annually repetitive time series & question regarding HBV-EC Emaluation

[Julian]

Greetings,
at the moment I am setting up a Model for a 450.000 km² large watershed. I have ~90 HRUs and it's not clear for me yet, how I best setup my HRU-Data regarding land use classes and vegetation classes.

For instance: One HRU has the following vegetation classes: 70% Conifer forest (20% cover) and 30% grass land.

How to set up this HRU?
##.rvh-File
:HRUs
LAND_USE_CLASS -> FORESTED01
VEG_CLASS -> CONIFER_FOREST
....

##rvp-File
:VegetationClasses
CONIFE_FOREST LAI et cetera

:LandUseClasses
FORESTED01 -> Impermeable Frac: 0; Forest Coverage: 0.14 ( 0.2*70 -> 14 )

Do I have to set up a Landuse class here for each of my HRUs (FORESTED01,..., FORESTED50, GRASLAND01,...) because they differentiate from each other just a bit or can i add this Coverage Fraction somewhere within :HRUs command in the .rvh-File

In the manual (A.2.1, p.124) it's written regarding forest coverage that it is recommendet setting either 0 or 1, and to handle partial coverage via HRU definition. Where can i do this HRU-definition? I couldn't find it in the .rvh-File.

Thank you very much in advance for you help,
Julian

[rchlumsk]

Hi Julian

I think there may be some confusion in the term HRU here. The way that Raven and most hydrologic models are setup are to define the subbasins, and have multiple conceptual HRUs within each subbasin. Each HRU has one land use type, one soil type, etc. However, within each subbasin there can be multiple HRUs, which are defined by the unique combinations of land use and soil type properties. Of course, there will likely have to be some lumping of land uses and soil types in order to reduce the number of unique combinations (e.g. combining clay types into just 'clay', lumping pasture and agriculture, depends on the purpose of the model and the data you have).

If we take your example, say you have the following land use and soil types:

LAND USE: coniferous forest, grass land
SOIL TYPE: sand, clay

Subbasin #1 can have a mix of these, so if overlay these GIS layers we may find the following unique combinations:
HRU 1: coniferous forest, sand
HRU 2: coniferous forest, clay
HRU 3: grass land, sand
HRU 4: grass land, clay

Subbasin #1 may have an area of 100km2, where the HRUs would each have a proportion of that area to sum to the total area of Subbasin 1. This proportion would define the hydrologic response based on the areal coverage of the combinations of land use and soil types.

Getting back to the model setup, in the rvh file you can specify the area of each HRU and which subbasin it belongs to. The rvh class will define the subbasins something like:

Code: Select all

:SubBasins                                       
   :Attributes   NAME   DOWNSTREAM_ID   PROFILE   REACH_LENGTH   GAUGED                     
   :Units   none   none   none   km   none                     
   1   sub1   2   1.1   1.00   0                     
   2   sub2   -1   1.1   5.00   0                     
   3   sub3   2   1.1   1.00   0                                          
:EndSubBasins

:HRUs                                       
   :Attributes   AREA   ELEVATION   LATITUDE   LONGITUDE   BASIN_ID   LAND_USE_CLASS   VEG_CLASS   SOIL_PROFILE ...
   :Units   km2   m   deg   deg   none   none   none   none   none   none   ratio   deg
   1   75.00   960.35   45.78910   -78.38   1   Coniferous_forest Coniferous_forest    SOIL1 ...
   2   25.00   973.45   46.5678   -78.22   1   Grassland Grassland    SOIL1 ...
   3   81.64   963.33   45.12345   -78.15   2   Coniferous_forest Coniferous_forest    SOIL1 ...
...
(more HRUs)
...

:EndHRUs


where the HRUs define that subbasin 1 is composed of two HRUs, and based on the areas of each HRU, Subbasin 1 is 75% coniferous forest and 25% grassland, with a total area of 100km2.

In the rvp file you only need to define the properties for each land use class/veg class and soil types. For example, it might look something like:

Code: Select all

:LandUseClasses   
   :Attributes         IMPERMEABLE_FRAC   FOREST_COVERAGE      
   :Units              frac               frac   
   [DEFAULT]         0.1         0.9
   Coniferous_forest   _DEFAULT         _DEFAULT               
   Grassland         _DEFAULT         0.05
:EndLandUseClasses


where the Coniferous_forest class is defined as 90% forested and the grassland still has 5% forest. Raven will then apply these parameters to the HRUs as defined by their single land use class.

Hopefully that clears it up, feel free to reply back if you have additional questions.

Cheers,
Rob

[Julian]

Thank you Rob for the answer!
This was very helpful.

Right now I have another question:
For instnace I have a forest class or another landuse like agriculture, with a attached Forest_Coverage (Vegetation).

