AceCAST 5: what's new
AceCAST 5 is available now. Here is what changed: WRF 4.8, four physics updates and four new model outputs.
A small miss in forecast wind speed is a large miss in power. What wind and solar operators need from a forecast, what local detail costs to compute, and what you can use today.

For a wind or solar operator, weather is both the fuel and the main hazard. A forecast that misses tomorrow's wind by a little can miss tomorrow's power by a lot, and the difference is paid for in reserve power, imbalance charges or curtailed output. A storm that arrives unannounced can damage equipment. Both problems come down to the same need: a forecast that describes the weather at the site, not the average over the region around it.
A wind turbine's output follows a power curve. Below a cut-in speed it produces nothing. Above its rated speed it produces a steady maximum. In between, power climbs steeply.

The steep part is where forecast skill pays. Researchers at the National Renewable Energy Laboratory and GE Energy put it this way in The Value of Wind Power Forecasting (2011):
"For typical pitch-controlled wind turbines, power output varies as the cube of wind speed over a significant portion of the power output curve."
A cube law magnifies errors. On that part of the curve, a wind speed 10 percent higher than forecast means about 33 percent more power than planned. The same paper noted that day-ahead wind forecasts of the time typically had a mean absolute error of 15% to 20% for a single wind plant. That figure is from 2011.
A forecast for a wind or solar site has four jobs.
Power system operators forecast demand for each hour of the next day and commit generation to meet it. The wind forecast is combined with the load forecast to decide how much other generation to schedule. The NREL paper describes what goes wrong in each direction. If the wind forecast is too low, too much conventional generation is committed, it runs less efficiently, and wind may have to be curtailed. If the forecast is too high, operators turn to quick-start peaking units, which cost more to run.
Turbines sit 80 to 160 m up. Wind at that height can differ a great deal from the standard 10 m forecast, especially at night, when a fast layer of air can form aloft while the surface stays calm.
Turbines shut down above a cut-out wind speed to protect themselves. Hail and lightning can damage blades and solar panels. A few hours of warning lets an operator plan crews and maintenance around the weather.
For solar, cloud is the variable that matters. Low cloud and fog that a coarse model smooths over can decide a morning's output.
A weather model divides the atmosphere into grid cells, and each cell holds one value of wind. Grid spacing limits the detail a forecast can represent. Our article on local weather prediction explains what that hides.
Finer grids need more computing. AceCAST, TempoQuest's product, addresses the hardware side: it runs WRF, the Weather Research and Forecasting model maintained by NCAR, on NVIDIA GPUs.
Two published sources give measured results for earlier AceCAST releases.
These are measurements of earlier releases on particular hardware, not promises. Work out the cost for your own case.
AceCAST 5 is available now: see what is new.
Fine-grid detail is not proof of forecast skill. See how to score a forecast, including a bounded historical comparison with lidar observations.

AceCAST 5 is available now. Here is what changed: WRF 4.8, four physics updates and four new model outputs.

Aerava runs a weather model around the site you pin and checks the result against your own storm, wind and turbine cut-out limits. Here is what you can run today, what it costs and what it says about accuracy.

One published 2023 benchmark and two Azure list prices, worked line by line: what a GPU VM costs against one CPU VM and against sixteen, and what the sums leave out.
Your forecast
Tell us the decision it has to serve. One short form reaches the team, and we review every request.