| | 5 | = 10/10/2013 - Technical meeting on the routing scheme = |
| | 6 | Present : Agnès Ducharne (AD ; secretary), Josefine Ghattas (JG), Betrand Guénet (BG), Matthieu Guimberteau (MG), Ronny Lauerwald (RL), Jan Polcher (JP), Camille Risi (CR), Ana Schneider (AS) |
| | 7 | |
| | 8 | The main equation of the routing scheme is found in subroutine routing_flow .f90:[[BR]] |
| | 9 | '''flow=MIN(i_reservoir/((topo_resid/1000.)*i_tcst*one_day/dt_routing), i_reservoir(i)-min_sechiba)'''[[BR]] |
| | 10 | i_flow(ig,ib) = MAX(flow, zero) |
| | 11 | |
| | 12 | - flow is in kd/dtrouting,[[BR]] |
| | 13 | - i_reservoir is in kg, [[BR]] |
| | 14 | - one_day and dt_routing are in seconds, and are equal by default,[[BR]] |
| | 15 | - i is a surrogate for fast, slow, or stream, and corresponding i_tcst values defined in the routing module, supposedly in day/m : slow_tcst_cwrr = 25.0 ; fast_tcst_cwrr = 3.0 ; stream_tcst_cwrr = 0.24 |
| | 16 | T_i = (topo_resid*i_tcst/1000.) gives the characteristic timescale of reservoir i, in days. In this parameter T_i, topo_resid describes the effect of the grid-cell morphology (via the streams' length and slope), while i_cst gives the characteristic velocity of the reservoir, with a dimension of [T/L] |
| | 17 | |
| | 18 | We had a first meeting on October 24th, with AD, MG, RL, and AS, in which we started to analyse the source code of routing.f90. [[BR]] |
| | 19 | Questions remained about the actual units of topo_resid (read from the file routing.nc) and i_cst, given the presence of a 1000 factor. |
| | 20 | |
| | 21 | After a careful look at the values in routing.nc, the scripts Jan developped to create routing.nc, classical velocities of large rivers found in the literature, and considering the fact that the routing scheme simulates satisfactory river discharges in off-line mode, we conclude that :[[BR]] |
| | 22 | - '''topo_resid is in km routing.nc''' [[BR]] |
| | 23 | - '''the characteristic velocity of the reservoirs is (i_tcst/1000.) in days/km''' [[BR]] |
| | 24 | The code comments should be corrected accordingly. We suggest to remove the division by 1000 from the code, and to integrate it directly when defining i_tcst in the module :slow_tcst_cwrr = 25.0/1000. ; fast_tcst_cwrr = 3.0/1000. ; stream_tcst_cwrr = 0.24/1000. !! in days/km |
| | 25 | |
| | 26 | '''We then discussed the possibility to use higher resolution topographical datasets to calculate topo_resid''', which is presently defined at the 0.5° resolution, based on the dataset produced at the same resolution by Vorosmarty et al. (2000) : ''Global system of rivers: Its role in organizing continental land mass and defining land-to-ocean linkages, Global Biogeochem. Cycles, 14, 599–621.'' |
| | 27 | |
| | 28 | The main two options are to start from Hydro1k (http://webgis.wr.usgs.gov/globalgis/metadata_qr/metadata/hydro1k.htm) or HydroSHEDS (http://hydrosheds.cr.usgs.gov/index.php), which are both hydrologically conditionned DEMs, at the 30 arc-sec =1 km and 15 arc-sec resolution respectively. Hydro1k misses data in Australia and does not use a global projection. HydroSHEDs does, but lacks information in Antartica, which is not very limiting since LMDZ does impose an icecap over this continent for the moment. |
| | 29 | |
| | 30 | '''We propose to undertake this work based on HydroSHEDS''', as part of the PhD thesis of AS, supervised by AS. JP will provide support to ascertain the precise information required by ORCHIDEE. Marie Sylvestre, specialist of Geographic Information Systems (GIS) in Sisyphe, will provide technical support. We will submit this work to the ORCHIDEE group is critical decisions need to be taken, and in any case to show the comparison with the original version. |
| | 31 | |
| | 32 | ---- |
