Wildly incorrect temperatures and rain forecast
-
We have been travelling down the Queensland coast (Australia) for the last 2 months visiting several Islands just off the coast. Ive noticed each time we are on an island Windy has shown a strangely fixed temperature range of 22max to 20 min. This has been completely wrong based on our lived experience on the island and totally at odds with the BOM (Australian Bureau of Meteorology). For example again here at Eurong on K'gari, Windy has said max 21-min 20 for the last few days and no rain. Well the last 2 dsts were in fact max25 min 15 and it just rained this morning. BOM says 24-16. And predicted today's rain for the last few days. Why is Windy so far out of whack? Something isn't right.
-
@RustyRig Can you please provide further info about the behavior? Can you send any screenshots from Windy and link to other sources, that could help us to check the situation?
-
@RustyRig said in Wildly incorrect temperatures and rain forecast:
Windy has said max 21-min 20 for the last few days and no rain.
Windy visualises weather models and the default model is the one of the ECMWF, with a 9km resolution. On a small island, with such a distance between grid points (the model resolution) some of these points are over the sea, that explains why the temperature doesn’t vary during the day as it varies above the land. The local forecast is interpolated between the grid points to show a weather map. So a high resolution model should be necessary in these conditions.
You could try ACCESS, a weather model from the BOM, but in Windy its resolution is 12 km, so it is probably not better over the same island. The BOM has higher resolution models but they are not used in Windy.
In Windy you could try METEOBLUE, this model is not gridded and it gives better temp accuracy over small islands. -
@idefix37 Pardon for jumping in at this discussion. Is the interpolation btwn points done at Windy based on the on-grid values provided by the forecasting entity (ACCESS, ECMWF, whatever)? I see Eurong is right on the eastern shore of a long narrow island facing the ocean and indeed both ACCESS & ECMWF forecast very narrow 2degC fluctuation btwn day and night, whereas just 4km West the fluctuation expands to 7degC. But MeteoBlue has this large fluctuation right on the shore, and--strangely--even 24km into the sea.
-
@hangInThere
Windy interpolation is applied between grid points based on GRIB files provided by the weather model. You know that forecast parameters are supplied in this format files on a grid at the resolution spacing. It is the reason why the temperature near the sea shore is an weighted average between temperature over the land and over the sea. So interpolation is smoothing the reality, but it allows to produce weather map.
Meteoblue is not a non-gridded model. The parameters like temperature is based on a Multimodel using the nearest grid point of several weather models. So this model is not able to plot a weather map as it provides discrete values. Note there is no Meteoblue map in Windy.
Furthermore, as I use to sail at sea, I discovered that the wind speed at short distance of the coast was lower by Meteoblue than that predicted by each individual model. Meteoblue explained to me that on seashore their model considers the nearest grid point on the land but not over the sea, probably because there are much more people wanting to know weather forecast on land rather over the sea. So for temperature on land near the sea, Meteoblue provides a better estimation than an interpolation between sea and land, but worst over the sea near the coast. Hope it helps and explanation is clear. -
I idefix37 referenced this topic on
