Modelling Rain to River over Africa - A Paper Review

By Chris Skinner @cloudskinner

I recently had my first research article published - "Hydrological modelling using satellite rainfall estimates in a sparsely gauged river basin: The need for whole-ensemble calibration". It has been accepted by the Journal of Hydrology, with which I am very pleased, and it is available to view for free until the 27th February 2015 here. If you are reading this after that date, I'm afraid you will need a subscription to the Journal to view it.

The problem the project was hoping to address is the issue of a lack of equipment available in many parts of the world which records rainfall. There are several methods of doing this, which I explain in an older post, but the most common ways are to use a network of rain-gauges or radar, both of which are expensive to install and maintain. For many nations, the measurement of rainfall is not a priority enough to invest in these networks but they would benefit greatly from having reasonable estimations of how much rain has fallen - it allows them to monitor water resources, forecast floods and droughts, and even predict how many crops will grow.

A Map of the Senegal River Basin. The rain-gauge network used for the study covered this wide area, yet the hydrological modelling focused on the Bakoye catchment (area in the south-east, containing both the Bakoye and Baoule rivers). (Image by Kmusser) 

"Senegalrivermap" by Kmusser - Own work, Elevation data from SRTM, drainage basin from GTOPO [1], all other features from Vector Map.. Licensed under CC BY-SA 3.0 via Wikimedia Commons - http://commons.wikimedia.org/wiki/File:Senegalrivermap.png#mediaviewer/File:Senegalrivermap.png

My research focused on a section of the Senegal River Basin, a large area with only 81 rain-gauges across the wider region with which to estimate rainfall, a density of 1 gauge per 7,000km². The density of the network is the equivalent of covering the UK with only 27 rain-gauges. There would be no way we could capture the complexity of our rainfall with so few gauges, and in reality the UK maintain a network of over 4,000 gauges with a density of 1 gauge per 76km². In addition to this, the UK also maintain 15 rain recording radar stations, yet none are available to the Senegal Basin region.

Currently, the best way around this and to fill in the gap is to use information from satellites. For continuous observation of the same area it is not yet possible to directly observe rainfall, but it is possible to monitor other factors that indicate rainfall, particularly the temperature of clouds. Unlike the UK, this area of Africa generally only receives one type of rainfall, so there is a strong relationship between the temperature of the tops of clouds and whether they are raining - if the cloud is below a specified temperature it is assumed to be raining - the longer it remains below that threshold, the more intense the rain is thought to be. This is all calibrated against the very little rain gauge data that is available.

Example of a processed satellite image showing the number of hours each pixel of the image is below the specified rainfall (here it is -20c). The number of hours is known as the Cold Cloud Duration (CCD) and this is related to rainfall. The rain gauges used for the project are represented by the white circles.

Unfortunately, as this is not directly observing rainfall it can be wrong on occasion, and especially so when a user tries to make use of it at a smaller scale than the density of the rain gauge network. When the estimates are used as information for models, such as a hydrological model to estimate river flows, the errors in the rainfall estimates are passed on to that model. What we do know, however, is just how wrong the rainfall estimate could possibly be and this allows us to try and represent this.

The increasingly common way of representing this, what scientists call uncertainty, in the rainfall estimate is to take the value of how wrong it might be and to randomly produce several hundred different versions of the possible rainfall - each different but equally possible based on the information available to us. The users takes this ensemble rainfall estimate and feeds each member individually into the hydrological model and produces an ensemble of river flow estimates. Statistics can be used to suggest the probability of river flows for each step in the record.

View of the across the River Senegal in the Kayes region, downriver and north of the study site. (Image by Bourrichon)

This is where my research came into the process. Hydrological models need setting up before they are used and this is done via a process called calibration. You need a period of the record with rainfall data and recorded river flow data, and you vary adjustable values within the model (these are called parameters) until you get the closest match between the recorded river flows and what the model estimates. You then test this against another period of recorded river flow that was not used in the calibration to test its performance (known as validation or verification).

It is good practice to calibrate a model using the same data you intend to drive the model with. For example, the model will not perform as well if you calibrate it using rainfall data as estimated by rain gauges, but subsequently run the model using satellite rainfall estimates. This poses a particular problem for when you intend to run a hydrological model using ensemble rainfall estimates. This has previously been performed using either using the original satellite estimate (disregarding its uncertainty), or an average derived from the ensemble members. However, I tried the calibration using all the ensemble members individually, but assessing the model performance with them as a whole - this is called whole-ensemble calibration and named EnsAll in the paper.

Graph showing the mean error from the Pitman model, run separately with each ensemble member using data for the period 1997-2005. The model was calibrated using data for 1986-1996 using the whole-ensemble method (EnsAll), the daily mean rainfall estimate from all the ensemble members (EnsMean), and the theoretical mean of the ensembles (EnsExp). EnsAll clearly produces less error than the other two calibration methods on this measure.

