July 1- 3, 2004  ENGREF- Paris
MOPEX Paris Workshop

Recently published papers

1         General Scientific Objectives

MOPEX (Model Parameter Estimation Experiment) was established to aid in the development of techniques for the a priori estimation of parameters used in land surface parameterization schemes of atmospheric and hydrological models. Workshops have been held each of the past three years to investigate various issues related to parameter estimation and regional applications. The 4th International MOPEX workshop will be held this summer in Paris, focusing on the estimation of parameters for ungaged watersheds in humid regions. To allow for maximum participation, the workshop is designed so that modelers can choose varying levels of involvement (based on resources and the number of basins they wish to study). A well defined set of science objectives are also planned to allow for a coordinated and meaningful analysis of workshop results. A last objective for the workshop is the development of an international database with basins from various countries. 
 

2         Workshop Structure  

Participation Level Choices:

  •  Level 1 - using 3 basins

  • Level 2 - using 12 basins ( including those from Level 1)

  • Level 3 - using 40 basins (including those from Level 2)  

Optional Basins for Study include 12 U.S. Basins (Humid Climatology).

These basins may be found at the NOAA/NWS ftp site: ftp.nws.noaa.gov/oh/gcip/mopex2004

List of actions for participants of the MOPEX Paris workshop  

1. Ungaged Watersheds

 

A Priori Parameter Estimation  
Estimate the parameters of the chosen hydrologic model without using hydrometric measurements for optimization  
Streamflow Prediction  
Predict the hourly streamflow in the ungaged watershed for the whole record period ((01/08/1995-31/07/2002). The first year will be considered as a warm-up period and will not be used in the error criteria calculations. 

Calculate the objective functions provided below  

2. Gauged Watersheds

 

Parameter Estimation through Calibration
Calibrate the hydrologic model on the first half of the time period (01/08/1995-31/07/1999) and predict the hourly streamflow for the second half (01/08/1999-31/07/2002). The first year will be considered as a warm-up period and will not be used in the error criteria calculations. 

A list of objective functions that should be used for model calibration and validation is provided below. 

The objective is to set the upper limit on what can be achieved on an ungauged basin  

3. Sensitivity

 

Parameter Sensitivity  
Provide a simple estimate of the sensitivity of model parameters. [A common approach for all participants is suggested below. Participants can individually pursue a more intensive analysis of this issue].

4. Analysis  

Analyze results based on a list of criteria   

A) List of criteria

Participants are free to choose their own objective function during the parameter calibration phase. However, model performance will be assessed based on the criteria provided below, both for the gauged-calibrated and the ungauged mode. 

Therefore, we strongly recommend the use of one (or more) of these criteria.

Where:
N is the number of time steps used, 
t  is the time step index, 
Q is the streamflow, sim and obs indicated simulated and observed flows and
is the mean annual discharge value.

 B) Sensitivity Analysis

 
We would like every participant to submit a basic sensitivity analysis in order to gain 
a better insight into the working of their models on a larger scale. 
A basic scheme for this analysis is: 
  • Vary values of each parameter around its best estimate
by +/-10% first, and then by +/-25%.
  • Calculate the statistics outlined above for each of the runs. This will produce
 Nx4 model runs, where N is the number of (calibrated) model parameters.
  • Submit the parameter sets with the corresponding objective functions in a text file
entitled 'XXXsensitivity.dat' (replace XXX by model name) 
 
 Those participants who are not able to undertake a full sensitivity analysis should provide
 a written description of their model, including all equations and conclusions of former
 parameter sensitivity analyses.

3      Presentations during workshop – Pre-defined elements that are required to be addressed by each participant  
a. Present the details of the model structure you applied. 
b. What approach did you use to estimate the a priori value of your parameters? 
c. What approach did you use to calibrate your parameters? 
d. Present the results of your sensitivity analysis, and of the streamflow predictions for the a priori and the calibrated case. 
 Files to be submitted for each basin:  
a. Hourly streamflow simulation runs using a priori parameters for the record (ungauged mode)
b. Hourly streamflow simulation run using calibrated parameters for the second half of the record (gauged mode)
c. Best parameter set for basins (gauged mode)

      * Participants who have chosen to undertake the sensitivity analysis detailed in section 2B should also submit the sensitivity runs and their corresponding parameter sets and statistics

Possible complementary research paths for the MOPEX Paris workshop  

  • Provide an estimate of the soil moisture state (time-series) of the watershed. 

