Australia Rain: Seasonal by State

I recently looked for trends in precipitation in New South Wales, searching for a regional rather than continental trend. There wasn’t much to find, no real evidence of change in the yearly average rainfall. But we can also look for trends on a seasonal rather than annual basis, and we can do so for all seven Australian states for which the BoM kindly provides precipitation data.

When we isolate seasons, we find some patterns strong enough to rise above the noise. For example, Victoria has been getting dryer during Autumn:

Meanwhile West Australia has been getting wetter during summer:

Using PCA (Principle Component Analysis), for all seasons the first PC turns out to be very close to the overall (continent-wide) average. The second PC highlights patterns of consistent differences between states. Just as an example, here it is for the Autumn season:

On the left are the loadings, showing that the time series pattern is much like the difference between Victoria+Tasmania(+some of New South Wales) minus West Australia+Northern Territories. On the right is the time series pattern with its evident recent drying.

Both the summer and autumn seasons show similar patterns, West Australia/Northern Territories contrasting with Victoria/Tasmania, and the way the two interact isolates the pattern we noted before: that the northwest has been getting wetter during summer while the southeast has been drying during autumn.

Thanks to Baerbel Winkler for a very kind donation to the blog. If you’d like to help, please visit the donation link below.

This blog is made possible by readers like you; join others by donating at My Wee Dragon.

15 responses to “Australia Rain: Seasonal by State

  1. Very interesting. Wheather patterns change in a way that can not bee seen at first glance, but are nevertheless significant.

  2. It’s the 30th year of the IPCC’s 1990 First Assessment Report (FAR). I’ve written elsewhere about FAR’s accurate predictions, but won’t link to that writing. Just wanted to share some information about FAR, in case Dr. Foster wanted to write about FAR’s accurate forecasts.

    Here are 2 links to FAR:

    Some pertinent information from there:
    1) graphs of projected greenhouse gas increases from 4 scenarios: figure A.3 on page 333 [business-as-usual = BaU = scenario A],
    2) graph of radiative forcing for those scenarios: figure A.6 on page 335,
    3) graph of projected warming for those scenarios: figure A.9 on page 336,
    4) graph of projected sea level rise for those scenarios: figure A.12 on page 337
    5) example of an uncertainty range for the BaU warming projection: figure 8 on page xxii

    Observed values to compare to those projections, with respect to radiative forcing and greenhouse gas increases: [update to: 10.1111/j.1600-0889.2006.00201.x ]
    10.5194/gmd-10-2057-2017 (supplemental figures include additional greenhouse gas concentrations)

    To cut a long story short: observed greenhouse-gas-induced forcing followed scenario B, not scenario A (i.e. not the “business-as-usual” scenario, which over-estimated increases in greenhouse gases). And we ended with about scenario B’s warming trend, so FAR accurately represented shorter-term climate sensitivity (i.e. warming per unit of radiative forcing). Sea level rise also likely fell between scenario B’s best estimate and low estimate. The previous point about FAR’s accurate warming projection is confirmed by the following two papers, the first of which is from earlier this year:

    10.1029/2019GL085378 [also important to check supplemental figure S6]

    If Dr. Foster is interested and has time, I’d like to hear his assessment of FAR’s forecasts for sea level rise and/or global warming, especially since he previously said he places more stock in trends with 30 data points or more, a mark we just reached with FAR’s projections.

    Below I’ve include people previously discussing FAR’s projections. I’ll start with potholer54’s (a.k.a. Peter Hadfield’s) solid assessment, followed by more flawed comparisons from denialists:

    from 1:02:36 – 1:06:28 :

    “A. Failed predictions
    1. Warming rate predictions
    1990 IPCC FAR”

    “Global Temperature Trends and the IPCC”

    • I hate to be cynical–dewy-eyed naivete is more fun–but it’s hard to believe McKitrick doesn’t know that he’s using the wrong forcing. IOW, he probably knows he’s making a misleading comparison.

  3. To some extent this is discussed in the Australian State of the Climate Report at . The autumn into early winter changes are even more nuanced. Research a few years back showed that an increasing proportion of rainfall in the “cool” season was coming out of “summer like” patterns, as the cold fronts and mid latitude cyclones contract polewards . The decline in baroclinic instability in the Australian region during autumn into winter is remarkable, and is a long predicted consequence of global warming.