The parameter values produced using the whole-ensemble calibration produced more accurate river flow estimates from the ensemble rainfall estimates than those produced by the other methods. In fact, the whole-ensemble calibration was the only method to produce 'reliable' estimates during validation, with the other methods proving no better than making an educated guess based on the rainfall record for the period.

I do hope that this provides you with a better understanding of the paper. If you want to know more of the technical details it is all in there, along with some nice statistics and some graphs to make Dave Gorman weep with joy. The research represents only a very small facet of the problem, which will require many more small facets to solve rather than one big one. I hope to produce a few more.

The paper can be viewed for free until 27/02/2015 using this link. The reference for the paper is - 

Skinner, C. J., Bellerby, T. J., Greatrex, H., and Grimes, D. I. F., 2015. Hydrological modelling using satellite rainfall estimates in a sparsely gauged river basin: The need for whole-ensemble calibration. Journal of Hydrology, 522, 110-122

Researcher Profile - Chris Skinner

Ok, here at GEESology we have decided to tell you all a bit about ourselves and to do this in the form of a ‘Researcher Profile’. For some reason I have drawn the short straw, to put it cynically, and have to go first. The flip side of the coin is that in going first I can set the benchmark for everyone else and have a fairly free hand in doing so. I guess it is really a chance for us all to share a little bit about ourselves and what is behind our research, in particular what motivates us and why we do it. Each of us will provide one of these posts to you over the coming weeks and months, so without any further ado, here’s me –

Who am I

I am Chris Skinner, currently working as a Research Assistant as part of the Dynamic Humber project at the University of Hull. My role is develop the CAESAR-Lisflood model for operation on the Humber Estuary with the aim of providing forecasts of changes in the estuary for the coming century.

Me before my remote sensing days

What I do and Why do I do it

Last year I completed my PhD research that looked into the effects of uncertainty in satellite rainfall estimates on hydrological models. These estimates are vital in Africa, where there is a real lack of raingauges and radar that we use in the UK to predict rainfall, but as they cannot directly record rainfall they are often a little bit wrong. This in turn affects the models that are used to forecast droughts and floods. This chance of being wrong is termed by scientists as ‘uncertainty’ and this has a major impact on the people who have to make decisions.

“There are known knowns; there are things we know we know.

We also know there are known unknowns; that is to say we know there are some things we do not know.

But there are also unknown unknowns – there are things we do not know we don't know.” 

Donald Rumsfeld perfectly, although unwittingly, describing the nature of uncertainty. 

Uncertainty leads to a lack of confidence and can mean that important decisions that influence millions of people can be delayed, sometimes at the cost of people’s lives. A recent example of this was the Horn of Africa drought in 2011, which was forecast several months before any aid began to be mobilised. My research interests are in looking at ways to either reduce the uncertainty, measure it better or just communicate better to people who have to make the difficult choices – I blogged about this on my (largely defunct) personal blog over two years ago in Why do we bother?

How do I do it

How I do this is by using a lot of statistics and numerical computer models that are far too complex (and not all that interesting enough) to be talked about in detail here, but the main method I use to show uncertainty is by using ensembles of forecasts – a set of possible futures, each equally likely yet different, within the bounds of what we don’t know. From this you can produce what is known as a probabilistic forecast. It’s the difference between Michael Fish telling you there is absolutely no chance of being hit by a hurricane, and him telling you there’s a 30% chance – subtle difference but results in different (and probably better) decisions.

How did I get here

Short answer, I walked. That’s very important, as my job before I started my PhD was as a Sustainable Transport Policy Assistant at a local authority in the Midlands, and a large part of my job was encouraging people to walk and cycle more. It was fun job on the frontline, getting to organise events such as bike rides, but I did not like the look of the career ladder ahead of me. I wanted to stretch my mind so in 2009 I decided to quit my job and focus on a career in Academia.

Sustainable Transport - It can be dangerous!

At this point I hadn’t chosen a discipline, I just wanted to do something that looked like it might help people and make a difference. In the end I got the perfect PhD back at Hull, which is where I got my undergraduate degree and close to where I grew up and my family live. I’m pleased to still be a part of such an excellent department but I know one day the Academic career will draw me away to pastures new.

Wow, 500 words isn’t a lot – I never got to tell you about the time I spent in the nappy factory, the garlic bread factory, painting student houses, data entering, on Job Seekers, as a Geotechnical Laboratory Technician or in the Planning Department...

Getting Animated

Getting Animated by Chris Skinner (@cloudskinner)

The formal presentation of research in academia is pretty traditional. I doubt it has changed much in the last 500 years, if not longer, and for a progressive sector of society it really does not look set to change. Basically, you get your results, write it up as a paper, some experts look it over and request more details or changes, you do them, they pass it, you get published.

