PILPS 2e Project Memos

December 20, 1999: Research Announcement for PILPS 2e

Dear PILPSer,

High latitude land processes, including snow accumulation and ablation, soil freezing and thawing, and extensive lake and wetland areas pose particular problems for the representation of the land surface in numerical weather prediction and climate models. These regions are the particular focus of PILPS Phase 2(e), which will carry on in the tradition of Phases 2(c) and 2(d) by conducting a multi-year, spatial intercomparison of hydrology and land surface processes in northern regions. The project will be a joint activity of GEWEX, through PILPS, and the Arctic Climate System Study (ACSYS).

The intercomparison will consist of three stages, focusing on the Torne River in Sweden, the Mackenzie River in Canada and the Lena River in Russia. Stage 1, utilizing data from the Torne River basin, will begin in winter 2000 with the distribution of a draft design plan for comment by all participants. Dennis Lettenmaier from the University of Washington, will be the lead-PI for stage 1.

Further information about the PILPS-2(e) experiment, including the science plan formulated by a joint ACSYS/GEWEX working group, can be found at http://www.hydro.washington.edu/Lettenmaier/CurrentResearch/PILPS-2e/index.htm. At this point we are looking for an expression of interest, so that we can estimate the likely number of groups that would be participating. Accordingly, we'd appreciate it if you could respond to Laura Bowling (lbowling@u.washington.edu) by January 31, 1999. Note that we expect that the "winter 2000", as stated on the web site in connection with release of data, should be interpreted as about February or March, 2000. As in previous PILPS projects, we anticipate releasing a short draft experiment plan to participants, this would occur in the near future (probably early January), once we have a list of likely participants. Finally, we have attached to this message a copy of the list of addressees of this message, we'd appreciate it if you could let us know of any potentially interested parties that do not appear on this list.


Dennis Lettenmaier and Laura Bowling

January 24, 2000: Registration Information Required

Thank-you for expressing interest in Pilps 2E. We are continuing to move forward while finalizing the participant list. I am sending this message to each person who responded for their group, and members of their project team when explicitly specified. If you have not already done so, please respond with the following information:

  1. The name of your primary contact if other than the initial responder;
  2. The name and email address of any members of your team who should receive all pilps 2E mailings;
  3. The name and acronym of the model to be used; and
  4. The organization which you are representing.


Laura Bowling

April 15, 2000: PILPS 2e Status

Pilps 2E participants,

I am writing to provide a status report for PILPS 2E. We have decided on a few changes in infrastructure, in the hopes of streamlining data handling and quality control for PILPS-2E (and future PILPS activities). We have agreed to design PILPS 2E in conjunction with ALMA (Assistance for Land-surface Modeling activities), which is part of a new GEWEX activity, the Global Land-Atmosphere System Study (GLASS), which is being coordinated by Jan Polcher (LMD/CNRS - Paris). The idea is to use a simplified infrastructure for land-surface scheme inter-comparisons, which should reduce the "overhead", and facilitate post-analysis. ALMA will use a standard data exchange format, most likely netCDF with the GDT convention. This will mean that model inputs and outputs will be distributed and archived in netCDF format. The PILPS-4C interface used this format, so there is some precedent. We will be receiving support from Jan's group in this effort, and in fact Laura Bowling, who will be coordinating the experiment from the UW end, will be meeting with Jan and his group next week to sort out some of the details. Among the items of discussion next week will be development of additional drivers and translation software to assist the needs of the various participants in providing us quality-controlled model output formats. One idea that will be considered is some kind of driver to "translate" from the formats used in previous PILPS experiments. To that end, we'd appreciate knowing how many of the PILPS-2e participants would be interested in such a translator. We will also be coordinating with Andy Pitman, who has extensive experience with previous PILPS experiments (and various quality control issues) as we move forward. Our intent is to make this transition as smooth as possible, but the additional infrastructure will delay the start of Pilps-2E somewhat. Our current intention is to distribute a preliminary design plan for comment by late May, with the distribution of forcing data by late June, 2000. Since the summer field season is often a difficult time to pin people down in front of their workstations, we will extend the time available for model simulations.