  4. Australia has three big rainfall sources, and they don’t align well with state boundaries. There are the westerlies that bring rain, mostly in winter, to SW WA, SA, Western Victoria and W Tasmania. There are the SE trades that bring rain with a summer tendency to the E coast, including E Victoria. And there are the tropical monsoons in the N, with a wet season starting about December.

    In terms of trend, the westerlies are not coming so far north, reducing winter rains in the SW. No very clear trend in the east coast, but the monsoons are coming further south, increasing rainfall in the north, and in Australia overall.

    A big region like WA has more than one thing going on. The rise in summer rain is in the north, from the monsoons. Victoria, not so large, is split between the westerly influence and the SE.

  5. It’s obvious but worth pointing out that timing of rainfall is important for groundwater reserves, water courses, and reservoirs. Much of the rain falling on a hot day will just evaporate; when it rains on cooler days, more of the water ends up in rivers, lakes, etc. So a reduction in rainfall during the austral autumn or winter should concern hydrologists.

  6. I know little about drought, drought in Australia, or for that matter Australian weather. But I do monitor statistical and machine learning methods (is there a difference?), and there has been a burst of new applications for watershed and hydrological modeling and forecasting, and droughts. Notably some of these have been applied to Australia:

    * Rahmati, et al (2019)
    * Sachindra, et al (2018)

    The best introduction to these methods, however, comes from a United States application,

    * Konapala, Mishra (2019)

    which uses Random Forests as their modeling engine, and clustering techniques to provide them training data. While I cannot address how good or bad these perform in the problem domain based upon this paper, it offers a good illustration of how to go about using these if one does.

    Note, too, that Random Forests are not as exotic as they might seem to be. For example, Lu and Hardin (2019) and their predecessors offer a means of estimating Random Forest predictive error by marshalling a conditional predictive error distribution function, moving Random Forest methods beyond ad hoc computational devices for providing estimates of miscasts.

    Indeed, these structures are familiar enough to theoretical statisticians that fundamental algorithms for parallelizing MCMC calculations have been devised based upon them.

  7. We have been revegetating our South Australian degraded mallee block with local native vegetation since 1993. In the early years our practice was to wait for the early winter rains and then plant a thousand or so seedlings, mumble a short prayer of hope and leave them to cope as best as they could.. The winter rains were able to get the hardier seedlings through the summer heat and dry and our survival rate was maybe 10% which we were pleased with. But over the years we noticed that the winter rains were decreasing. Just an anecdotal observation but supported by my neighbour who keeps very accurate, albeit amateur and unofficial, weather records.The survival rate declined. In more recent years there has appeared to be a slight relative increase in our summer rainfall as we more frequently catch the tail end of summer monsoons that come way down south to us. But that was less effective rainfall, the soil capacity was depleted. the evaporation rate higher. And the total annual rainfall decreased slightly over that period from about 275 mm to way less than 260 mm in the last 10 years or so [yeah I know, not a sufficient time for a trend but I’m describing this as a tree planter] Until the last 2 years. 2018 was the driest year recorded by our neighbour in 24 years and last year was worse – only 87mm.
    All this anecdotal stuff is in line with this BoM report for our region in general …”A continuation of the trend of decreasing winter rainfall is projected with high confidence. Spring rainfall decreases are also projected with high confidence. Changes to summer and autumn rainfall are possible but less clear. For the near future, natural variability is projected to dominate any projected changes’ [link: ://]
    The result is that we have simply given up planting. Our last few plantings of some hundreds of seedlings per year were total failures – not a single survivor. Too much work and no return and we’re not young anymore. Bloody depressing.

    • @Shea NcDuff,

      My son who visited Australia for a long conference and then spent a couple of weeks hiking the south-southwest with a friend, observed today in a phone chat, when I mentioned the smoke problems at the Australian Open, that the problem will eventually be self-limiting when there is nothing left to burn, because it is all gone. He was, of course, being particularly dark in his humor.

    • Just a suggestion Shea, a handful of fully wetted water holding crystals at the bottom of each planting hole might be enough to get your seedlings through that critical first summer.

  8. There’s a paper coming that looks at changes to seasons since 2000 for Western Australian wheatbelt:

  9. Similar results showing actual and projected autumn/winter rainfall declines in the south east and south west here –