FOREST_1 -> Coverage 0.5
AGRI_1 -> Coverage 0.33

Then I have my HRUs with different Vegetation Classes:
HRUs:
ID - LAND_USE_CLASS - VEG_CLASS
1 - FOREST_1 - FOREST_VEG
2 - AGRI_1 - CROP_1

One half of my Forest HRU is populated with trees and one third of my Agriculture HRU is populated with some Crops. For these populated areas I would have defined some parameters for each Veg.-type within the VegetationClass. However how is RAVEN handling the other 50% respectively 66% in this case?
Since I have to generalize quite a bit, there could be for instance either bare soil or shrubs.

I appreciate any help.
Best,
Julian

[rchlumsk]

Hi Julian

I think I understand your question but let me know if it goes unanswered. The way Raven handles the various parameters depends on the process choices made, but let's take the example of HRU 1 with FOREST_1 land use and 50% forest coverage. HRU 1 is still one HRU area with a parameter of FOREST_COVERAGE, it is not divided into 2 HRUs. For example, when Raven calculates the canopy evapotranspiration rate for HRU 1, it will use the FOREST_COVERAGE parameter with a value of 0.50 to estimate the canopy evapotranspiration as a fraction of the PET (see page 46 of the manual, v2.8).

If you want to split the FOREST_1 land use into two areas, say one with 100% forest cover and another with 0% forest cover, you could do that and adjust the areas and FOREST_COVERAGE parameter (and other parameters) accordingly. However, for the sake of being able to reduce the number of HRUs in a model, you can lump, for example, partially forested areas into one land use and assign parameters accordingly.

The important point here is that Raven does not split HRU units any further, a single HRU is by definition one unit, but can be described by different parameter sets (set by the land use, vegetation and soil classes) to make the response different from other HRUs. Let me know if that makes sense of if you would like further clarification!

Cheers,
Rob

[Julian]

Thanks a lot Rob,
that cleared up a few things.

Well right away to another question:
How would you proceed without information about channel geometry? My model consists of 13 subbasins and so far I am only using the "No Routing"-option for the In-channel routing.

Thanks in advance!

Best,
Julian

[rchlumsk]

Hi Julian

Depending on the model application, No Routing option may be fine - particularly if you want a more conservative estimate of peak flows. However, I imagine the routing will play a large role in matching observed data in calibration and is an important part of most hydrologic models.

Without channel geometry, you likely have some combination of three options:

1) Determine other routing options for in-channel routing that do not require a full cross-section, such as Simple plug flow, although you will still require the estimation of some channel properties such as reference celerity and channel length
2) Search for data that you can obtain to estimate cross-sections, such as open source LiDAR data from the province (assuming surveying is not an option). You may also be able to contact whichever agency manages the hydraulic models in your area to obtain hydraulic models and use their cross-sections
3) If you have some idea of bankfull flows, you may generate a sample cross-section dataset for use in the model as a generic, representative cross-section using normal depth with Manning's equation.

In short there is not really an easy answer, it depends on the area you are modelling, the purpose of the model, and the type of information you are able to obtain. Likely either sticking to No Routing or generating a sample cross-section based on a bankfull flow and/or LiDAR data is your best bet if you are not able to obtain more accurate representative cross-sections. Hope that helps!

[Julian]

Jep Rob,
that helped, thanks a lot! :)

If I would want to track the source of my discharge at the basin outlet (surface runoff, baseflow, etc.), how would I achieve this goal?

Thanks in advance!

Best,
Julian

[rchlumsk]

Hi Julian

The solution to the tracking of sources is to basically setup a 'contaminant' concentration in the source you wish to track with a fixed concentration of 1.0. When the contaminant tracking is included in the Raven model, the results will produce the concentrations in the state variables.

An example of this is in the Raven Tutorial #3 (tutorials available for download on the Raven page, http://raven.uwaterloo.ca/Downloads.html) which you may find useful. The syntax in the .rvi file for tracking snowmelt is this. Note that in this example, SNOWMELT is a name that was assigned as a generic contaminant name, this can be replaced with whatever name you feel fits the item you are working on.

Code: Select all

:Transport SNOWMELT
:FixedConcentration SNOWMELT SNOW 1.0
:FixedConcentration SNOWMELT SHALLOW_GW 0.0
:FixedConcentration SNOWMELT DEEP_GW 0.0

[Julian]

That worked fine,
thank you! :)

[Julian]

I am trying to implement Permafrost into my model. At the moment I am just trying to geht the :TransientParameter -Command to run:

Code: Select all

:TransientParameter [PARAM_NAME] [class] {(optional) ClassName}
  [date yyyy-mm-dd] [time hh:mm:ss.0] [interval] [N]
  {double value} x N
:EndTransientParameter

For instance for the code below I get the following error:
'ERROR : CommonFunctions::FixTimestep: timesteps and time intervals must evenly divide into one day

Code: Select all

:TransientParameter IMPERM LANDUSE
  [1980-11-01] [12:00:00] 30 4
  0.5
  1.0
  1.0
  0.5
:EndTransientParameter

How do [interval] and [N] (number of entries) work? How can I apply this every year?

My target would be a dynamic active layer in which my soil and landuse parameter are changing over the length of a winter.

Thanks a lot in advance!