Methods to establish the quality of simulated climatological data produced by numerical mesoscale modeling techniques

Many planning applications and simulations that are impacted by the environment require values in much greater detail than traditional observed climate databases can support. A method has been developed to simulate high-resolution historical conditions using a limited-area high-resolution numerical model initialized from available global gridded databases. The simulated conditions are then used to generate local climate statistics around the globe. With its database of the surface characteristics of the earth and the basic principles of physics, the model dynamically generates estimates of the historical conditions from which local climate statistics can be calculated at locations for which no observational data is available. This technique has been given the name CLImate statistics by a dynamical MODel (CLIMOD).

Research has now been completed for several different climate regimes around the world at model resolutions of 10 and 40 km. Recently the DOD Air and Space Natural Environment Modeling and Simulation Executive Agent (ASNEMSEA) has sponsored research to determine the quality of, and to establish a user confidence index for, the simulated climate statistics. A specialized method has been developed to help understand the quality of the simulated data and provide this information to CLIMOD users. There are three distinctive subdivisions to the method employed. (1) First, characteristic model biases are established for the various general climate regimes around the globe. (2) Second, verification of the model bias for the specific target climate area is performed. (3) Finally, an objective estimate of the accuracy and confidence a customer can have in the simulated climatology is determined.

The method developed to reasonably estimate the confidence a user can have that the simulated climate data depicts the actual climate of a given point, region, or grid involves four factors: (1) A General Climate factor. This factor accounts for a reduction of the confidence because of the typical model performance (bias) for that general type of climate. This information would be based on criteria for runs made in simulations made in analogous regions. This factor is designed to allow a user to input the knowledge gained from running in similar regions in the past. (2) Local Variability Factor. This factor is an attempt to quantify the ability of the model to resolve local variability for such factors as terrain, land-water distribution, soil type and vegetation. For example, in flat plain areas with uniform vegetation the model may be able to resolve 98% of the surface, but in very mountainous regions only 75%. (3) Target Area Comparison. This is calculated objectively by calculating the absolute value of the model bias found in the simulated versus observe climate comparisons made for the target climate region. This number is also used a quality control when running developing the simulated climatologies. If the number exceed a given threshold it would alert the user that a possible model running problem exists. (4) Subjective Adjustment Factor. It is recognized that there certain factors that still can't be effectively evaluated so this factor is settable by the user and a value entered based upon the experience of the user. In most cases this factor would be set to 0 most of the time, but it would give the model operator the option to make subjective adjustments.