Global and United States Southeast Assessment of Precipitation
Total Page:16
File Type:pdf, Size:1020Kb
Computational Sciences and Engineering Division
Global and United States Southeast Assessment of Precipitation: Comparison of Model Simulations from the Intergovernmental Panel on Climate Change with Reanalysis-based Observations
Eduardo Ponce Mojica Research Alliance in Math and Science (RAMS) Polytechnic University of Puerto Rico
August 2009
Auroop R. Ganguly
Prepared by OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37831-6285 Managed by UT-BATTELLE, LLC for the U.S. DEPARTMENT OF ENERGY Table of Contents
I. Introduction…………………………………………………………………1 II. Methods……………………………………………………………………. 2 III. Future Work………………………………………………………………...4 IV. Discussion…………………………………………………………………..4 V. Acknowledgments………………………………………………………….. 5 VI. References…………………………………………………………………..5 VII. List of Figures………………………………………………………………6 Global and United States Southeast Assessment of Precipitation: Comparison of Model Simulations from the Intergovernmental Panel on Climate Change with Reanalysis-based Observations Eduardo Ponce Mojica1*, Auroop R. Ganguly2, Shih-Chieh Kao2, and Karsten Steinhaeuser2 1Research Alliance in Math and Science, Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831 2Geographic Information Science and Technology Group, Computational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831 3Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, IN 46556 *Corresponding Author Information: Eduardo Ponce Mojica Polytechnic University of Puerto Rico, 377 Ponce de León Ave.; Hato Rey, PR 00918 Phone: (787) 392 – 7257; Email: [email protected]
Abstract Climate change is likely to cause significant stresses on the environment, ecology, and society owing to increased intensity-duration-frequency of extreme hydro-meteorological events and large regional changes in weather and hydrology. However, precipitation measurements remain among the hardest to predict from climate and atmospheric models. We compare and evaluate two of the most comprehensive and commonly used climate models used by the Intergovernmental Panel on Climate Change (IPCC) in terms of their ability to project precipitation patterns. The Community Climate System Model version 3 (CCSM3) and the Hadley Center Coupled Model version 3 (HadCM3) models are compared and evaluated. Statistical approaches for climate data, developed using the MATLAB software package, are used to compare and contrast the model simulations in addition to visual comparisons. The Southeastern region of the United States and the Earth are selected as the case study regions, with a focus on the time period from 1948 to 1999. Given the importance of precipitation on climate change adaptation and mitigation, the comparison of two of the most commonly used climate models may offer guidelines and uncertainty bounds for stakeholders and policy makers.
I. INTRODUCTION
Climate changes have been a big concern for humanity for the past decades. Complicated phenomena have occurred all over world, injecting fear in people’s minds. Global warming, hurricanes, floods, droughts, earthquakes, tornados, and other natural events are thought to be increasing. All of these extreme events occur because of atmospheric, ocean, and land processes. Climate is composed of numerous meteorological elements, like temperature, atmospheric pressure, wind, rainfall, and evapotranspiration. Many scientists attribute anthropogenic effects to be the principal suspect of creating these phenomena. Anthropogenic is defined as processes or materials derived from human or human-related activities [1]. For example, the large expulsion of greenhouse gases into our atmosphere and the decreasing of vegetation and animals. Changes in atmospheric concentrations of greenhouse gases and aerosols, land cover, and solar radiation alter the energy balance of the climate system. One topic that stands out and is an important aspect to take a close look at is the increase or decrease in precipitation. Precipitation itself is any product due to condensation in the atmosphere that drops onto Earth’s surface [1]. Rain, ice pellets, snow, and hail are the main examples of precipitation. The study of precipitation and its effects has been of great interest to many class world scientists, as well as to nations all over the globe. It has come to such an interest that intergovernmental agencies are being created to find and study plausible solutions to mitigate irregular climate changes. One agency that is responsible for releasing articles, journals, and climate data is the International Panel on Climate Change (IPCC). The Intergovernmental Panel on Climate Change is the leading body for the assessment of climate change, established by the United Nations Environment Programme (UNEP) and the World Meteorological Organization (WMO) to provide the world with a clear scientific view on the current state of climate change and its potential environmental and socio economic consequences. The IPCC is a scientific body. It reviews and assesses the most recent scientific, technical and socio economic information produced worldwide relevant to the understanding of climate change [2]. The IPCC exposes precipitation data for any researcher to enable him the conveyance of new projects. Climate models have been created with extensive algorithms to try to reproduce precipitation behavior for the next few years. The drawback with modeling and simulating is that precipitation is very difficult to predict due to its immense complexity. Atmosphere processes are still being studied and only, incorporating assumptions, simulations with a degree of uncertainty can be realized. As powerful as climate models are, there is much about Earth’s climate that still cannot be simulated. Too many parameters are taken into account during the construction of precipitation models, e.g. ocean circulation, land surface, sea ice, concentration of atmospheric gases, electromagnetic radiation, etc. Also, the complexity of meteorological physics abruptly increases uncertainties in the models. Scientists need to understand mass and energy transfer and radiant exchange. On the other hand, precipitation events may be studied for a specific region, or across the whole Earth. The models are used to make global and regional comparisons and conclude what is happening and what will happen in respect to precipitation. During the following pages, two climate models will be compared with reanalysis based observations. Statistical methods or analysis will be used to describe the models. Finally, conclusions and uncertainties will be extracted from the climate models and from the observations.
