NWP vs. MCP
In the post “Assessment 101,” I mentioned that Scott Eichelberger and his ilk get a lot of grilling while staffing the 3TIER booth, and not just about the location of the restrooms, or which booths give away the best tchotchkes. People want to know how our method, numerical weather prediction (NWP), compares to other techniques. Many of our competitors are using a method called measure, correlate, predict, or MCP. Here’s the skinny on some key differences between the two.
Both NWP and MCP address the same question: “How windy is it here?” You learned about NWP in the Assessment 101 post. MCP is a statistical method that is one of the wind industry’s standard tools, but its applicability can be limited.
Relationships Matter
MCP practitioners gather a year’s worth of data from a met tower at a wind farmer’s site, perhaps atop a ridge, then look for a nearby site with a larger data set, like an airport that’s been dutifully recording the weather. MCP makes statistical relationships between the two data sets, tailors the airport’s larger data set accordingly, then combines the year’s worth of on-site observational data with the now-statistically-corrected decade of airport measurements. The result is a representation of past weather at the wind farmer’s site.
“MCP works well if the two points are very near, but if they are separated by a wide distance or are in different wind regimes, there may be no real good correlation between the two,” Scott says. For example, “airports are usually in non-windy places, while ridge tops are usually windy.”
If, for some reason, an airport-type source of a larger data set with which to correlate the on-site observational data set doesn’t exist, MCP can’t be performed. This is a particular problem in developing nations.
Airport Not Required
NWP, on the other hand, can offer clients detailed knowledge of a site without relying on that second, larger data set. NWP can also provide assessment information even in the absence of any observational data at a prospective wind farm site, although with less certainty.
NWP has a few other things up on MCP:
- An NWP model can be run far back in time — a time before airport data collection. We can provide wind resource assessments going back 10, 25, or 40 years.
- An NWP model can be run in finer detail, time-wise. NWP can represent wind conditions from hour to hour, even minute to minute.
- NWP modeling can reveal wind resources over a broad area, to help clients look for potential wind farm sites. MCP is better-suited to understanding the wind resource at a particular site.
MOS-Covered
If you remember the Assessment 101 post, you’ll recall that we use model output statistics (MOS) to bring the output from all the computer models into line with on-site observational data. And so you’re probably wondering: How is the statistical approach of MOS applied to NWP output different from the statistical approach of MCP? It’s apples and oranges. MOS compares the computer model’s picture of apples at a particular site to the real apples on site, whereas MCP attempts to make oranges from another site look like apples at the site of interest.
Some of our clients make use of both NWP and MCP, perhaps prospecting with NWP and MCP-ing once they’ve decided on a site, or in concert with each other if their due diligence process is particularly diligent.
Tags: Assessment, measure-correlate-predict, numerical weather prediction
Author: Leah Kauffman
