There was an issue when the first value above 0 was received, the graph would make the entire plot except for the new data points, the same as the first point after 0.

What was wrong:
The original code for rendering the chart iterated through the data points received from the APIs and plotted points directly at the timestamps when the data arrived. This meant:

• The chart points were not guaranteed to be exactly every 5 minutes.
• When data was sparse, there would be gaps or irregularly spaced points on the chart.
• The "Real Aurora Score" (red line) didn't explicitly "carry forward" the last known value if no new data arrived for a particular time interval. It relied on finding the "closest" point, which isn't the same as maintaining the previous value.

What was done to fix it (Bullet points):

• Sorted Data: The data received from both APIs was sorted by timestamp in ascending order. This makes it efficient to find the latest value at any given time.
• Fixed Time Intervals: Instead of iterating through the incoming data points, the code now iterates through every 5-minute interval for the last 3 hours. The start time for this iteration is rounded down to the nearest 5-minute mark (e.g., if 3 hours ago was 6:53, the chart starts at 6:50).
• Find Latest Value: For each 5-minute interval in the loop, the code checks the sorted data to find the latest data point (for both datasets) that occurred at or before that specific 5-minute time.
• Carry Forward Value: The identified latest value for each dataset is stored in a variable (lastValue1, lastValue2). For each 5-minute interval, this lastValue is added to the chart's data array. This ensures that if no new data arrived since the last 5-minute mark, the previous value is used, effectively "carrying it forward" as requested. If the last known value was 0, it remains 0 until a new, non-zero value is received.
• Consistent Labels: The time of each 5-minute interval is added to the chart's labels array, ensuring the x-axis has points exactly every 5 minutes regardless of data arrival times.

In essence, the fix changed the approach from "plot the data you got" to "plot a point every 5 minutes using the latest data available up to that time".

I've wanted to include a bit more stats to the site for a while but I wanted to make sure it was user friendly, so I've incorporated an emoji system to help with this. Happier the emoji, the better the stats!

What's new:

• Added stats:
  • Solar Wind Speed
  • Solar Wind Density
  • Hemispheric Power
  • Bt
  • Bz

What Changed:

• Removed colour elements from background

We've added an interactive New Zealand Aurora Forecast Map! This is based off of our aurora forecast score with no lunar or sun influence. You can see this being plotted on the 3-hour graph. This is still a 2 hour forecast, not an immediate forecast. Please be patient.

If you live out of the West Coast and don't think this lines up, please get in touch with us via the forecast feedback form. This is pre-release, and is only up for testing.

What's Changed:

• A new interactive graph element on the website that visualizes where aurora activity may be.

After a tumultuous start, we are finally live, with a final product. Updates will be made as necessary, or to add extra functionality.  

How does the forecast work?

• NOAA publishes space weather data
• A local Home Assistant server records the data
• Home Assistant calculates the data into 4 different scores;
  •  Aurora Score - No Lunar or Solar Influence
  • Final Aurora Score
  • Aurora Score - No Lunar or Solar Influence (30 Minute Average)
  • Final Aurora Score (30 Minute Average)
• The two 30 Minute Average scores get sent via API to a online hosting service
• I embed the result in my website and add some CSS and graphing to make it a bit more visually appealing

How is the score made?
Hemispheric Power is a metric that is essentially potential energy to be burnt as Aurora Substorms. IMF (Interplanetary Magnetic Field) is a metric that severely effects Hemispheric Power. If Bz is negative, then that increases Hemispheric Power. 
My Aurora Score - No Lunar or Solar Influence takes hemispheric power and converts it to a percentage (200GW = 100%), and boosts the score if Bz is negative. I then tested (for 2 months) at what percentage is visibility on the West Coast, specifically the Greymouth region. Then labeled the score as such. 
My Final Aurora Score takes into account when the sunsets, sunrises, moonrise, moon set and illumination for the West Coast. 
The 30 Minute Average Scores are there to flesh out a very dynamic score. This updates every 5 minutes, and provides my website with a 30 minute average. 

Displaying the data
I tried to make this as easy to understand as possible, so having a little bit of code that tells us exactly what visibility it is expected (eg none, pro cameras, phones, eyes etc), and using emojis really helps, the happier the emoji, the better chance there is to see aurora! I have made a 3 hour graph available, which shows the viewers what's happening. Is this growing or is it dying off or is it stable? Seeing the graph will help visualize this.