This repository contains the code for the creation of gridded fields of Density of Ocean Heat Content (dohc) in the frame of the ME4OH initiative, using the DIVAnd software tool.
DIVAnd stands for Data-Interpolating Variational Analysis in n dimensions (Barth et al., 2014). The tool performs an n-dimensional variational analysis/gridding of arbitrarily located observations. It is a generalisation of DIVA (Troupin et al., 2012; Beckers et al., 2014), which was focused on two-dimensional interpolations only.
The main advantages of the method are:
- The computation time doesn't depend on the number of observations, make it possible to work on the World Ocean at several depth or for different time periods.
- The interpolation takes into account physical boundaries, such as the coastline and the bathymetry, without having to apply a mask a posteriori.
The code is written in the Julia programming language (Bezanson et al., 2017).
The method has been applied in several European projects such as SeaDataCloud, EMODnet Biology, EMODnet Chemistry and BlueCloud. It was also applied to produced regional and local climatologies (e.g., Belgacem et al., 2021; Shahzadi et al., 2021).
While the initial developments were targeted on oceanographic data, any datasets including coordinates can be used as an input for DIVAnd.
Different input files are provided and the goal is to interpolate the observations using different techniques (inclusing DIVAnd) and compare the results with a set of ground true field.
Different configurations are applied for the computation of the gridded fields:
- The observations are interpolated layer by layer and time period by time period, i.e. we don't introduce any influence from one period to the next or from one layer to another.
- Same as 1., except that the variable
dohc_mask_by_en4_maxdepthis used to remove some of the measurements. - Same as 1., except that we now introduce the time as a 3rd dimension, i.e., the observations from the months from before and after the month of interest have an influence on the field for that month.
Note that for 3., we didn't perform any optimisation of the temporal correlation length value and set it to 30 days. The computation time is increased in this configuration (approx. 35 seconds per time step, vs. 0.3 second for the configuration without the time correlation).
Scripts:
In experiment-B we create a climatology consisting of 12 monthly fields, then perform the interpolation on the anomalies with respect to that climatology.
Again in this experiment we perform analysis with and without the use of the variable dohc_mask_by_en4_maxdepth.
The dohc product is computed in 3 steps:
- A monthly climatology of temperature is computed using all the available observations.
- Temperature anomalies (with respect to the climatology) are interpolated onto the 51 vertical levels specified in the variable
ts_z, leading to a 4-dimensional field for the time period of interest. - The ocean heat content density of computed for the 3 depth layers of interest.
Not performed yet.
Barth, A., Beckers, J.-M., Troupin, C., Alvera-Azcárate, A. and Vandenbulcke, L.: divand-1.0: n-dimensional variational data analysis for ocean observations, Geoscientific Model Development, 7, 225–241, 2014. DOI: 10.5194/gmd-7-225-2014
Belgacem, M., Schroeder, K., Barth, A., Troupin, C., Pavoni, B., Raimbault, P., Garcia, N., Borghini, M. and Chiggiato, J.: Climatological distribution of dissolved inorganic nutrients in the western Mediterranean Sea (1981–2017), Earth System Science Data, 13(12), 5915–5949, 2021. DOI: 10.5194/essd-13-5915-2021
Bezanson, J.; Edelman, A.; Karpinski, S. & Shah, V. B. Julia: A fresh approach to numerical computing SIAM Review, SIAM, 2017, 59, 65-98. DOI: 10.1137/141000671
Shahzadi, K., Pinardi, N., Barth, A., Troupin, C., Lyubartsev, V. and Simoncelli, S.: A New Global Ocean Climatology, Frontiers in Environmental Science, 9, 2021. DOI: 10.3389/fenvs.2021.711363
Troupin, C., Sirjacobs, D., Rixen, M., Brasseur, P., Brankart, J.-M., Barth, A., Alvera-Azcárate, A., Capet, A., Ouberdous, M., Lenartz, F., Toussaint, M.-E. and Beckers, J.-M.: Generation of analysis and consistent error fields using the Data Interpolating Variational Analysis (Diva), Ocean Modelling, 52-53, 90–101, 2012. DOI: 10.1016/j.ocemod.2012.05.002