The published article then goes into a journal. Most of these are still printed but are available, usually as a PDF file, electronically. This is where the embrace with the modern world ends. I mainly read articles either on my computer or my tablet – most articles are formatted into two columns on a page which makes it very awkward to read off a screen. So optimisation for electronic presentation is not high on publishers’ agendas it would seem.

But are we missing out? A magazine I have been reading since I picked up my first copy in October 1993 has changed many times in the last two decades. It isn’t a science publication but is related to a hobby of mine, and last year they started publishing a version of the magazine optimised for the iPad. They could have just bunged out a PDF of the paper copy, but they knew that the new technology provided them with a platform to support more content. In place of a photo there is an interactive 360º image, instead of a price list for new products there are hotlinks direct to their entry on the online store, plus there’s additional videos, interviews and zoom panels. If the magazine contains typos or erroneous details, it is automatically updated. The company have started rolling out this idea to their other printed materials.

What if these ideas were used in academia? What sort of content could we include? The most immediate thing that springs to my mind is animations. I produce tonnes of them, and conference presentations aside, they rarely get seen outside of my research group. Why do I make them? Because they are useful for very clearly showing how systems work, if your model is operating how it should or demonstrating patterns in data - (*Thanks to @volcanologist for pointing out that animations can sometimes be submitted, and hosted on a publisher's website).

Take for example some work I have been doing on historic bathymetry data from the Humber estuary. Bathymetry data are readings of water depth at the same tide level, and I use the data to create maps that show the shape and elevation (heights) of the bottom of the estuary. To find out more about what estuaries are, take a look at Sally's previous blog.

Provided by ABPMer, the data spans a period between 1851 and 2003 – I processed the data, calculated rates of elevation change between each sampling period, and from this produced yearly elevation maps. By putting these together as an animation I could see the evolution of the data (it is important here to stress the difference between ‘data’ and reality - not all areas of the estuary were sampled by each survey, and the number and locations of reading varied. Much of the change seen in the video is because of this and not because the Humber has actually, physically, changed in that way).

What immediately struck me was the contrast between the middle and the inner estuary. The middle estuary is the part between the Humber Bridge and the sea, where the estuary’s course deviates southwards – it is remarkably stable over the 150 or so years. The inner estuary, from the Bridge towards Goole, sees lots of internal changes – driven by interactions between the river inputs and the tides – but overall very little change. The Mouth of the Humber, the part closest to the sea, looks to see little overall change, but most of the variations seen in the animation are due to differences in sampling point in the data, and not actual changes. Similarly, changes around the banks of the estuary observed in the animation are most likely caused by sampling difference in the surveys, rather than actual elevation changes.

I have recently been continuing work on adapting a landscape evolution model, Caesar-Lisflood, to model the Humber estuary, and a big step towards this is to accurately model the tides as they are observed by tidal stations recording water depths. Numerically we can do this, but it is important to check that the model is representing the tides in a realistic way - this is a very important step in making a model as it has to be able to accurately simulate observed behaviours before you can experiment with them. Again, animations are a really useful tool for doing this.

The video above shows the variations of water depth throughout several tidal cycles, as modelled, with light blues as shallow and dark purple as deep water. The model changes the depth of the water at the right hand edge in line with water depth data recorded from the Spurn Point tidal station near there. The water then 'flows' from there, down the length of the estuary as the depth increases, and vice versa - this simulates the tides going in and out.

From this I can tell that the model is operating well, as the tide is advancing (coming in/going up/getting deeper) and receding (going out/down/shallower) as expected, throughout the whole region and not just at the points where the tidal stations are located. You'll notice that the early part of the animation shows the estuary filling up with water - this is part of something called 'spin-up', where you let the model run for a period of time to get the conditions right before you start the modelling. In this case it is a 'day' as the water levels gradually builds, filling the estuary.

Another check would be the velocity of the flow as the tide floods and ebbs - this is the speed with which the water is moving (both in or out). The velocity should increase as the tide advances or recedes, but slack water (where the water is hardly moving at all) should be observed at high and low waters. If the model is working as expected, the area of slack water should progress from the sea and up the estuary towards Goole. From the video below, this is seen to be the case. Light blue shows low flow speeds, and darker purples higher flow speeds. The video shows the same modelling procedure as the previous video.

This type of content is really useful to me as a modeller. It is also really useful for presentations as I can show a group of people something that takes a few seconds, yet would probably take a lot of slides and quite a bit of explaining. If academic publications were to begin to include enhanced content in peer-reviewed publications, I believe this could advance the communication of research, not only to other researchers but also to the wider public. For now, Blogs, like the GEES-ology one here, are the best outlet. I hope you enjoyed the animations!