URBA6006 Tsang Nok Sze 3035776660
Evaluation of Climate Model – Bias and Uncertainty in Climate Prediction
Academic Paper – Climate Model
Paper Title Model 1 Quantitative urban climate mapping based on a geographic database: A simulation approach using Hong Kong as a case study (Chen & Ng, 2011) GIS-based simulation approach – Means of SVF and FAD simulation 2 Applying urban climate model in prediction mode – evaluation of MUKLIMO_3 model performance for Austrian cities based on the summer period of 2019 (Hollósi et al., 2021) MUKLIMO_3 3 Reanalysis-driven climate simulation over CORDEX North America domain using the Canadian Regional Climate Model, version 5: model performance evaluation (Martynov et al., 2013) Canadian Regional Climate Model 4 Evaluation of extreme climate events using regional climate model for China (Ji & Kang, 2014) Regional Climate Model Version 4.0 5 Extreme climate events in China: IPCC-AR4 model evaluation and projection (Jiang et al., 2011) Regional Climate Model – IPCC AR4 6 A future climate scenario of regional changes in extreme climate events over China using the PRECIS climate model (Zhang et al., 2006) PRECIS, regional climate model system 7 Climate change in China in the 21st century assisted by a high-resolution regional climate model (Gao et al., 2012) Regional Climate Model version 3 (RegCM3) 8 Regional climate model downscaling projection of China future climate change (Liu, Gao & Liang, 2012) Regional Climate Model version 3 (RegCM3) 9 Changes in Extreme Climate Events in China Under 1.5°C–4°C Global Warming Targets: Projections Using an Ensemble of Regional Climate Model Simulations (Wu et al., 2020) Regional Climate Model (RegCM4) 10 Climate Change over China in the 21st Century as Simulated by BCC_CSM1.1-RegCM4.0 (Gao, Wang & Giorgi, 2013) Regional Climate Model (RegCM4)
Introduction
The climate model is an extension of weather forecasting, it usually predicts how average conditions will change in a region over the coming decades (Harper, 2018). To understand how to evaluate a climate model, we should understand the components of a climate system. A climate system is a system combining the atmosphere, ocean, cryosphere and biota, therefore, there are lots of parameters that will affect the climate situation of a region.
The climate model is usually used by researchers to understand complex earth systems. The model inputs will be the past climate data which acts as a starting point for typical climate systems analysis and a model can be created and used to predict the future climatic situation as the model output. Therefore, the more we learn from the past and present climatic situation, the more accuracy of the model to predict the future climatic situation.
Model accuracy and precision depended on the following three major parts, including input, which is related to the data quality and quantity; model which depended on the quality and quantity of parameters, temporal and spatial extent settings; and output, which is about the accuracy and precision of the forecasting of the model.
URBA6006 Tsang Nok Sze 3035776660
Evaluation
A) Complexity of model
Problem of parameters
There are increasing statistical methods of multimode climate projections, the complexity of the model in analyzing different parameters also hence to enhance to predict different possibilities of the future climatic situation. However, most of the researchers mentioned in this paper are only interested in ranking the importance of the different parameters in affecting and controlling the climate system. They will try to do some correlation between the parameters and the climate result to find which parameters should be included in the climate model for prediction and analysis. However, what we need to focus on is how these models predict the changes in the climate of the region, their ability to predict the accurate trends of the climatic situation. It is important to note the complexity of the climate model is not in a linear relationship with its accuracy in predicting future trends.
B) Uncertainty and Bias of the model
The uncertainty of the model causing overestimation and underestimation of the model in predicting the temperature and precipitation.
The issue of uncertainty and bias are the core parts of the climate change prediction problem. Due to the complexity of these issues on both concept and specialty, uncertainty and bias will remain an inevitable issue in the debate of climate change.
The problem of topography
As indicated by much research on climate models based in China, the problem of topography is the major limitation for the collection of data in the first stage. China is known as a country with complicated topography, including mountains, basins, plateaus, hills, and plains. It is important to note that complicated topography largely affects the climate model’s stability (Mesinger & Veljovic, 2020), and this topography characteristic has been reviewed by Martynov et al. (2013), Jiang et al (2011) and Zhang et al (2006) as the barriers in data collection.
For example, as stated in research of Martynov et al (2013), the horizontal resolution in the climate simulation is insufficient for such a complex topographic situation, while the vertical interpolation of the pressure gradient simulation is also affected by the complex topographic factors. Similar to the results as stated in the research of Jiang et al (2011), the complexity of the topology in China also affects the accuracy of the model in predicting future precipitation, especially for the case of topography-driven precipitation, the related data is not well measured and recorded by the coarse resolution model. Mountainous regions of China also induced bias issues. Some weather stations located in the valley or low elevation regions may also result in the cold bias of the climate modelling results. As reviewed in the regional climate model in research of Zhang et al (2006), the operation of complex topography in China with the strong monsoon system causing a large spatial variability in the prediction accuracy of the climate system.
The problem of humidity
Both humidity and temperature are the major components in the climate model while humidity has long struggled in the climate models in whether it has been adequately represented the cloud systems to tropospheric humidity in the calculation of the climate system. In the research done by Ji & Kang (2014), the factor of humidity in the formulation of climate systems becomes the greatest uncertainty in climate model prediction. The climate model stated in Ji & Kang (2014) research also indicated the relative humidity prediction appears to be much less credible and show a large variety of model prediction skills.
