NbS: Analyse drought risk

When identifying the risk of non-permanent, an important natural risk to consider is the frequency of droughts.  This block calculates the number of drought years that have occurred within a pixel area (11km square) over the previous 40 years.  You can define what threshold defines a drought year (as a percentage of the mean annual rainfall for the pixel).  The example area is Cambodia, but you can choose whatever country (or area) is of interest. 

TABLE OF CONTENTS

• Demonstration Video
• What should I use this template for? 
• What data has been used? 
• What do the outputs show me? 
  • Map Outputs 
  • Dashboard Outputs 
• What are the blocks used in the workflow?

Demonstration Video

What should I use this template for? 

You should use this template project to look at historical drought trends for a given location. This could either be a large area (as in this example, where we are looking at Cambodia) or in one or more project sites (here we are looking at 5 different project sites). Then drought can be calculated based on average rainfall since 1981. 

A drought can be defined in different ways, depending on the country or region.  In this example template, the default is set at 75%. That is, a year is categorised as a drought year if the annual rainfall is less than 75% of the average annual rainfall since 1981. The calculator block allows you to define your own threshold. 

What data has been used? 

This template uses the ERA5-Land climate dataset. It is a reanalysis dataset providing a consistent view of the evolution of land variables over several decades at an enhanced resolution compared to ERA5. 

ERA5-Land has been produced by replaying the land component of the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 climate reanalysis. Reanalysis combines model data with observations from across the world into a globally complete and consistent dataset. Reanalysis produces data that go several decades back in time, providing an accurate description of the climate of the past. This dataset contains 50 climate variables. The data used here is a subset of the full ERA5-Land dataset post-processed by ECMWF. 

This dataset does not make future predictions regarding changes to precipitation but rather allows you to see historical data to understand the long-term trends for your particular location.

What do the outputs show me? 

Map Outputs 

This template project has only one map output layer. This shows the total number of drought years for each pixel (since 1981).  The colour scale will default to min-max so that the darkest red corresponds to the highest number of drought years. See the legend to determine what that value is.  You can adjust the scale using the setting button on the Add Map Layer block.  

Dashboard Outputs 

For this template, the dashboard tab will show a table with the mean value (of the pixels) for each project area.  You can export these values as a CSV file. 

What are the blocks used in the workflow?

Here we will go into more technical detail about what the workflow is doing.

The first block brings in the climate data set that contains information on monthly precipitation on an 11km grid. Since this data set contains lots of bands, we use a Filter block to select only "total_precipitation" so that we don't waste computer resources analysing all the other bands. 

The following two blocks then select the area and the time period. 

This is then followed by the first of two Aggregate Images in Time blocks.  This one collects the data together into individual years and calculates the mean for each year. The data is monthly, but we want to look at annual trends, year on year, not monthly. 

The first of the two calculator blocks define two variables. One is just the annual mean for each year as calculated by the first aggregator block ("YEAR").  The second is the mean over all years ("MEAN").  By calculating 100*YEAR/MEAN it calculates the annual precipitation for each year as a percentage of the mean across the entire period.  Note that this is calculated for each year and each pixel.  The results are saved as a new band: "THRESHOLD-PRECIP".

The second calculator block uses a logical operator ("<" meaning "less than) to set a threshold.  The default threshold is 75, meaning 75% of the average. This means that for each year, a pixel that meets these criteria (<75 is TRUE) then the value of the pixel is set to 1.  Otherwise, it is set to zero. 

The second Aggregate Images in Time block then combines every year of threshold data using the sum.  That is, it adds up all the years for each pixel, where 1=drought year, and 0=not a drought year.   The final data layer is therefore the total number of years of drought.  Each pixel value represents the number of years that met the drought threshold over the time period selected.

Warning:  When the long-term average for a pixel is very low, then even small magnitude fluctuations can cause big variations in the percentage rainfall compared to the average for a given year.   This method is therefore not optimum for areas of low annual precipitation.