a clear pattern in the progress of a drought, or other climate event, is to be always several months behind the action.

Other droughts and deluges

Graphs like that at the top of this page feature in all the posts of the category “Manilla NSW/3-year climate trends”. Each post has six graphs in the same format. All have smoothed daily maximum temperature anomaly along the x-axis. The y-axes have smoothed anomaly values of not only total monthly rainfall, but also percent cloudy days, early morning dew point, daily temperature range, daily minimum temperature, and subsoil temperature.
From May 2010 the 3-year climate trends are updated each month. Before that date, posts back-dated to September 2002, September 2004, September 2006, and September 2008 show all smoothed anomaly data from September 1999.

The whole sequence of rainfall droughts and deluges from 1997 is shown in a graph in the post: 17 Years of “Droughts and Flooding Rains” at Manilla.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s