This post outlines a process chain for comparing the variations between rainfall and Vertical Water Balance (VWB) on the one hand and the Earth´s land water content on the other. The results are presented in the next post.

Prerequisites

To follow the processing steps in this post you must set up a Spatial Data Integrated Development Environment (SPIDE), download and install Karttur’s GeoImagine Framework from Github. You must also have completed the processing of Tropical Rainfall Measurement Mission (TRMM) rainfall, the VWB, and the equivalent water depth from the Gravity Recovery and Climate Experiment (GRACE), as described in earlier posts of this project.

Drivers of Earth’s water storage

In the water cycle, precipitation is the ultimate source for the water content in land areas. Not all precipitation that falls contributes towards filling the Earth’s water content. A fraction of the precipitated water is directly transferred back to the atmosphere by evapotranspiration. Theoretically the Vertical Water Balance (VWB) should thus be a better predictor for the variations in the equivalent water pillar depth compared to gross precipitation.

In this section, the correlation between precipitation and VWB as drivers (masters) and the equivalent water pillar as a slave is tested using graphics and local (per cell) cross correlation. The correlations ignore the redistribution of water over land from upstream to downstream.

Process chain

The principal steps for creating estimations, maps and animations relating water input (precipitation and VWB) to water content (equivalent water pillar depth) over Africa using Karttur’s GeoImagine Framework include:

Overlay comparison
Time series processing
Mosaic
Export media

In the Framework a process chain can be built as a series of calls to xml coded instructions. This section contains the calls and the remaining parts of the post details each called xml. As noted above, the results are available in the next post.

Overlay comparison

As part of the processing of TRMM rainfall, VWB and GRACE equivalent water pillar depth the absolute changes of each has already been calculated. Here, a special overlay process is applied comparing two overlapping layers depicting changes or trends. The overlay identifies the congruence between the two change layers in 9 classes:

			1st layer
		Increase	No change	Decrease
	Increase	both increasing	1st no change 2nd increasing	1st decreasing 2nd increasing
2nd layer	No change	1st increasing 2nd no change	No change in either	1st decreasing 2nd no change
	Decrease	1st increasing 2nd decreasing	1st no change 2nd decreasing	both decreasing

The overlay can be done with any two change layers and here the overlay comparisons are done for layers including changes for all regions, and layers with only significant changes.

TRMM vs GRACE

VWB vs GRACE

Time Series Processing

The time series process is restricted to cross correlations between monthly variations in a master (rainfall or VWB) driving a slave (equivalent water pillar depth). The cross correlations are done at pixel by pixel time series. Following the results from comparing different smoothing algorithms cross correlating climate indexes and TRMM, the layer to layer cross correlation was done using a 12 month naive kernel. The allowed lag was restricted to slave responses following 0 to 6 months after the master signal.

Precipitation as driver of water content

VWB as driver of water content

Mosaic

In this projects, the mosaicking is only done for the exports in the next step. In the mosaic process, the tiles are first concatenated and then cut to the actual coordinates of the defining region. Cell values and data type remain the same, but the data can be reprojected on the fly.

Mosaic TRMM vs GRACE cross correlation

Mosaic VWB vs GRACE cross correlation

Export Media

The main reason for exporting the mosaicked layer is to allow visualization of both the data and the results.

Export TRMM vs GRACE cross correlation

NOT DONE

Export VWB vs GRACE cross correlation