From models to forecasts and what humans add in-between (part 2)

(continued from earlier... this post discusses some of what humans do when creating a river forecast)

Next: “Is there something human-related that is going to affect the floods?” Nature is the second hardest thing to predict in the world. People are the hardest. It is so difficult that my agency in the US would only make forecasts of “natural flow” (i.e. how the river would flow if there were no humans around, with all the effects of dams and farms removed). It would be up to the user to figure out the rest, to translate how “natural” flow will turn into “actual” flow of wet water flowing in the real stream.

Flood forecasters don’t have that easy-out. You wouldn’t want to pile your sandbags up to the “natural” flow level. People just care about if they are going to have “actual” water flowing through their living room. The problem is that there are so many ways that people affect the flow in rivers and often there is not enough data to quantify how the humans are having an impact.

An obvious impact would be a dam that holds or releases water. River forecasters are often in close communication with dam operators. A forecaster might tell a dam operator “expect this much water coming into your reservoir” to which the operator would reply “if that comes true, I’ll be releasing 50 cubic meters per second until Tuesday.” The forecaster would then work that outflow scenario into his forecasts of other locations downstream of the dam. It can be a bit of a circular problem, the releases depend on the forecasts and the forecasts depend on the releases.

This relationship between forecaster and dam operator can be so close that it causes misperceptions among the public. When I worked for a US forecasting agency, I would feel a twinge of resentment a couple times a year on reading newspaper quotes such as “The US Bureau of Reclamation forecast for the inflow to Lake Powell is 115% this summer”. The stories were mostly about how the reservoir level would change as a result of the river forecast, so the reporter would call the Bureau of Reclamation (the reservoir operator and forecast user, not the forecast maker) to look for quotes. It is very easy to forget about attribution and recognition of which agency is doing what part of the process.

The reverse can happen; the Manila forecasters faced angry pleas of (translated and paraphrased) “Why are you flooding us? Please stop!” where the public thought that the reservoir releases were causing high water and that the weather service was the one opening the flood gates. From what I understood, the forecasters only suggested what the future inflow would be. The water managers (a separate agency) then decided that it would be safer to release extra water in advance of the second typhoon.

But even those human impacts can be unclear. Some of the above-mentioned Manila reservoir releases were small compared to what was flowing in to the rest of the natural river. So even if the reservoir operators stopped releasing water, it wouldn’t make much of a difference far downstream. Indeed there can be flat-out misunderstandings; I heard the story of one reservoir being blamed for a flood in a village, but the reservoir wasn’t upstream of the village, it was on a parallel fork and its outflow joined the main river elsewhere downstream. Hydrologically, it would be the equivalent of your doctor saying “I know you think you have a burst appendix, but I assure you it’s not that. The pain is on the wrong side of your body.”

Again, unfortunately, there are many many human factors that fall in this “could be important, might not… 
might happen, might not” category. Does the garbage in the river make the floods worse (by how much? How much garbage is out there now?)? Do the water lilies floating down the channel clog the river? How much of the catchment has urbanized since the last flood? How much of the forest in the headwaters has been cut down? Are any of the levees going to fail? Almost none of the above is ever quantified in hydrologic models, so it’s up to the human forecasters to intuit what the situation is and how things will unfold.

As an aside, if you think that the real answers to these questions don’t actually matter, or you’re happy with the answer “they’re all very important”, consider the consequences to someone trying to come up with an action plan to prevent floods. Do you invest in fixing the water lily problem or preventing deforestation? Similarly, some may say that there is no harm in alternative medicines that don’t work. This is untrue; those resources could go to pursuing a real cure.


(continued again in the final part)