Chapter 6 Part two (Research Degree)

First, validation of precipitations derived from space-based observations with in situ data will be performed using rain gauge data from archives of BoM, to examine accuracy of satellite remote sensing data over Australia. Specifically, performance of satellite precipitation estimates in quantifying the extreme rainfall events measured over different spatial scales (e.g. 0.25deg, 0.5deg, 1.0deg etc.) will be examined. Following on the relations between heavy precipitation and floods will be examined.

There are some issues; Satellite data is gridded in 0.25deg, 0.5deg, 1deg, there are numerous satellite products, the TRIMM, PERSIANN and CMORPH from a constellation of satellites collected in different intervals (3hrs to 24hrs). There are eight metrics that will be used

Statistical Index Equation Perfect Value Probability of Detection (POD)

  • Frequency of Hit (FOH)
  • False Alarm Ratio (FAR)
  • Critical Success Index (CSI)
  • Heidke Skill Score (HSS)
  • Relative Root Mean Square Error (RRMSE)
  • Correlation Coefficient (CC)
  • Relative Bias mean(RS)

6.1 Data Sources

Data will be sourced from organizations such as the Bureau of Meteorology, Japanese Aerospace Explo- ration Agency and the National Oceanic Atmospheric Agency; Data will be provided in the form of high resolution satellite imagery and data from approximately the last 20 years.

High-resolution (0.25deg by 0.25deg) gridded daily precipitation dataset obtained from The NOAA CPC Mor- phing Technique (CMORPH) will be used. The dataset covers the period of 2014-2018. This dataset is available at ftp://ftp.cpc.ncep.noaa.gov/precip/global CMORPH/daily 025deg/

The In situ precipitation dataset was obtained from The Bureau of Meteorology in Australia from approximately 7000 rain gauge stations across Australia. The rain gauge precipitation data was collected on a daily basis and it covers the period from 2000-2018. This dataset is available at http://www.bom.gov.au/climate/data/index.shtml?bookmark=136 These datasets have been utilized to firstly validate remote sensing data against in situ data. Matching In situ rain gauge stations with (0.25deg by 0:2deg) gridded daily precipitation satellite observations was done using the Haversine formula.

6.2 Mathematical Techniques

Our Data is the Spatial and Temporal variability in extreme precipitations in Australia.

  • There are issues as quanitfying what is considered extreme (here we take rainfall at or above 99 quantile for that area in duration, intensity and both)
  • The data is Zero Inflated, so traditional techniques in time-series analysis will fail
  • Zero Inflated models are techniques that will be used starting with the exponential family (Poission, Negative Poisson etc)

6.3 Data set source and description

6.4 Data pre-processing and preliminary analysis

6.5 Discussion