II. METHODS
In this research two climate models were analyzed and compared with the solely purpose of establishing which model fits better the reanalysis observations. For the research, several resources were needed and will be mentioned promptly. The resources are: two climate models simulations, one climate observations data, MATLAB, and Microsoft Excel. The first climate model simulations data, Community Climate System Model version 3 (CCSM3), were produced in the United States. This data is publicly available and open source and distributed by the United States Department of Energy (DOE) from their Earth System Grid (ESG) database. The other climate model, also publicly available, was done by the United Kingdom and the data was distributed by the Intergovernmental Panel on Climate Change (IPCC) from their Project for Climate Model Diagnosis and Intercomparisons (PCMDI). Both of these countries, U.S. and U.K., work and collaborate to promote better science and contribute to climate issues currently concerning humanity. The climate observations, National Centers for Environmental Prediction reanalysis 1 (NCEP1), was also distributed and created by the United States. The National Oceanic and Atmospheric Administration (NOAA) was the agency responsible for obtaining the climate observations and publishing the data. Prior to computing statistics and creating plots and graphs, the dataset from the climate models and observations had to be normalized. Normalization refers to having the data of coinciding dimensions. The dimensions incorporated into the datasets were latitude, longitude, and time. Since the models have to be normalized with respect to the observations, a linear interpolation was conveyed. The observations’ precision was 94 latitudes and 192 longitudes. The climate models were 128 latitudes and 256 longitudes. The datasets were interpolated and the results were two climate models with 94 latitudes and 192 longitudes. The case study regions used to compare the models are the Southeast United States and a general study of the Earth. The states that are included in the Southeast region are Mississippi, Georgia, Alabama, North and South Carolina, Virginia, West Virginia, Florida, Kentucky, and Tennessee. The boundaries of the region are latitudes between 24° N and 41° N and longitudes 95° W and 74° W. Both case study regions are studied during the time range window from 1948 to 1999. It is in all these years that the model and observations were analyzed. Furthermore, some statistical methods or analysis were obtained from each model and then inserted into a table, see Figure 14. The statistics were applied to both, the Southeast U.S. and the global scope, and included the mean, standard deviation, skewness, median, bias, and variance. The models and observations were manipulated in MATLAB to produce those values. MATLAB is a numerical computing environment and fourth generation programming language. Developed by The MathWorks, MATLAB allows matrix manipulation, plotting of functions and data, implementation of algorithms, creation of user interfaces, and interfacing with programs in other languages[1]. The following lines will describe and compare the climate models. Also, a brief discussion of the observations of each figure will be established. Figure 1 shows the global average precipitation rate in mm/s of the observation data NCEP1. The lighter the blue color the less precipitation rate was measured for those regions. It can clearly be seen that the tropical and equatorial regions have more precipitation. The graph is an average of all the years between 1948 and 1999. The Polar Regions demonstrate a light blue which indicates less precipitation rate. Figure 2 displays the same graph as Figure 1 but it displays the result of the CCSM3. The same rules of colors apply to this figure and to the next three figures as well. A similar pattern of precipitation rates can be clearly seen in the figure but the color is darker. This darker blue means that the model simulated more rates in precipitation than actual observations. This over prediction may be due to the uncertainties injected through the interpolation method or the complexity of the atmosphere itself. The HadCM3 also over predicted for the global area and even more than the CCSM3. By observing the graph, the model almost over predicted by doubles the observation’s rate. The same graphs and analyses were done to the regional graphs. The first, Figure 4, refers to NCEP1 results and shows quite high rates for such a small area. The highest measurement is about 0.8 mm/s and a lowest of 0.1 mm/s. The CCSM3 graph, Figure 5, of Southeast U.S. clearly demonstrates under prediction. The land surface is fairly