Geostatistical merging of weather radar data with a sparse rain gauge network in Queensland

Sharp, C., Schepen, A., Das, S., and Everingham, Y. (2021) Geostatistical merging of weather radar data with a sparse rain gauge network in Queensland. In: Proceedings of the 24th International Congress on Modelling and Simulation. pp. 644-650. From: MODSIM 2021: 24th International Congress on Modelling and Simulation, 5-10 December 2021, Sydney, NSW, Australia.

[img]
Preview
PDF (Published Version) - Published Version
Available under License Creative Commons Attribution.

Download (1MB) | Preview
View at Publisher Website: https://doi.org/10.36334/modsim.2021.K6....
 
46


Abstract

Many parts of Australia, including much of Queensland and Northern Australia, tend to have sparse rain gauge coverage. To provide rainfall information across Australia, several gridded daily rainfall datasets such as those available through the Australian Water Availability Project and Scientific Information for Land Owners services have been developed. These daily grids are produced by interpolation of rain gauge data and therefore can provide unrealistic rainfall estimates in areas that have few rain gauges. To obtain rainfall data at a higher spatial resolution, weather radars and satellites can provide coverage over a large area although their measurements come with considerable uncertainty.

Various approaches have been developed to adjust radar and satellite data and statistically merge them with rain gauge measurements in interpolation schemes, the goal being to retain the information on the spatial distribution of rainfall provided by remote sensing while also taking advantage of the greater accuracy of the rain gauges, but many of these techniques have been applied primarily on shorter time scales of an hour or less. This paper applies some existing methods for geostatistical merging of radar data with sparse rain gauge networks and evaluates the performance of the approaches using the Mt Stapylton radar in Brisbane and 15 surrounding rain gauges. Summer and winter data from 01/12/2013 to 28/02/2018 are considered. The radar data is corrected for mean field bias using quantile mapping and is used to develop the variogram models for use in Kriging. The performance of Kriging the gauge data using the radar variogram is compared with conditional merging and Kriging with radar values introduced as a drift variable. Leave-one-out cross-validation is used to evaluate the performance of the methods.

We find some disagreement between all radar-based approaches and the validation gauge measurements with typical daily root-mean-square errors being between 10mm and 20mm for all approaches. Some outliers with substantially higher RMSE are noted for some days in the unadjusted radar data as well as in the corrected and interpolated data. For winter data the bias-correction and interpolation steps increased the agreement between the radar data and the validation gauges, but this improvement was not observed in the summer data. In addition, due to the low number of gauges the performance of the interpolation is extremely sensitive to the rain gauge values, with certain combinations of rain gauge values and choice of validation gauge leading to extremely large cross-validation errors. The results indicate that while incorporating the radar data makes it possible to perform Kriging with few gauges ona single day's data, this is not an ideal approach for quantitative precipitation estimation and further steps should be taken to improve the radar-gauge correlation.

Item ID: 73119
Item Type: Conference Item (Research - E1)
ISBN: 978-0-9872143-8-6
Keywords: Weather radar, Kriging, rainfall, quantile mapping
Related URLs:
Copyright Information: These proceedings are licensed under the terms of the Creative Commons Attribution 4.0 International CC BY License (http://creativecommons.org/licenses/by/4.0), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you attribute MSSANZ and the original author(s) and source, provide a link to the Creative Commons licence and indicate if changes were made. Images or other third party material are included in this licence, unless otherwise indicated in a credit line to the material.
Date Deposited: 30 Mar 2022 03:34
FoR Codes: 37 EARTH SCIENCES > 3701 Atmospheric sciences > 370199 Atmospheric sciences not elsewhere classified @ 25%
49 MATHEMATICAL SCIENCES > 4905 Statistics > 490501 Applied statistics @ 75%
SEO Codes: 28 EXPANDING KNOWLEDGE > 2801 Expanding knowledge > 280118 Expanding knowledge in the mathematical sciences @ 50%
28 EXPANDING KNOWLEDGE > 2801 Expanding knowledge > 280107 Expanding knowledge in the earth sciences @ 50%
Downloads: Total: 46
Last 12 Months: 2
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page