Dennis Lettenmaier

May 23, 2000: PILPS 2e Draft Experiment Plan

Dear PILPser,

Attached is a draft version of the experiment design, in pdf format, for the PILPS 2e Stage 1 application in Sweden. A postscript version is also available from the PILPS 2e web site: http://www.hydro.washington.edu/Lettenmaier/CurrentResearch/PILPS-2e/index.htm

It is intended that this document will serve as the main instruction set for this experiment. We request that all participants carefully review this plan and ensure that all of the necessary information is provided for each participant to perform the outlined simulations. Please provide any comments or suggestions to Laura Bowling (lxb@hydro.washington.edu) by June 22, 2000. Any significant points of discussion which arise will be posted on the PILPS 2e web site.


Laura Bowling

July 5, 2000: Experimental Design and Response to Comments

Thank-you to everyone who took the time to review this document and submit your questions. A summary of the questions and comments we received is provided below. The revised experimental design will be posted on the PILPS 2e web site (http://www.hydro.washington.edu/~Lettenmaier/ CurrentResearch/PILPS-2E/index.html), as soon as a few outstanding issues are resolved. A copy of this document is also available on the PILPS 2e web site.

Response to Comments:



July 21, 2000: PILPS 2e Data

The PILPS 2e forcing data and instructions are available via ftp!

ftp to: norway.hydro.washington.edu/pub/lxb/pilps2e

For convenience, the meteorological forcing files are seperated by year, look to the ALMA web site for instructions on combining the yearly files in time.

My apologies to anyone waiting for the monthly greeness fraction files, they will be ready sometime next week.

Datasets on cd-rom should also be received by those that requested them by the end of next week.



July 28, 2000: PILPS 2e Updates

The revisions and clarifications of the experimental design have begun. A FAQs page has been added to the PILPS 2e web site (www.hydro.washington.edu/~Lettenmaier/CurrentResearch/PILPS-2E/index.html), and I will continue to update this as I receive questions. The instructions document will only be updated for major changes, so it is best to check the FAQ list periodically to see if any minor changes have been posted. That being said, the instructions document on the ftp site has been updated as of 7/27/00. The changes include:

  1. Specification of the time axis for output files;
  2. Revision of the Rawls et al. (1982) porosity values;
  3. Instructions for vertical interpolation of wind speed;
  4. Correction regarding averaging of forcing variables (they are actually backward-averaged, not forward-averaged); and
  5. The addition of 6 more output variables.

Please see the PILPS 2e web site, or download a new version of the instructions for details.

In addition, the mean monthly green vegetation fraction file (tor_gvi_mon.nc) is available from the ftp site (ce.washington.edu; pub/lxb/pilps2e).

Here's hoping we won't have this many changes every week!

Laura Bowling

August 19, 2000: PILPS 2e Met forcing revisions

I have found several small errors with the meteorological forcing files which requires that they be replaced. The changes should not result in any major changes to the results you have received to date. I apologize for the large inconvenience.

New forcing files for 1979-88 and 1989-98 and new cloud cover files for 1989-98 will be posted on the ftp site as of August 19, 2000. The errors include:

  1. Incoming shortwave radiation was calculated with respect to local time, rather than UTC. This resulted in a shift of solar noon by 1 to 2 hours. All met variables are now given with respect to Greenwich time.
  2. There was an error in the disaggregation program for the data sets converted from 1 degree. This resulted in a spatial error in the Psurf, Qair, SWdown and CloudFn data sets. The relative magnitudes of these quantities do not change greatly.

CDs of the new datasets will be shipped to those who have requested them next week.

In addition, there are many updates to the FAQs page on the Pilps 2e web site!

***Please note***

I will be checking email only sporadically from now until August 25th. After August 25th, I will be unavailable until September 10th. I will do my best to respond to all comments and concerns promptly on my return.

