Read on Twitter
Let’s talk about aircraft observations (obs) and numerical weather predictions. A recent letter ( https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2020GL088613) asks some useful questions about the extent to which weather forecast accuracy decreased due to less planes flying. But were weather forecasts really (1/)
worse and can we quantify how worse they were (and where)?
First, some background: aircraft obs are highly important for weather predictions, as are radiosondes. For global predictions, satellites are collectively the highest contribution to accuracy, in part because there (2/)
First, some background: aircraft obs are highly important for weather predictions, as are radiosondes. For global predictions, satellites are collectively the highest contribution to accuracy, in part because there (2/)
are so many of them.
While satellite obs are fairly evenly spread (in space and time) around the world, the same cannot be said for aircraft and radiosonde (weather balloon) obs. Radiosondes need a place to launch from, which usually means land, and airplanes are not (3/)
While satellite obs are fairly evenly spread (in space and time) around the world, the same cannot be said for aircraft and radiosonde (weather balloon) obs. Radiosondes need a place to launch from, which usually means land, and airplanes are not (3/)
profiling the atmosphere (landing) over the ocean either, even though they may provide upper-air (jet stream) information.
Aircraft are also equipped with different sensing packages. There are more aircraft that measure and report wind than moisture. In addition, even though(4/)
Aircraft are also equipped with different sensing packages. There are more aircraft that measure and report wind than moisture. In addition, even though(4/)
there has been less air traffic over the United States since March, there was still much more air traffic than over the oceans and remote areas. Those logistics companies have been busy!
Where obs are dense, the value of each ob tends to be less. It is a bit analogous to (5/)
Where obs are dense, the value of each ob tends to be less. It is a bit analogous to (5/)
supply and demand. Meteorologists tend to need less additional obs to make an accurate forecast where they already have a lot.
Obs are also complementary. Even though satellites and aircraft do not provide the same type and quality of information, they provide similar (6/)
Obs are also complementary. Even though satellites and aircraft do not provide the same type and quality of information, they provide similar (6/)
measurements. Removing the aircraft obs from a numerical model will result in more reliance on satellites and radiosondes.
In addition, more obs around a certain city or within a country doesn’t mean that the 7-day forecast will get better there. Mid-range forecasts depend (7/)
In addition, more obs around a certain city or within a country doesn’t mean that the 7-day forecast will get better there. Mid-range forecasts depend (7/)
on the obs of the weather that is upwind. The weather of the eastern United States now is largely a function of weather over the Atlantic and Pacific Oceans a week ago.
Now, back to the decrease in aircraft obs: Can we easily compare the error to the past years and make a (8/)
Now, back to the decrease in aircraft obs: Can we easily compare the error to the past years and make a (8/)
judgment?
Maybe, but it requires all of the other variables and configurations to be constant, the baseline to be extensive, and the impact to be substantial.
The best opportunity to determine the value (impact) of a certain type of ob is when you have complete set of obs. (9/)
Maybe, but it requires all of the other variables and configurations to be constant, the baseline to be extensive, and the impact to be substantial.
The best opportunity to determine the value (impact) of a certain type of ob is when you have complete set of obs. (9/)
Then you can run a series of models to see how accurate the resulting prediction was – both with and without the ob under assessment (aircraft or otherwise).
This is known as a data denial experiment. Without that, there may be anecdotal evidence that forecast accuracy is (10/)
This is known as a data denial experiment. Without that, there may be anecdotal evidence that forecast accuracy is (10/)
decreasing, but it is quite difficult to quantify. This is because different weather scenarios are simply more difficult to predict than others. The time scale of the prediction can also have an impact because forecasts become less accurate with time.
Where would I look to (11/)
Where would I look to (11/)
see the possible impact of less aircraft obs? First and foremost, the southern hemisphere (SH). There is less land down there (less radiosondes), so aircraft obs have greater opportunity for impact. Weather predictions in the SH tend to already be less accurate than the NH(12/)
but the gap is lessening.
In the NH, I would suspect that parts of Canada may have been prone to accuracy decreases due to less transoceanic routes between Asia-NorthAm and NorthAm-Europe. Look at those flight tracking sites to see where planes fly – and where they do not. (13/)
In the NH, I would suspect that parts of Canada may have been prone to accuracy decreases due to less transoceanic routes between Asia-NorthAm and NorthAm-Europe. Look at those flight tracking sites to see where planes fly – and where they do not. (13/)
Some countries in Europe also started launching radiosondes two additional times per day. What impact did this have? The US increases the number of balloon launches when there are hurricanes that have the potential to make landfall.
Aircraft obs returning can only be a good(14/)
Aircraft obs returning can only be a good(14/)
thing for forecast accuracy. The summary of science here is that all obs have their value but the global investments in weather satellites stabilize weather prediction accuracy for forecasts several days ahead. Standard microwave channels like 24 GHz on satellite instruments(15/)
