I recently did some of Coursera’s data science modules as way to sharpen my skills in big data. The modules use R to process the data and do the statistical analysis. My goal is to use this knowledge, I will try to analyze Tobago’s precipitation and temperature for the period 1980 to 2015. The data can be found at the Trinidad & Tobago Meteorological Service. This analysis will be in two parts: Rainfall trend and Temperature trends.
For the past few (10?) years Tobago have been experiencing rather hash drought like conditions. Each year, the Water and Sewerage Authority (WASA) has implemented its Water Supply and Conservation plans in order to ensure a supply. Some of these included water rationing, a ban on watering plants or washing cars with a hose.
As a farmer, I felt the effects of the heat and lack of rain fall during the wet/rainy season. I had to use pipe borne (via a tank) water almost everyday. Fortunately, I was able to harvest my crops though I believe the yield would have been higher if the rain fall was “normal”.
The main reason WASA has given for the short fall in water is that rain fall has been steadily declining over the years. While the public believe WASA did not plan for increase usage by the population. So which is true? No planning for increased usage or decrease rainfall over time. Hey, maybe both!. WASA has not given any “numbers” to back up their assertion. While there has been a marked increase in usage due to a higher number of visitors to the island especially during July/Aug months.
So is WASA’s reason for the short fall true?
Some back ground:
- There was no meta data on the site.
- Found no literature on this topic on Tobago specifically
- The Met office collects data from only one location the airport.
- The code -9999 was used for cells without valid data.
- The variables are: Date, Precip, Tempmax and Tempmin
- Temperature Units: Celsius
- Rainfall Units: Millimeters
- Rainy season: June thru December.
- Wettest month: July
- Driest month: March.
- On average, the warmest month is July.
- On average, the coolest month is November.
- Petit Carême, which is similar to what is known as Indian Summer, normally happens between mid September to mid October.
Summary of what I did
Review of Data
Downloaded and imported the data into R (actually used Rstudio). A review of the data shows there were:
- 13,332 records with 4 variables (Date,PRECIP,TEMPMIN,TEMPMAX)
- Need to convert Date from character class to date class
- Need to convert missing data code (-99999) to NA’s
After tidying (i.e. clean and prep for analysis) the data, I thought as a first pass to see whether WASA’s assertion of the drop in rainfall, I looked as the rainfall for the wettest month (July) of the year in 5 year gaps. Also, I looked at the total rainfall for the years 2000 to 2015. Graphs shown below. I wanted to see if there was a noticeable drop in the rainfall over time.
A review of the graphs shown below indicated that other than the year 2000 and 2015, rainfall over the years (for July) has been pretty consistent. Rainfall for the year 2000 was extremely low (under 10 Millimeters) for the month. And for the year 2015, although the rainfall was more than in year 2000, it was still below 10 Millimeters.
The yearly totals of rainfall graph shows a consistency until the sharp drop from 2014 to 2015.
Full details can be found here
Unfortunately, I can’t draw any conclusions for the entire island only for the area where the data were collected. As I mentioned above, the yearly rainfall graph shows no significant decline in rainfall over the years. This begs the question: What’s the cause of lack of water in the dry season. Some reasons could: be an increase in population in the area without a corresponding increase in water collection; Lower rainfall in other catchment areas that feed the area where the data was collected; lack of maintenance of infrastructure (there are a lot of complaints about leaks).
In searching whether data were collected from other areas of the island, I found out that WASA has about 50 stations across Tobago. I was also told the may be gaps in the data due the lack of proper maintenance of the equipment in the stations.