September 12, 2000: FTP site access

As some of you have noticed the pilps 2e ftp server has met an untimely demise. The forcing data for pilps 2e can now be found at ftp.ce.washington.edu /pub/HYDRO/lxb/pilps2e/

I am back in the office only briefly until next week, so please be patient if I do not get all of your questions answered promptly.

September 22, 2000: Updates

Sorry about the blank message, I was just updating the address list. If anyone was curious about the complete pilps 2e mailing list, now you have it!

Since I am writing anyway, I will provide my updates now:

  1. The answers to many additional questions have been posted on the reformatted pilps 2e FAQs web site, as of yesterday.
  2. In particular, the fractional areas for the calibration basins now include the actual grid cell area (necessary in order to calculate discharge volume). PLEASE NOTE, THE FRACTIONS HAVE CHANGED SLIGHTLY SO THAT THE DIGITAL AREA MORE CLOSELY MATCHES THE REPORTED GAUGED AREA. PLEASE UPDATE YOUR FILES ACCORDINGLY.
  3. The promised quality control screening program is currently undergoing it's own qc testing. It will hopefully be available on the ALMA web page by the end of next week.
  4. The acceptable range of output variables is being updated in response to user comments. Updates will be incorporated in the QC screening program, and posted on the pilps 2e web site soon.
  5. I think that I have now caught up with all questions I received while I was gone. If you feel I have overlooked your comment, please let me know.

I hope the experiment is going well for everyone. I look forward to seeing some results in the next few weeks!


October 11, 2000: Returning Results

We are getting close to the end! Unfortunately, there are many small updates which may effect production of your final results. I have updated the FAQ page as of 10/11/00.

Some things to notice:

  1. The definition of Ground heat flux, Qg, should no longer include the change in soil heat. This is included in DelSoilHeat.
  2. There was an error in one of the fractional areas of the Kaalasjarvi validation catchment. The updated value is on the web site.
  3. I have revised the dimensions for the subsurface variables. The values of the z-axis should correspond to the depth from the surface to the bottom of each soil layer.
  4. Clarification: SnowT, IceT and SAlbedo should take on no data values when the snow and/or ice are not present. SnowFrac, IceFrac and SDepth can be zero.
  5. I have posted requirements for the format of the output netcdf files. In particular, to pass the QC checks, all variables must include the attributes: units, associate and missing value, as well as the global attribute SurfSgn_convention.

Quality control:

The beta version of the promised quality control program is attached to this mail, written in Fortran 90. The variable ranges in this release have been revised, as indicated on the FAQs page.

Returning results:

I encourage you to return your results to me by hard media (dlt or 8mm tape, cd or dvd). Considering the large volume of the data, FTP transfer of the files is very likely to affect access to our network in a negative way, and to fail for one reason or another before completion. Tapes and CDs can be sent to me at the address given below.

If you must use the FTP method:

  1. please try to avoid peak times on our local network (1700 - 0200 GMT)
  2. compress your data (tar and/or zip archived)
  3. create a subdirectory with your model acronym in "incoming" directory of our FTP server, ftp.ce.washington.edu. (Note, you will not be able to see this directory once you create it.)
  4. place the archived file in the new directory
  5. inform me so that I can retrieve it promptly.


I have not forgotten about documentation, so please don't forget what you have done during calibration! I will send out a more detailed request for documentation of your experience at a later date.

Time Schedule:

Please let me know now which method of data transfer you will use.

Due to the late release of the qc program, we will extend the deadline for return of data by two weeks. We hope to see all of your results by October 29th!

October 13, 2000: ALMA QC

Dear PILPS_2e participants,

The first day of utilization of almaqc has already revealed a number of bugs and problems, how could it have been otherwise !

I have now created a Web page for almaqc which can be accessed from the ALMA bazaar web page : http://www.lmd.jussieu.fr/~polcher/ALMA/alma_bazar.html There, you will be able to get the latest version and see which bugs were corrected.