clear, approximately 0.2 mm/s in overall. The east part, Atlantic Ocean, is represented by a darker blue stating more precipitation. Similarly, Figure 6 shows the HadCM3 rates of the regional section. The average precipitation rates were also under predicted by this climate model. The same behavior as the CCSM3 can be observed. Another analysis done with MATLAB was the bias between the models and the observations. In the following descriptions, the red color states an over prediction and the darker the blue color states under predictions. In Figure 7, the CCSM3 was compared with NCEP1. An average of the biased precipitation rates was obtained and showed under predictions in land areas and over predictions in ocean areas. This is an interesting case because maybe oceanic attributes cause the models to over predict. The Figure 8 is the bias between the other model, HadCM3 and the observation, NCEP1. This model has a lot of irregularities in its color representation. It has under predictions in both land and water, over predictions in land and water, and several spots with no bias in land and water regions. Besides using MATLAB, another software was used, Microsoft Excel. Excel was used to create plots and trend lines to observe patterns and trends. The first plot, Figure 9, shows the Southeast U.S. precipitation rate per year of the CCSM3. A repetitive behavior can be observed were the rates range from 60 mm/year to almost 120 mm/year. The linear regression is represented by an almost horizontal line indicating a fairly constant result. Figure 10 shows the same data but for the HadCM3. The rates range higher, from 65 mm/year to about 160 mm/year. The trend shows an increasing pattern as years pass by. The following plot displays the averaged precipitation of the Southeast United States. The NCEP1 result shows a decrease until 1975 and then continues constant until 1999. An interesting observation is that the line for the HadCM3 shows an almost constant pattern and merges with the NCEP1 line from 1975 on. On the other hand, the CCSM3 remained in lower measurements throughout all the years. This low values show under prediction for the CCSM3. The Figure 12 demonstrates the Southeast U.S. averaged precipitation rates for each month throughout the time window (1948-1999). The NCEP1 has high values during the months of June, July, August, and September. Then, the values drop for the remaining moths. Both climate models took a similar pattern, increasing. The difference is that the values for CCSM3 are lower in almost all the months. Finally, the last graph, Figure 13, shows the global average precipitation rates for each month. Here not too many differences can be seen since when the data is analyzed, some regions of the Earth are summer, other fall, and other winter. This discrepancy causes that all the lines look alike.
III. FUTURE WORK
The study of precipitation earlier discussed leaves some aspects that still need to be clearly looked at. Since the only one region was studied with the CCSM3 and HadCM3 it is unwise to assume that similar results will be obtained in other regions of the Earth. Future researchers could examine and compare metropolitan areas in comparison with urban areas. This will provide a better overview and less uncertain deductions. By gaining better understanding of precipitation and its, safety measures for high precipitation areas can be proposed, thus, reducing risks and concomitant damage to property and people’s lives. Also, after designing a more reliable and complete climate model simulation can be done to forecast future trends in precipitation. Simulations done ranging from 2000 to 2100 will be a great acquisition for researches and policy makers.
IV. DISCUSSION
After observing and comparing the climate model, CCSM3 and HadCM3, with the observations, NCEP1, it was adequate to state that Earth climate is very complicated and improvements still need to be done. Both models demonstrated over predictions of precipitation rates when studying the global scope. Nevertheless, the CCSM3 showed greater accuracy globally and produced more reliable statistics. On the other hand, when analyzing the regional scope both climate models under predicted. The interesting finding was that in this region the United Kingdom model, HadCM3, was more accurate. Another issue to take into account is that more accurate results and statistics are obtained when studying smaller regions than the whole globe. By comparing smaller regions, uncertainties are greatly reduced. It is for these results that a better understanding of atmosphere and ocean physics is needed. Several governmental agencies collaborate together to keep enthusiasm among researchers and promote better researches.