  • Develop new a priori estimates for model parameters based on the characteristics of the whole watershed sample. 

  • Compare and discuss criteria used for the assessment of model results. 

  • Assess how results depend on the time step used by the model (compare for example the results obtained at the daily time step, depending on the time step that has been used for the precipitation input: hourly vs. daily). 

  • Multi-model analysis 

  • Calibrate the hydrologic model using only ten days chosen in the first half of the time period and predict the daily streamflow for the second half. (Participants may use a combination of an a priori parameter estimate and a calibration on these ten days). 

4         Instructions for interested participants  

The MOPEX group welcomes all new participants who would like to contribute to the Paris workshop with their own model, using the common data sets. Data and instructions for the naming of file may be found on the Cemagref ’s ftp file. To access the data participants should first download sign and return the data exchange agreement. Once the document is received, instructions to access the site, logins and passwords will be emailed. Result files should be submitted before the 22nd of June.

convention_MOPEX.doc  pdf

5   Organization

The Paris workshop is organized by the Cemagref in collaboration with the ENGREF, Météo France, National Weather Service and the SAHRA/University of Arizona. 

Members of the local committee are :  
France : Vazken Andréassian et Nanée Chahinian (Cemagref), Florence Habets (Météo France), Éric Parent (ENGREF). 
USA: Qingyun Duan (Lawrence Livermore National Laboratory), Hoshin Gupta (SAHRA/University of Arizona), Terri Hogue (UCLA), George Leavesley (USGS) John Schaake (National Weather Service), Thorsten Wagener (SAHRA/University of Arizona). 

6    Registration  

All participants are requested to fill in the registration form and return it by June 22nd:  2004. The registration form can be downloaded using the following link.

Registration Form pdf

Invitation Letter for Visa pdf

7 Travel information (maps)    

The workshop will be held at the ENGREF, 19 avenue du Maine 75015 Paris. 
Nearest metro station is Montparnasse-Bienvenüe. Air France buses can transfer you directly from CDG and Orly airports to the gare de Montparnasse train station located 500 m from the workshop venue » 10 minutes walk away. 
From “Place Raoul Dautry” head to “Boulevard Vaugirard”, walk on straight for 81 m, turn left on “Place Bienvenüe” and head towards the Avenue du Maine. 

The following links might be help you get around in Paris : http://www.mappy.fr provides maps and iteneraries ; http://www.ratp.fr/ the Parisian public transport service site

8      Hotel booking

Each participant should take care of his/her own hotel booking. A list of hotels located within close distance of the venue has been supplied by the ENGREF. Participants can also search for accommodation through : 
ELYSEES WEST RESERVATION Tél : + 33 1 47 75 92 90 Fax : + 33 1 47 75 92 99 
The Paris tourist board : http://www.paris-touristoffice.com/index_va_v2.html 
We advise you to book your accommodation well in advance. 

Hotels selected by the ENGREF  

Name   Address   Rate (*)  
HOTEL DE LA TOUR   19 boulevard Edgar Quinet 75014 PARIS
  Tel : + 33 1 43 20 67 09- Fax : + 33 1 40 47 04 31		 35 (5)  
ETAP HOTEL PARIS PORTE D'ORLEANS

15-21 boulevard Romain Rolland 75014 PARIS  
Tel : + 33 1 46 56 26 27- Fax :+ 33 1 46 56 57 40 

42,5 (3,80)  

PETIT PALACE HOTEL  

131 avenue du Maine 75 014 PARIS
Tel : + 33 1 43 22 19 75  -Fax : + 33 1 43 21 79 01  

54 (5)  

HOTEL MISTRAL  

24 rue de Cels 75014 PARIS  
Tel : 01 43 20 25 43 Fax : 01 43 21 32 59

56 (6)  

CAMPANILE MONTPARNASSE  

146 avenue du Maine 75 014 PARIS  
Tel : + 33 1 53 74 60 00 - Fax : + 33 1 40 47 01 49  

67 (6,42)  

TRIANON/ IBIS  

22 avenue du Maine 75 015 PARIS
  Tel :  + 33 1 45 48 95 52  - Fax : + 33 1 45 48 82 44  

79 (6)  

Rates are for a single room with shower and WC. Breakfast is not included but can be purchased for the sum given in parenthesis.

(*) The Rates are in Euros as stated on April 24, 2003.  

          

                   

 retour page colloque