Beside the usual bugs another more fundamental problem showed up : some of us do not have a F90 compiler. Two options are available here, either provide binaries or generate a C version of the code. The first solution is not too difficult to implement but the second one would take too much time for me to do and could generate some code maintenance headaches. Help and suggestions would be welcome on this point.

Best regards


October 19, 2000: Output Variable Clarification

There have been enough questions about definitions of the ALMA variables as they pertain to each model (especially the energy balance components) that I think detailed definitions are in order. If these definitions seem to reflect departures from previous explanations I have given, please follow these definitions.

To start, the balance equations which I will be using follow. If in doubt, please define your variables so that these equations will work:

energy balance:

DelSoilHeat + DelColdCont = SWnet + LWnet + Qle + Qh + Qa + Qg

water balance:

DelSWE + DelIntercept + DelSoilMoist + DelSurfStor = Rainf + Snowf + Evap + Qs + Qsb

total water in column = SWE + SurfStor + CanopInt + summation(SoilMoist)

Specific definitions follow:

0.1 Energy balance components

0.2 Water balance components

Evap = ECanop + TVeg + ESoil + EWater + SubSnow

0.3 Surface State Variables

0.4 Subsurface state variables

0.5 Evaporation components

0.7 Cold season processes

January 8, 2001: Inquiry of participants about workshop

Dear PILPSer -

As we have hinted in earlier correspondence, we are considering having a participant's workshop in Seattle for discussion of PILPs 2e results, possible additional exploratory runs, and publication of results. This workshop would be patterned after the one held at Princeton at the conclusion of PILPS-2c. The dates we are considering are March 19-20, at UW in Seattle. These dates coincide with our spring break, which will make it much easier for us to access meeting rooms in the immediate vicinity of our own building, computers, and so on. Topics for the meeting would include:

  1. Review of summary results, and comments by individual participants about performance of their models, any peculiarities, problems with data, and so on;
  2. Discussion of possible need for, and nature of, future re-runs. This topic inevitably comes up in pilps and similar experiments, typically either because of problems with the data sets and/or experimental design, and/or the need for additional diagnostics to help uncover the reasons for differences in model performance;
  3. Publication of results;
  4. Other topics (please suggest)

Based on the PILPS-2c workshop, we think two days will be enough. North American participants would be expected to provide their own travel support. We will attempt to secure some funding to defray the costs of attendance of overseas participants. To assist us in planning, please return to us as soon as possible:

I would be able to attend on the 3/19-20 dates

_____ yes

_______ no (If no, is there a reason, like a schedule conflict?)

If overseas,

I would _______

would not ______ need travel support.

(Travel support will be made available for no more than one participant per team).

We hope to see you in Seattle in March! (And yes, it will be raining.)


Dennis Lettenmaier and Laura Bowling

January 11, 2001: Documentation

Dear folks,

Attached is a plain text file containing a list of questions that we think will help in our analysis of the PILPS 2e results. Formatted versions (with check boxes and everything!) are available on the web site: www.hydro.washington.edu/Lettenmaier/CurrentResearch/PILPS-2e/index.htm

We would appreciate it if you could look over the questionnaire briefly and advise us of any problems or additions in the next week. I will notify you when the final list of questions is ready for you to fill out!

To those of you who have submitted results but have not heard from me personally recently, that means so far so good! I am still in the process of checking file integrity and water and energy balances. A summary of these preliminary tests will be posted on the web site once this phase has been completed.


January 30, 2001: PILPS re-runs and other issues

PILPS participants -

Please read this entire email carefully, as it addresses several important issues, including:

  1. Model re-runs; and
  2. Definition problems with models returned to date;
  3. PILPS Workshop

Model Re-Runs:

Two issues have been identified with the forcing data which may have a significant input on some model results. The problems are detailed in the attached pdf document: pilps.bias.report.pdf. Please review this document and comment on the proposed fixes to the forcing data set as soon as possible. I will post the revised forcing data if no significant issues have been raised in one week.

