This guest post was written by Mark Webb, research scientist at the Met Office.
I work at the Met Office Hadley Centre and part of my job is to evaluate the representation of clouds in the computer models which we use for climate predictions.
One way in which we do this is to compare the clouds simulated by the models with observations of clouds from satellite instruments in orbit around the Earth. Unfortunately this process is complicated by the fact that satellites do not have a perfect view of all clouds. For example, low-level clouds are often not visible from space because of other clouds above them.
For this reason, we use satellite instrument ‘simulators’, which are computer codes designed to simulate what a satellite would see if it were observing our climate model. Simulator outputs can then be compared with observed satellite products in a quantitative way. This overcomes the ‘apples and oranges’ problem of comparing climate model clouds on all levels with observed clouds with some low-level clouds missing.
A number of simulators have been developed over the years, for a range of satellite instruments. These include operational weather satellites, which make ‘passive’ measurements of energy radiating from the Earth at infra-red and visible frequencies. More recently the approach has been extended to include new active instruments – for example, a cloud-profiling RADAR which analyses the time and intensity of a return from a pulse of radio waves sent into the atmosphere from the satellite.
I co-chair the Cloud Feedback Model Intercomparison Project (CFMIP), and as part of this I have been involved in an international effort to develop a software package which will bundle different simulators together and provide a consistent interface. This package is the CFMIP Observation Simulator Package (COSP). COSP had its first production release earlier this year, and is now being implemented in a number of climate models around the globe as part of the CFMIP.
COSP currently runs to about 15,000 lines of FORTRAN code, contributed by a number of academic and government organisations around the globe. The COSP glue layer and all but one of the instrument modules are available under a BSD licence. Although we liked to think of the project as being quite open, until recently we have had no formal governance model, and have been running the project under an informal form of benevolent dictatorship.
This all changed following a few conversations with Steve Lee of OSS Watch (who I know socially). Steve outlined a number of benefits to having a formal governance model. The first benefit that struck me was that a governance model would lay out clearly a number of roles and responsibilities which could encourage those who may not want to develop code to contribute in other ways, e.g. by testing, improving documentation, or helping with user queries. The other is that it lays out a clear decision-making process.
About the same time, our group of developers were discussing the possibility of securing some funding to employ someone to take responsibility for developing and improving the code. It seemed to us that a governance model would provide a useful foundation for any bids for funding, and would ensure that control of the project remained with those who have contributed over the years.
We found the OSS Watch information on governance models very useful for this. We had considerable debate about whether to move to the benevolent dictator or meritocratic governance model, but we eventually agreed that we would adopt the meritocratic model. We have now formed our Project Management Committee (all developers were invited) and appointed two co-chairs. We will be having our first teleconference to discuss the future of COSP in a couple of weeks.
Already I am seeing evidence of more shared responsibility for decision-making on the project, and I am looking forward to seeing the project grow, knowing that we now have a process in place for making decisions – hopefully one which will scale well as the project grows. I’m very grateful to Steve Lee and OSS Watch for the support that they have given – I doubt we would have made this transition without them.