V. ACKNOWLEDGMENTS
The Research Alliance in Math and Science program is sponsored by the Office of Advanced Scientific Computing Research, U.S. Department of Energy. The work was performed at the Oak Ridge National Laboratory, which is managed by UT-Battelle, LLC under Contract No. De- AC05- 00OR22725. This work has been authored by a contractor of the U.S. Government, accordingly, the U.S. Government retains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for U.S. Government purposes. The author would like to thank Dr. Auroop R. Ganguly for the opportunity to work on this research. Also, the author would like to thank Shih-Chieh Kao, Karsten Steinhaeuser, the GIST Group, and Rashida E. Askia for their continued support. Finally, special thanks go to Debbie McCoy, who made provisions for this research experience along with exceptional professional support.
VI. REFERENCES
[1] http://www.wikipedia.org/.
[2] Intergovernmental Panel on Climate Change, Fourth Assessment Report: 2007, 2007. [3] Rudra Pratap, Getting Started with MATLAB 7: A Quick Introduction for Scientists and Engineers, New York, New York: Oxford University Press, Inc., 2006. [4] Gabriele Villarini, Francesco Serinaldi, and Witold F. Krajewski, “Modeling Radar- Rainfall Estimation Uncertainties using Parametric and Non-parametric Approaches,” Advances in Water Resources, August 2008.
[5] Kuhn, G., S. Khan, A. R. Ganguly, and M. Branstetter, “Geospatial-temporal Dependence among Weekly Precipitation Extremes with Applications to Observations and Climate Model Simulations in South America,” Advances in Water Resources 30, 2007.
[6] Kevin E. Trenberth, Aiguo Dai, Roy M. Rasmussen, and David B. Parsons, “The Changing Character of Precipitation,” American Meteorological Society, March 2003.
[7] Auroop R. Ganguly, Shih-Chieh Kao, Karsten Steinhaeuser, Esther S. Parish, Marcia L. Branstetter, David J. Erickson III, and Nagendra Singh, “Uncertainties in the Assessments of Climate Change Impacts on Regional Hydrology and Water Resources,” 2009 (In review). [8] Alex Kirby, “Climate in Peril: A Popular Guide to the Latest IPCC Reports,” Birkeland Trykkeri, Norway: GRID-Arendal, 2009. VII. LIST OF FIGURES
N C E P 1 A v e r a g e P r e c i p i t a t i o n R a t e - 4 i n m m / s f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 2 8 0
6 0
4 0
2 0 e d u t
i 0 1 t a L - 2 0
- 4 0
- 6 0
- 8 0 0 - 1 5 0 - 1 0 0 - 5 0 0 5 0 1 0 0 1 5 0 L o n g i t u d e
Figure 1. Global average precipitation rate graph of NCEP1 C C S M 3 A v e r a g e P r e c i p i t a t i o n R a t e - 4 i n m m / s f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 2 8 0
6 0
4 0
2 0 e d u t i 0 1 t a L - 2 0
- 4 0
- 6 0
- 8 0 0 - 1 5 0 - 1 0 0 - 5 0 0 5 0 1 0 0 1 5 0 L o n g i t u d e
Figure 2. Global average precipitation rate graph of CCSM3
H a d C M 3 A v e r a g e P r e c i p i t a t i o n R a t e - 4 i n m m / s f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 2 8 0
6 0
4 0
2 0 e d u t i 0 1 t a L - 2 0
- 4 0
- 6 0
