Hydrological modelling is dependent upon meteorological input data. The main variables are temperature, precipitation and evaporation. Temperature is usually relatively easy to handle because the spatial variations are smooth and the temporal variations are characterised by distinct cycles (mainly over the day and over the year).
Because of these regular and smooth variations in time and space, temperature can be observed with high accuracy and is also well described in atmospheric models. Precipitation and evaporation are substantially more difficult to handle.
Different types of precipitation
There are two fundamental types of precipitation, generated by different mechanisms and associated with different scales.
Precipitation caused by frontal passages is relatively evenly distributed in both time and space; generally a stable, relatively low intensity covering a large area. This type of precipitation can be accurately observed even with a sparse gauge network and is also well described in atmospheric models.
The other type of precipitation is caused by local convection, which produces locally very high intensities for short periods of time (e.g. thundershowers). This type requires a very dense gauge network to be accurately observed, or different types of instruments that measure over large areas (e.g. weather radar), and also a short time step (preferably on the order of minutes).
Further, this type cannot be fully described in atmospheric models because of their insufficient spatial resolution. A lot of research and development is required to interpret and process both observed and modelled precipitation in order to make it as suitable as possible for hydrological applications.
Precipitation in hydrological forecasts: from very short to very long
Hydrological forecasts are issued with very different time horizons.
- The operational forecasts are made for the coming 10 days.
- Shorter forecasts, e.g. for the coming 48 hours, may be useful, especially in flood situations.
-Longer forecasts, for the coming 3-6 months, are issued mainly for application within the hydropower industry.
- Extremely long forecasts (scenarios), until the end of the century, are made to assess the impact of climate change on e.g. water resources and flood risk.
The different types of forecasts require different treatment of the precipitation input. Short forecasts and high-resolution hydrological modelling require a careful analysis of different kinds of precipitation observations.
The operational 10-day forecasts use metyeorological ensemble forecasts, i.e. a number of different forecasts that describe different future probable evolutions of the weather.
For seasonal forecasts, precipitation data are extracted from selected historical years, that are estimated to be most representative for the coming 3-6-month period. For hydrological climate change impact modelling, generally an adjustment of the climate model precipitation is required to get a credible result.
How much water evaporates, really?
A main objective of hydrological modelling is to describe water balance over longer periods of time. Then the amount of water that evaporates becomes an important component. This amount is however difficult to estimate, mainly because evaporation is difficult to measure and consequently difficult to describe in models.
Based on different observaions, relationships of varying complexity have been propsed, expressing evaporation as a function of e.g. temperature, wind, radiation and humidity. However, it is difficult to prove that more complex relationships are more accurate than simpler ones, and futher some of the explanatory variables may be difficult to estimate.
One aspect of uncertainty concerns whether the same relationship can be used for all temperatures, this is not least relevant in continental-scale modelling over different temperature zones and in impact studies of global warming.
Ongoing research aims at developing robust and ‘climate proof’ relationships for evaporation.