URBA6006 Tsang Nok Sze 3035776660
It is necessary to include a comprehensive analysis of the dynamic cloud processes so to evaluate the humidity effect in the climate model. Moreover, humidity is highly variable over small scales of time and space, which is a huge uncertainty for the regional climate model, this will lead to a large range of potential results in the future, directly affecting the forecasting ability of the model. (Maslin & Austin, 2012).
The availability of observational data
Climate observations are used as a baseline for accessing climate changes. As revealed in some researches, complicated topography that falls within a large range of elevation largely affect data quality and quantities of climate data collected. For instance, the temperature and humidity related data are hardly collected. For example, for the Hollósi et al (2021) research on applying climate models for Austrian cities, the problem of uneven distribution of weather stations is found. In other cities of Austria, because of the limited number and sparsely placed data collection stations, there are much less observational data of some rural regions. Even if the cities have a relatively high amount of weather stations, due to the building geometry differences between rural and urban cities environmental setting, some patterns such as heat load is not properly investigated and monitored.
Therefore, the quality and quantities of the observational data are not stable and reliable for some climate modes, resulting in large uncertainties and difficulties when analyzing the climatic difference between urban and rural areas.
C) The forecasting ability of the model
The limited forecasting ability of the climate model is not inevitable. It is so hard to predict climate changes, which highly depends on the data quality measured and captured by the measurement stations or equipment (Maslin & Austin, 2012). Also, our atmospheric structure is so complicated and the climatic situation is affected by many external factors that cannot be analyzed and found out by one single climatic model (Herrington, 2019).
The problem of using past climatic data in predicting extreme weather
It is important to note that climate has changed so extremely and intensely that the frequency of past extreme events is no longer able predictor, especially for the human-induced warming has on the extreme events. Hence, the use of temporally lagged periods of extreme events probably will probably underestimate the historical impacts, and also underrate the risks of the occurrence of extreme weather.
As stated by Foley (2010), the technique that using historical observation data to calibrate future model projections is not precise enough when the model is trying to simulate and validate a state of the system that has not been experienced before. This is an inevitable barrier for the model computations of the natural systems.
Researches done by Ji & Kang (2014), Jiang et al (2011) and Gao, Wang & Giorgi (2013) tries to predict extreme weather by using the historical data at different ranges, basically using the range of the temperature as the observational data as the input of the model. Sometimes the problem of complicated topography of China will also induce large biases in the collection of climatic data, includes the daily mean temperature and the records minimum and maximum temperature. As mentioned by Sillmann et al., (2017), predicting extreme weather needed to depend on the presence of large scale drivers, which should be the major contributors to the existence of extreme weather.
Therefore, instead of using the separate dynamic and physical processes in the predictive model to predict climate changes as stated in research Ji & Kang (2014), Jiang et al (2011) and Gao, Wang & Giorgi (2013), the researches should focus on the interrelationship between the processes, a better understanding of the processes can allow us to realize the underlying drivers of the results of extreme weather.
URBA6006 Tsang Nok Sze 3035776660
Overestimation and Underestimation
The climate models overestimated the interannual variability of temperature. As indicated in the Ji & Kang (2014) research, the network of precipitation patterns that are processed from stations in the arid areas may underestimate the precipitation over the northern topography of China. While the Jiang et al (2011) research indicated the regional climate model tends to overestimate the precipitation situation in the northern and western parts of China where intense precipitation is rarely found. On the other hand, the climate model also underestimated the precipitation that will exist in the southern and northeastern parts of China in the future. A similar result was also found in the Zhang et al (2006) research, which indicated that the climate model underestimated the existence of extreme precipitation events in the southern part of China.
For the climate model researches done in Hong Kong (Chen & Ng, 2011), only building geometry is taken into consideration in climate simulation, both topography and vegetation cover are not included, indicated that the results may overestimate the real temperature for the location located in higher elevation with large vegetation cover.
Limitation of the Regional Simulations in Regional Climate Model
Most of the researches indicated in this paper focus on the regional climate model, which is the higher resolution model compared to the global climate model. Therefore, with a finer resolution of the regional climate model, scientists can have a higher ability in resolving mesoscale phenomena that contributing to heavy precipitation (Jones, Murphy & Noguer, 1995). However, as the regional climate model only cover certain parts of the continental, the lateral boundary condition is required in the model simulation. Therefore the accuracy of regional simulations is highly dependent on the boundary conditions of the observations. When the regional climate model is affected by some cross-boundary external forcings, uncertainties must have easily existed when the climate model trying to forecast or project the future climate in boundary conditions. (CCSP, 2008)
Conclusion
Formulation and using a climate model to analyze the climate data and making the prediction is becoming a new trend for scientists and researchers to enhance our understandings of the earth we lived on. With the increased complexity of the climate model, more and more factors are putting into considerations when we trying to predict the climate situation. However, despite the climate model are more sophisticated in today’s society, biases and uncertainties still existed, but we should also need to understand that there is no perfect model with no bias and uncertainty. As long as the climate model is able to ensure and decide the sensitivity of the actual climate system to small external drivers, the weight of scientific evidence is already enough to give us the information and make it acceptable prediction of the climatic conditions of our world.
URBA6006 Tsang Nok Sze 3035776660
Reference List:
CCSP, 2008: Climate Models: An Assessment of Strengths and Limitations. A Report by the US Climate Change Science Program and the Subcommittee on Global Change Research [Bader DC, C. Covey, WJ Gutowski Jr., IM Held, KE Kunkel, RL Miller, RT Tokmakian and MH Zhang (Authors)]. Department of Energy, Office of Biological and Environmental Research, Washington, DC, USA, 124pp