Column and surface measurements of H\(_2\)O and HDO at Darwin, Australia



The added value of stable isotopes to interpret atmospheric hydrology has been of recent interest following the introduction remote sensing products, in-situ measurements and increasingly sophisticated models. In-situ measurements provide point measurements with good temporal coverage and resolution, but due to small scale of some of the processes influencing stable isotopes, it can be difficult to evaluate regional and global climate models. Satellite based remote sensing products have the potential to provide datasets with appropriate spatial scales but with poor temporal resolution, and although there have been recent attempts to assess the accuracy of these techniques only a couple use well calibrated in-situ measurements. To date only a couple of assessments of dataset accuracy have been provided using well calibrated in-situ measurements from aircraft platforms (Herman 2014, Schneider 2015). Beyond these studies assessment dataset accuracy has focused on using ground based remote sensing Fourier Transform Spectrometers (FTS), however, these have been shown to suffer from large measurement biases (Schneider 2015). There is a need to both provide validation studies of these ground based FTS measurements and to evaluate a broader range of products used to study water isotopes in the hydrological cycle.

The ground based remote sensing of water isotopes has been performed through the Total Column Carbon Observing Network (TCCON; Wunch et al., 2011) and the MUlti-platform remote Sensing of Isotopologues for investigating the Cycle of Atmospheric water (MUSICA) project using the Network for the Detection Atmospheric composition Change (NDACC) Fourier Transform Spectrometers (Schneider 2012). Schneider et al. (2015) presented an evaluation of the accuracy of the MUSICA sites was provided through comparison against in-situ measurements of water vapor isotopes from an aircraft platform. They showed large positive biases for FTS retrievals. No similar comparison with in-situ measurements has been conducted for the TCCON network, although the TCCON stations have been used to determine the accuracy of a number of satellite products (Frankenberg 2013, Scheepmaker 2015, Boesch 2013).

Large positive biases consistent for all comparisons with in-situ observations (Worden 2011, Herman 2014, Schneider 2015). Comparison between satellite SWVI products and the TCCON network have shown the satellite illustrate a negative bias (Risi 2012, Boesch 2012), whether this is caused by inaccuracy in satellite or TCCON retrievals remains uncertain. Risi et al. (2012) showed that whilst the TCCON column retrievals were enriched relative to \(\delta\)\(^2\)H from the SCIAMACHY instrument, the latitudinal gradient was preserved. Boesch et al. (2012) compared the TCCON isotope retrievals with the GOSAT satellite and showed the bias between the 2 datasets varied with season and location. They showed the GOSAT retrievals were very sensitive to surface characteristics and the shape of the H\(_2\)O and HDO a priori column. Interestingly the largest biases were observed in Darwin, Australia where complex water vapor and isotope profiles would be expected due to the strong vertical mixing in this tropical region. To better assess the errors associated with satellite based retrievals determined from these studies, the TCCON measurements would greatly benefit from comparisons against accurate in-situ observations.

These are sutiable for comparison - Ground based remote sensing

  • why is it useful (validation of satellites and can be more easily calibrated using in-situ measurements)

  • what datasets are there (NDACC and TCCON (Wunch 2010))

  • if at all how have they been calibrated.

  • random errors for total column isotope ratios can be as large as the variability (Frankenberg 2009), so also need validation

TCCON dataset
- What is TCCON – purpose, where, spectroscopy
- TCCON has been used to validate satellite remote sensing products (Scheepmaker, Boesch), but TCCON dD has not itself been validated using independent measurement techniques.
In-Situ measurements
- how can these add value (accuracy)
- difficulties of such comparisons. Worden et al. (2011) used surface in-situ measurements from on top of Mauna Loa, Hawaii to evaluate the accuracy of the Tropospheric Emission Spectrometer (TES) isotope retrievals. The top of Mauna Loa was the region the TES retrievals were most sensitive. Similarily, the TCCON observations are most sensitive to the surface (ref for averaging kernels), indicating that the surface in-situ measurements could be used to provide quantitative estimates of the TCCON SWVI accuracy.
- poorly constrained hydrology
- driving force of moisture transport
- lack of measurements
- isotopically show added value not seen elsewhere (no relationship between dD and H2O – convection) – can models and remote sensing products reproduce these effects and the seasonality
- TWP is unique area strongly influenced by MJO and pacific warm pool so is an area of intense convection during the wet season.
The Darwin TCCON site and co-located Picarro instrument provides a unique opportunity to study both how tropical hydrological processing effects stable isotopes and assess the accuracy of the Darwin dD TCCON measurements. This paper analyses the added value from isotopic measurements for interpreting tropical hydrology in wet and dry seasons. We examine the roles that different dehydration processes influence the local hydrology. In addition we use in-situ measurements to assess the accuracy of dD TCCON retrievals and and determine how uncertainty influences the intepretation of seasonal cycle and atmospheric hydrological processing. Finally we examine the agreement between the modelled surface and column datasets and determine if they are sensitive to the hydrological processes observed in the in-situ surface observations.

Investigate smoothing for TCCON dD

dD from in-situ combined with Rayleigh model for a priori profile (is it better than ocean equilibrium + Rayleigh)

Do models and remote sensing observe the same hydrological processing as in-situ?

\(\delta\)D from in-situ combined with Rayleigh model for a priori profile (is it better than ocean equilibrium + Rayleigh)

Profile Construction

Looking at the \(\delta\)\(^2\)H.H\(_2\)O plots for the in-situ data suggests Rayleigh models will not be appropriate. During the dry season the data matches very closely with a mixing line , wet season super-Rayleigh (ie much more depleted than predicted by Rayleigh) (Noone 2012), and the transitional seasons observe both processes but nothing like a Rayleigh model. These plots could be the justification for using a mixing model to represent the profile in the dry season, a super rayleigh model for the wet, and a threshold which separates these 2 models for transitional periods.

  • What do the H_2O profiles look like for super-rayleigh?

  • Do we actually have any TCCON data for super-rayleigh cases?

  • How the hell do you construct a profile with a super rayleigh model - no relationship between \(\delta\)\(^2\)H and H\(_2\)O??

  • Do models see the same?