How well forecast in advance was the new all-time record for peak NEM-wide wind?

We already noted this morning that there was ‘A new all-time record for NEM-wide wind production on Wednesday 1st July 2026’.

With memories still fresh of how poorly the batteries performed* in South Australia on 21st and 22nd June 2026 (in comparison with a totally impossible, but nevertheless irresistible, hindsight-driven charging regime!) it’s useful to use this Case Study as another case to think (once again) of the increasingly important question of  ‘how forecastable is VRE?’.

* remembering that the issue there was not primarily due to forecastability of VRE, but mainly because of forecastability of other aspects of the energy system (like hot joints and hot fixes).

Firstly some background, before the keyboard warriors at both ends of the Emotion-o-meter load in to shout their pre-configured lines, like broken records:

1)  We’ve previously written ‘Is VRE forecastable? (Part 1 in what might be a series)’ to explain some of our perspective

(a)  for instance, why we’ve invested a non-trivial amount of time and money in a wind forecasting organisation, ETSI.

(b)  by also why it’s not a solved problem … and (in our view, and why we invested) not as good as it will need to be into the future.

2)  And we’ve collated a growing range of analysis under the ‘predictability of VRE’ category here

… to which we add this one.

In this article we’ll start to explore this issue, with two different representations of the ‘Forecast Convergence’ widget in ez2view late on Thursday afternoon 2nd July 2026.

Step 1)  We’ll start with the ‘Grid’ view, selecting the NEM-wide Wind UIGF metric over the ST PASA time horizon (i.e. back ~7 days and forward ~7 days):

Remember that this widget allows one to ‘look up a vertical’ in order to see ‘that other dimension of time’

Focusing on the peak wind yield of late Wednesday 1st July 2026 and early Thursday 2nd July 2026  we can see a solid block of red ‘up that vertical’

 … but hold onto your horses, because we see that that swatch was decidedly paler back on Friday 26th June 2026.

Also worth noting, as an aside, that the current forecasts for Wednesday 8th July 2026, Thursday 9th July 2026 and Friday 10th July 2026 are for a reversion to modest wind yield (quite low at times).

… but, as we saw above, that’s a few days away now and the current forecasts are quite rubbery.

We flip to chart view and see the differences:

In the earliest forecasts from ST PASA (i.e. the faintest ones) we see that the miss was greater than 2,000MW at times (i.e. more than 20%).

Before the keyboard warriors weigh in with the binary “YES, it is forecastable” or “NO, it is not forecastable” it might be more useful to consider more meaningfully:

Q1)  Under what circumstances might a miss of that size, that many days in advance not be material;

Q2)  But, memories of poorly performing batteries ( as one type of shaping/firming capacity)  fresh in mind, under what circumstances might a miss of that size actually be material?

… especially as we increasingly move towards a much more ‘just-in-time’  type of energy equation (with risks tested only earlier this year).

… and that’s where we will leave this, for now.