6.5 GLOBAL WATER BALANCE ESTIMATED BY LAND SURFACE MODELS PARTICIPATED IN THE GSWP2
Taikan Oki, Naota Hanasaki, Yanjun Shen, Sinjiro Kanae Institute of Industrial Science, the University of Tokyo, Tokyo, Japan K. Masuda Frontier Research Center for Global Change, Yokohama, Japan and P. A. Dirmeyer Center for Ocean-Land-Atmosphere Studies, Maryland
1. INTRODUCTION year averaged annual mean value were The one of the objectives of the second phase of compared with previous estimates. Global mean, the Global Soil Wetness (GSWP-2) was targeted continental mean, and zonal mean values of to produce the best estimate of global water precipitation, evapotranspiration, and runoff were cycles for years from 1986 through 1995. Offline compared with historical publications, other simulation of the energy and water balance at datasets, and previous estimates. Global annual land surface was calculated by more than 10 precipitation increased 17% compared to the land surface models (LSMs), and sent to the forcing data used in the first phase of GSWP Inter-comparison Center (ICC) of the GSWP2, (GSWP-1). It is significant in Europe, where the founded at the Institute of Industrial Science, increase is approximately 60%. Corresponding University of Tokyo. It is expected that simple to this, annual runoff estimated by a simple ensemble mean or some statistical processing of Bucket model in Europe increased from 1,680 the submitted outputs from the LSMs will provide km3/y of GSWP-1 to 5,911 km3/y by GSWP-2. the state-of-the-art information on the global This value is extremely large relative to the energy and water balance. estimates of 2,900 km3/y by WMO (1997) and 2,770 km3/y by the University of New Hampshire 2. ANNUAL MEAN WATER BALANCE (1998). The forcing data of precipitation for the AVERAGED OVER THE GLOBE GSWP-2 was corrected considering the under Firstly, the water balance components of 10 catch by wind considering the rain gauge type in each country. It is suspected that this correction Corresponding Author Address: Taikan Oki, algorithm might have some deficiency since Institute of Industrial Science, University of national borders can be seen in the precipitation Tokyo, 4-6-1 Meguro-ku, Komaba, Tokyo data particularly in Northern Europe. It is 153-8505, Japan. possible that the original precipitation data in the Email: [email protected] corresponding European region had already corrected for wind under catch, and the GSWP-2 remaining LSMs are estimated considering the forcing data in the region was over-corrected. similarity of the LSMs. Linear weighting confirms More examination and sensitivity study should the conservation of mass in the water balance at be needed on this issue. each grid point. However, it was found, for 3. GLOBAL TERRESTRIAL WATER example, that the annual runoff is reduced due to BUDGETS AND THEIR INTER-ANNUAL the skewed probability distribution of runoff AND INTER-MODEL VARIATIONS among LSMs. Qualitatively, the difference The similarity of the global water balance among between the average of all the LSM runoff at a the outputs by various LSMs are examined for grid point and the highest value is larger than the annual mean field of runoff, annual amplitude of difference between the average and the lowest soil moisture in the root zone, the maximum soil value, and the excluding extreme values at both depth, ratio of interception loss in the total ends decreases. Even though it is not clear evapotranspiration, and the surface runoff ratio. which is the cause and which is the result, The misaligned outputs by a few LSMs were evapotranspiration becomes larger than simple excluded from further data analysis, and mean of the LSMs. weighted ensemble mean of outputs from
Global Terrestrial Water Budget
Unit: mm/year Legend 103 771 Inter- 71 97 108 annual 771 range 89 727 739 765 513 727 765 388 600 739 Snow 706 498 489 706 Rainfall ET Average of Inter-model 12 models range (1986-1995)
111 253 21 105 242 362 96 232 471 98 Surface Subsurface runoff 117 337 633 209 runoff 307 163 Total 51 161 343 runoff 160 Soil water storage
Figure 1: Simple Ensemble Mean of Annual Global Water Balance (mm/y) Averaged for 1986-1995. The inter-annual and inter-model variations are presented. This “Olympic Style” averaging method was and the rest (232mm/y) gradually discharge to introduced not to exclude whole data of any LSM the sea. Transiently, the water is stored as soil from the analysis, but more sophisticated moisture, and the annual mean value of the algorithm is certainly required. Therefore, simple stored water in the soil layer of the LSMs is 161 ensemble mean of all the models are used for mm. In total, approximately 40% of precipitation calculating global annual water balance over becomes runoff (337mm/y) on average, and the land (Figure 1). Even though the same rest is the evapotranspiration component precipitation forcing data were given for all the (498mm/y). Inter-model variation is dominant LSMs, the differentiation of snow and rain were particularly runoff and its components. Even done by each LSM, and there are discrepancies though the separation of surface and of the rainfall and snow fall in the final estimates. sub-surface runoff is more philosophical and On average, approximately 12% of annual difficult to observe, regional validation whether precipitation (836mm/y) precipitates as snow these surface and sub-surface runoff separation (97mm/y), and rest precipitates as rainfall over is realistic should be carried out using daily river land. Approximately 14% (105mm/y) of rainfall discharge data. (739mm/y) directly runs off with fast response,