- 8 0 0 - 1 5 0 - 1 0 0 - 5 0 0 5 0 1 0 0 1 5 0 L o n g i t u d e
Figure 3. Global average precipitation rate graph of HadCM3 N C E P 1 S o u t h e a s t e r n U . S . A v e r a g e P r e c i p i t a t i o n R a t e - 4 i n m m / s f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 1 4 0 0 . 9 3 8 0 . 8
3 6 0 . 7
3 4 0 . 6 e d u t
i 0 . 5 t
a 3 2 L 0 . 4 3 0 0 . 3 2 8 0 . 2
2 6 0 . 1
2 4 0 - 9 5 - 9 0 - 8 5 - 8 0 - 7 5 L o n g i t u d e
Figure 4. Southeast U.S. average precipitation rate graph of NCEP1 C C S M 3 S o u t h e a s t e r n U . S . A v e r a g e P r e c i p i t a t i o n R a t e - 4 i n m m / s f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 1 4 0 0 . 9 3 8 0 . 8
3 6 0 . 7
3 4 0 . 6 e d u t
i 0 . 5 t
a 3 2 L 0 . 4 3 0 0 . 3 2 8 0 . 2
2 6 0 . 1
2 4 0 - 9 5 - 9 0 - 8 5 - 8 0 - 7 5 L o n g i t u d e
Figure 5. Southeast U.S. average precipitation rate graph of CCSM3 H a d C M 3 S o u t h e a s t e r n U . S . A v e r a g e P r e c i p i t a t i o n R a t e - 4 i n m m / s f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 1 4 0 0 . 9 3 8 0 . 8
3 6 0 . 7
3 4 0 . 6 e d u t
i 0 . 5 t
a 3 2 L 0 . 4 3 0 0 . 3 2 8 0 . 2
2 6 0 . 1
2 4 0 - 9 5 - 9 0 - 8 5 - 8 0 - 7 5 L o n g i t u d e
Figure 6. Southeast U.S. average precipitation rate graph of HadCM3
S o u t h e a s t e r n U . S . A v e r a g e o f B i a s e d P r e c i p i t a t i o n R a t e i n m m / s - 5 b e t w e e n t h e C C S M 3 a n d t h e N C E P 1 f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 2 4 0 1 . 5 3 8 1 3 6
0 . 5 3 4 e d u t
i 0 t
a 3 2 L
3 0 - 0 . 5
2 8 - 1
2 6 - 1 . 5
2 4 - 2 - 9 5 - 9 0 - 8 5 - 8 0 - 7 5 L o n g i t u d e
Figure 7. Southeast U.S. average of biased precipitation rate graph between NCEP1 and CCSM3 S o u t h e a s t e r n U . S . A v e r a g e o f B i a s e d P r e c i p i t a t i o n R a t e i n m m / s - 5 b e t w e e n t h e C C S M 3 a n d t h e N C E P 1 f r o m 1 9 4 8 t o 1 9 9 9 x 1 0 2 4 0 1 . 5 3 8 1 3 6
0 . 5 3 4 e d u t
i 0 t
a 3 2 L
3 0 - 0 . 5
2 8 - 1
2 6 - 1 . 5
2 4 - 2 - 9 5 - 9 0 - 8 5 - 8 0 - 7 5 L o n g i t u d e
Figure 8. Southeast U.S. average of biased precipitation rate graph between NCEP1 and HadCM3
Figure 9. Southeast U.S. plot of yearly precipitation rate of CCSM3
Figure 10. Southeast U.S. plot of yearly precipitation rates of HadCM3
Figure 11. Southeast U.S. yearly averages of precipitation rates
Figure 12. Southeast U.S. monthly averages of precipitation rates Figure 13. Global monthly averages of precipitation rates
Southeast United States Global HadCM CCSM3 3 NCEP1 CCSM3 HadCM3 NCEP1 4.23E- Mean 3.37E-05 05 4.29E-05 2.66E-05 2.74E-05 2.66E-05 1.90E- Std. Dev. 1.41E-05 05 2.30E-05 2.78E-05 3.23E-05 2.93E-05 4.00E- Variance 2.31E-10 10 6.20E-10 7.76E-10 1.05E-09 8.62E-10 0.81535 0.63958 2.21221 2.48467 1.87750 Skewness 2 0.69339 5 7 3 1 3.91E- Median 3.01E-05 05 3.91E-05 1.90E-05 1.77E-05 1.78E-05 5.58E- -2.90E- -8.22E- Bias 9.20E-06 07 08 07
*All values are expressed in mm/s
Figures 14. Statistics table of global and southeast U.S.