Hello and rainfall estimation

[Andy Turner]

Hi Everyone,

I am new to the Group, so I thought I would send a brief message to let you know what I am trying to do with the DataPoints data. I have read a few related posts so I can see that others are trying to do something similar to me, so maybe you can help and we can work together...

I am developing an open source Java program to estimate (both observed and forecast) total daily rainfall at a 1 km resolution. The plan is that this program will run continually and pull DataPoints data within free fair use limits to get data for the region around the Teifi river catchment in Wales and the Wissey river catchment area in the Anglian region.

This work is part of a research project that is trying to estimate agricultural runoff risk in real time and help both farmers and water companies manage risk with regard agricultural chemicals that are applied to fields/crops.

In the last couple of weeks I have been getting to grips with the DataPoints API and I have developed some Java code that can pull point and image data that I can use as a basis for the estimations. The most important data is rainfall data, but in the future wind and other variables might also be used. My plan is to use the highest resolution WMTS data, but I might start with a coarser resolution until I've got all the main parts working..

The two issues I want to resolve next:

1) For the WMTS data I have to get the bounding boxes for the images. I've seen posts from others that have struggled with this. So, my plan is to use the WGS84 projected data and assume that the higher resolution data covers the same region but with the resolution effectively halved each time. I will work on these data and finally project into the OSGB projection that I want to have the total rainfall estimates for.

2) Prior to projecting the data, I have to convert the colours into numerical values and aggregate the values for daily periods. Initially I will aggregate by simply adding the values without attempting any clever interpolation. (The estimates will likely be wrong in places, especially for the 3 hourly forecasts, whereas the 15 minute observations should give a more accurate aggregate estimate. Anyway, I plan to compare the grid values around the locations where there are point observations and forecasts near the two study areas. Based on the comparison, I will rescale the values based on this. As a final step I'll reproject the data and the result will be fed into the risk estimation models.

I'll let you know how I get on in due course.

If anyone has any advice or wants to collaborate, please let me know.

Best wishes,

Andy 

[Michael Saunby]

Andy,

Have you also looked at data available from ceda.ac.uk and ecmwf.org ?  That's where I'd start These days if developing rainfall estimation.

Michael Saunby

--
You received this message because you are subscribed to the Google Groups "Met Office DataPoint" group.
To unsubscribe from this group and stop receiving emails from it, send an email to metoffice-datapoint+unsub...@googlegroups.com.
To post to this group, send email to metoffice-datapoint@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Michael Saunby]

Andy,

Make that http://apps.ecmwf.int/datasets/

Although focused on African rainfall, flooding, etc. there's much to be learned from the very long running TAMSAT project at University of Reading http://www.tamsat.org.uk

Michael

To unsubscribe from this group and stop receiving emails from it, send an email to metoffice-datapoint+unsubscribe...@googlegroups.com.

[Craig McHale]

Hi Andy,

I was wondering if you managed to get further with this? I'm trying to get similar information but to a slightly larger resolution so would be interested to know if its worthwhile continuing!

Thanks for any info

Craig