precipitation-VOD-ISV

Code for Harris et al. "Satellite-Observed Vegetation Responses to Intraseasonal Precipitation Variability"

This repository contains code to reproduce the analysis and figures of Harris et al. (2022) Satellite-Observed Vegetation Responses to Intraseasonal Precipitation Variability. Geophysical Research Letters 49(15) e2022GL099635: https://doi.org/10.1029/2022GL099635.

Note that this analysis relies upon Fortran code containing an algorithm for the Lomb-Scargle periodogram, from:

Weedon et al. (2015). Evaluating the performance of hydrological models via cross-spectral analysis: case study of the Thames Basin, United Kingdom. Journal of Hydrometeorology, 16(1), 214--231.

This code is not included here, but it is freely available for to anyone to use from the Met Office Science Repository Service at https://code.metoffice.gov.uk/trac/lmed/browser#main/trunk/benchmarking (last access: 29 June 2022). Access requires registration for an account and this will be supported by a member of the JULES group. Requests for new accounts can be made by emailing [email protected] with details of the user’s name, email address, institution and purpose for requiring access.

Cross-spectral analysis

The code can be used to perform the cross-spectral analysis described in Section 2 of the paper as follows:

• Obtain the Fortran code containing the Lomb-Scargle periodogram algorithm from Weedon et al. (2015) (csagan1.1.f from the Met Office repository described above).
• Enable multiprocessing with csagan_multiprocess.py following steps in multiprocessing_setup.txt
• Compile csagan (requires linking to netCDF libraries)
• Run bash script csa_multiprocess_tiles. This will run the cross-spectral analysis for all pixels/seasons, saving the results in three latitude-band tiles for each season. The save directory is set in csagan_multiprocess.py. A single season takes approximately 8 hours to process and produces ~3--5 GB of saved data.
• Run significant_coherent_intraseasonal_relationships.py for each tile ('tropics', 'northern', 'southern'), season and frequency-of-variability band. These variables have to be set at the top of the .py file before running. This computes and saves details of the coherent relationships that are 95% significant based on the 3-neighbour condition described in Section 2 of the paper.
• Run lag_subplots_calc.py to prepare data for Fig 2.

Figures

• Figure 1: produced by running csa_illustration.py (requires Lomb-Scargle Fortran code)
• Figure 2: lag_subplots_plot.py
• Figure 3 and Figure S6: lag_by_land_cover_calc.py
• Figure 4: first run isv_events_surface_water_tiles.py to save standardised anomalies of the observed variables and save the dates of intraseasonal precipitation events at each pixel. Then isv_composites_surface_water_filter_tiles.py generates the figure.
• Figure S1: Created using function filter_pixel() in vod_surface_water_filter_best.py with kwarg demo_figure=True
• Figure S2: inundation_mask_maps.py. This also saves data on the number of valid observations before and after inundation masking, which is required by lag_by_land_cover_calc.py to plot Figure S6.
• Figure S4: save_example_pixel_time_series.py saves the time series at the example locations to file. Then plot with example_pixel_time_series_plots_multiyear.py
• Figure S5: percent_rainfall_variability_in_isv.py

Other files

• vod_filter_monthly_global.py is used to mask random single-timestep spikes in the VOD data, as described in Section 2.1
• vod_surface_water_filter_best.py masks inundation events from the VOD data, as described in Section 2.1
• read_data_iris.py contains functions to conveniently read in IMERG/VODCA/SSM/SWAMPS data from file, assuming the dataset is stored in yearly files
• read_csagan_saved_output.py contains helper functions to read the output produced by csa_multiprocess.py