Essential for Hydrologists

Some anonymous asked what is for me the  "Essential for a hydrologist".  Actually I tried to delineate this all along the blog. However, here below I will try a decalogue (with eleven statements actually ... it seems common):

GET KNOWLEDGE

0 - know which kind of hydrologist are you 

Are you studying all hydrological processes ? Or do you need to forecast (operationally) a particular part of the hydrological cycle ? Are you a modeller or an experimenter ? Do you study surface waters or groundwater ? Hydrology is very specialised and each sub-disciplines tends to use its own tools. One of my masters, Ignacio Rodriguez-Iturbe used to say that no hydrologist knows the whole hydrology. The water cycle, in fact, is so pervasive on the earth that it is difficult to know equally likely any hydrological subject.

 1-  In any case, know the fundamental processes (Hydrology in the XXI century is a physical science); remember that all in hydrology is a budget (mass, energy and momentum), and they are conserved.  Use sound theories and avoid unessential empiricism (this is what inspired my Slides about Hydrology)

2 - Be aware of scales. At which scale do you work ? Hydrology covers phenomena at closely molecular scale to continental scale. Methods for studying (models and measures) could be differentiated at different scales. Processes dominant at one scale can be negligible at other scales. Scaling characteristics are present everywhere … so actually some equations could be re-parametrized when used at a certain scale. Other equations simply emerge … (even if a bottom up statistical mechanics of hydrology is missing: lot of work for researchers!).

3 - Be aware of heterogeneities. Most of hydrological characteristics wildly vary in space, and this affects parameterisation of the phenomena at the scale of interest. Parameters could be sometimes thought as random variables (As in the case of hydraulic conductivity).

GET TOOLS (Select your ones, but if you can live with a community is better)

4 - know a GIS (e.g. gvSIG, open source is possible and better^1). 

Since hydrology is space varying and scale varying, you need something to visualise it properly. Spatial representation of phenomena and heterogeneities is, IMHO, one important key for understanding (so blessed is who discovers ways to identify spatial patterns and quantify them).

5 - Get a tool for quick calculations and visualisation (e.g. R^2) 

  Since Hydrology is making budgets you need some tool for making calculations. (Refrain: if it is open-source, it is better).

6 - Get a programming language, possibly Object Oriented  (e.g. Java. Other choices are possible indeed ^3) and put your hands in models^3b.

Learning is a process in which repetition and  calculation are important. Tools for quick calculations are not always suitable for more complex tasks, in which many people interact, exchange data, and explore complex non-linear physical environments like the hydrosphere. OO programming highly helps in maintaining and evolve complex code. (Java has a multitude of tools for doing that. In fact they were able to build a GIS and entire modelling systems in Java)

7 -Learn statistics (for data analysis and models outcomes  inspection)

Someone sees statistics as a tool for a forecasting. I tend to see in it more a tool for understanding (and building null hypotheses).  Data handling and understanding is at a core of any physical science. Maybe you do not perform experiments or field works, nevertheless a physical science has to do with data (huge amount of noisy data indeed), and you have anyway to cope with it. To do statistics you need tools (e.g. R). 

BEHAVE

8- Read the best papers of the best  researchers (learn from  smart people, and copy them). Remind that there are "poets" (original thinkers) and "translators" (followers).  Read the poets, even if the translators can be good. (e.g. see Benchmark papers - Scientists -WRR best papers)

9 - Avoid models and theories that "just work"^4  (but use them for surviving).

Programs and models that just work are useful to engineers who have to give answers and numbers. But at a certain point they fail miserably to give the right answer. Work for the right answer for the right reasons (anyway getting puzzled by the reasons something works despite the known evidence is important).

10 - Do not give up to improve your knowledge, and search for switching paradigms before paradigms reveal obsolete.


P.S. - One thing I forgot to say is that all model are obviously imperfect, and quantifying the forecasting error should be an exercise that any hydrologist should do any time he/she uses a model. Uncertainty in models, is certainly an issue.

___________________________________________________________________________________

^1  - I have to say that for professional printing people, even many open source users, eventually use Arc* to create maps. Printing options are certainly a weak point in open source GIS

^2 - Many use Matlab. I use to be a proficient Mathematica user. Others use IDL. All very good commercial product. But I decided to go open source.

^3 - Now is the moment of Python, which is, however, used conjointly with FORTRAN and/or C for the real task. So I believe that using Java for (almost) everything, and paying a little bit in performances,  is an economical choice.

^3b - If you work for an administration or a company, at the end, you will probably not program by yourself. You will simply use programs made by others (HEC-RAS, SWAT, SHETRAN to name a few). So you will turn your expertise in using them. But, obviously you have also to know the core hypotheses on which these models are based, and you certainly experiment that they will fail in some of your critical task. So somewhere there must be someone that, for you and for others, eliminate these drawbacks and push modelling on. That is the reason I strongly support modelling by components.

^4 Here a link to the Klemes Paper: "Dilettantism in Hydrology, Transition or Destiny", Water Resources Research, Water Resources Research, 1986, that further expands this topic.