Definition Problems:

Many of you have been contacted by now regarding problems with output variable definitions or netcdf file format. Altogether these discrepancies make the uniform processing of results impossible at this point and several of the groups will have to re-run their models in order to fix the discrepancies. Therefore, I strongly recommend that all participants re-run their models with the new forcing data immediately and use this opportunity to fix the output file definitions.

Since the energy balance definitions have proven the most problematic, I am including a screening program, EnergyBalanceCheck.c. This program will read through the ten annual energy balance netcdf files to calculate the average-annual basin-wide energy balance residual. It outputs a netcdf file containing the average annual values of each of the energy balance components, including error. The output netcdf file can be easily plotted in free software, such as Ferret, to help identify the errors with the energy balance definitions.

You will find that some things are hard-wired in this code. That is in part intentional - if the code will not work on your files, chances are they are defined incorrectly! Nevertheless, I cannot guarantee that there are no errors, and the code will likely evolve over time.

PILPS workshop:

The workshop has been set for March 19th and 20th in Seattle. So now we have a crunch in order to have some results to discuss! Workshop logistics will be posted here as they become available.

Lodging information is currently listed on the web site. Please make your reservations promptly, they will not be held past February 17th, 2001!



February 8, 2001: PILPS 2e Re-runs

Thank-you to all of you who read and commented on the analysis of forcing data irregularities. We realize that for some of you it is simply impossible to perform full or even partial re-runs prior to the workshop and we cannot guarantee that discussion at the workshop will not bring to light further problems with the experimental design. However, for those of you who have indicated that you could perform the re-run, we think it would be very useful to have at least some new results for the entire basin to present at the workshop. If you cannot perform the full re-run, maybe you would consider performing re-runs for just the calibration basins at this point. In short, we will make our best effort to analyze whatever we receive prior to the workshop.

The new forcing data can be downloaded from ftp.ce.washington.edu /pub/HYDRO/lxb/pilps2e/rerun_files/ and then either sub-directory 'tarred/' or 'individual/'. Please let me know immediately if you have trouble accessing our site via ftp. For security reasons I now have to specifically provide access for each of your domains, so it is possible that your domain is not on the 'allowed' list.

The revised documentation forms are also available from the PILPS 2e website (text version attached). Please return this form at your earliest convenience. The survey results will be summarized on the web site as they are received.

A couple of new issues with the data have been brought to my attention, so I will summarize the changes:

Solar Radiation:

Incoming solar radiation has been recalculated using a relationship for clear sky radiation that has been calibrated based on station observations. The new radiation is roughly 10 to 20% higher on an annual basis.


Precipitation has been statistically disaggregated to hourly, taking into account hourly cloud cover. Annual totals are the same, except in the region of the Palkem station (Ovre Lansjarv basin), where the gauge catch factor was reduced.


It was brought to my attention that wind values seemed unreasonably low, with a suspicious north-south gradient. I discovered a processing error in the gridding program. The revised wind values are higher on average in the lower half of the basin, and there is a more reasonable distribution between the upper and lower portions of the basin.

Runoff Ratios:

A couple of people have mentioned suspicious runoff ratios in the Abisko catchment, with precipitation barely exceeding runoff in some years (yielding a basin-wide evaporation of 6 mm for 1996). However, if this calculation is performed in terms of water years (October 1 - September 30), the problem is not nearly as pronounced (minimum evaporation of 80 mm in the water year beginning in 10/1996). Based on water years, the mean annual evaporation for 1989-1998 is 185 mm, which we believe is reasonable for this area. The relatively low 1996 value of 80 mm is due to high runoff in a year with "average" precipitation. Proportionately more of the precipitation falls in January of this year than in other years. It is still open to discussion whether the "high" runoff is a natural consequence of a larger than average snow pack due to this January precipitation, or if the "average" precipitation is the result of insufficient gauge catch correction values for the month of January. In any case, we do not see any insurmountable problems and the precipitation fields HAVE NOT been changed as a result. Perhaps you may not want to be too concerned with model performance in water year 1996 when performing your analysis.

Other concerns:

The only other concern I am aware of is a general wariness of longwave radiation data which we have no way of verifying. I don't believe that there is any outright suspicion of discrepancy and nothing further has been done to date.

Should you have any other concerns regarding the forcing data or model parameters, please let me know. Although at this date these concerns will be discussion points at the workshop, I believe any discussion will be more productive if I have some fore-warning to generate summary analyses.

February 12, 2001: PILPS Errors

A few notes:

  1. In the new forcing data, I have inadvertantly swapped the y-axis direction. I switched from using dimensions of 'lat/lon' to dimension names of 'x/y' because the more general names are preferable to ALMA. Unfortunately, I did not realize that I was swapping the row order when the data is output from netcdf. I am very sorry for the confusion this caused. I would, however, prefer that all new output files are defined in this way (i.e. x, y dimensions, with y=1 corresponding to the lowest latitude). I currently have received a mixture of definitions with sometimes y=1 corresponding to the highest latitude (c standard, I realize), sometimes the lowest.
  2. The documentation file I sent out attached to the email last week was the old one. The new file is attached and the website versions were correct all along.
  3. The individual forcing file for 1983 was missing from the ftp site until this morning. If you downloaded the tar files everything should have been included.
  4. I will be contacting each of you individually regarding minor changes I would like to see in the format of your output files, whether you have told me you are doing the re-runs or not. If you are not doing the re-runs now, you can simply file this information away until after the workshop.
  5. If it is easier for you to return your completed re-runs via ftp, that is perfectly acceptable. File transfer for the first round went much more smoothly than I anticipated.
  6. If you are coming to the workshop, please remember to reserve your hotel this week. We may have difficulties finding rooms for you after February 17th.

February 19, 2001: PILPS 2e: water and energy balance checks

To all,

Attached is a new version of the program to perform basin-wide energy balance checks for your model output. I have also included a similar program to perform water balance checks. The new version is a little smarter than the original - the original energy balance program would invert the latitude axis in the output file in the case when maximum latitude was written first in the input file. This version will preserve the order of the input file.

I think that's all for now,


March 5, 2001: PILPS 2e Workshop

To all PILPS 2e particpants,

Thank-you to all who expressed concern regarding the recent earthquake in Seattle. Damage at the University was minimal and the workshop will proceed as scheduled.

Workshop agenda: An agenda for the workshop follows. This is meant to provide a framework; we anticipate that the actual agenda will be very fluid.

Workshop attendees: A list of workshop attendees follows the agenda. Please note that we would like the representative from each modeling group to give a 5 minute presentation on your experience with this project. Rather than just a review of model structure, this would include aspects of model structure which particularly pertain to this application and may render your results unique, as well as any problem spots you experienced.

Results reminder: We cannot guarantee that results received after March 11th will be included in the workshop analysis. I would still like to receive documentation from all models prior to the workshop.



May 31, 2001: PILPS 2e Status

Since I haven't written a group email since March, I will start by thanking everyone who participated in the workshop and provided ideas, suggestions and feedback.

In the weeks since the workshop I have been in contact with many of you and I have received rerun results and revised base run results in many cases. As you will recall, at the workshop, we set May 15th as the deadline for final submission of results. So let this serve as a reminder to those of you with outstanding results or unresolved errors in the base runs submissions!

The draft workshop report is now available on the PILPS 2e website. The figures include the rerun results for the models reporting back as of two weeks ago. I will update the figures once all of the results are finalized. Please pay special attention to the brief publication plan given in Section 4.5 and feel free to provide me with any comments. I will be back with more discussion on this in the coming weeks.

I know of a couple of participants that have ideas for additional related papers. I would encourage anyone writing additional papers using this dataset to consider submitting them simultaneously with the three group submissions, to maximize visibility.



Laura Bowling
Last modified: Wed Jun 20 14:39:21 PDT 2001