Global Composition of Evapotranspiration
Unit: mm/year Legend 513 Inter- 388 600 annual 498 771 range
489 727 739 765
ET 706 Average of Inter-model 12 models Es Ei Ew Et range (1986-1995) 85 31 145 81 225 114 344 77 219 17 213 0 16 28 188 15 27 369 180 175
Figure 2: Simple Ensemble Mean of the Components of Annual Evapotranspiration (mm/y) Averaged for 1986-1995. The inter-annual and inter-model variations are presented. Figure 2 illustrates the components of the among the separation of evapotranspiration into evapotranspiration in the annual water balance. these components among LSMs are very large, On average over land, 16% (81mm/y) of total and should be presenting the large uncertainties evapotranspiration (498mm/y) is from in our current understanding of the processes. intercepted water. Evaporation from water Nevertheless, the inter-annual variation of the surface is comparatively low as 3% (16mm/y), total evapotranspiration is relatively small but it is partially because some LSM does not compared to that of precipitation and runoff. consider this component, but the maximum Figure 3 illustrates the snow related components estimates by a LSM is still small (28mm/y). Major averaged over snow possible area of 60.8 million components of the evapotranspiration are km2. For snow possible region, annual snow fall transpiration from vegetation leaves through is approximately 300 mm/y, and annual mean stomata (44%, 219mm/y) and bare soil snow storage is 55 mm on average. evaporation (36%, 180mm/y). The variations
Snow Balance in Cold Regions
Unit: mm/year
318 Legend 220 334 Inter- 299 annual 276 771 range
Snow fall 727 739 765 63 706 0 55 122 Average of 20 44 Inter-model 12 models Snow storage -17 4 26 range (1986-1995) -9 Change in Snow storage
Snow area: 60.8 x 106 km2
Figure 3: Annual Snow Processes (mm/y) averaged for 1986-1995 mean over snow possible area. The inter-annual and inter-model variations are presented.
Table 1: (A) Renewable water supply (km3/y), (B) Aggregated water use (km3/y), consists of (B1) Industrial (km3/y), (B2) Municipal/domestic (km3/y), (B3) Agricultural (km3/y), (C) Mean ratio of use to supply, namely (B) divided by (A), (D) Falkenmark’s “crowding index of people in terms of water” namely population per 10-3 km3/y water supply, and (E) number of people (millions) the ratio of use to supply (C) is more than 0.40. VARIABLE ASIA FSU Latin North OECD Sub-saharan Global America Africa (A) (km3/y) 7846 5902 11160 367 12058 4815 42294 (B) (km3/y) 1516 380 243 271 934 79 3423 (B1) 120 118 22 13 432 3 708 (B2) 99 34 30 20 131 8 321 (B3) 1297 228 192 238 372 68 2394 (C) 0.19 0.064 0.022 0.739 0.077 0.016 0.081 (D) (c/106m3/y) 384 48 42 929 115 115 133 (E) (106 capita) 712 56 84 91 16 16 1123
4. APPLICATION OF GLOBAL WATER climate change and societal changes such as BALANCE ESTIMATES FOR WATER population and economic growth that will affect RESOURCES ASSESSMENTS on water usage in the future. Practically, current As an application of the global estimates of GCM simulation results are not easy to be used annual water balance, the obtained natural water in the world water resources assessment cycle is contrasted with socio economic quantitatively because its biases in water cycles. information such as population and water Therefore reliable current estimates are often withdrawals, and analyzed in order to provide used as “baseline” of the reliable water cycle on basic information for world water resources the earth, and changes in the water cycles assessments. Table 1 summarizes the simulated in the GCMs will be applied to the assessment results for various regional “baseline” field of water cycles. Consequently, aggregations. Values related to anthropogenic how to estimate reliable current water cycle is of activities on water and social statistics are interest, and the frame work of global offline derived from Oki et al.(2001), and regional simulation like GSWP should be one of the aggregations follows Vörösmarty et al. (2000). promising methodologies. Such an analysis is important not only for current assessment and help promote awareness on 5. Summary water issues, but also for future assessment of Global offline simulations by more than 10 land world water resources considering both global surface models were done under the Global Soil Wetness Project Phase2, and the ensemble Technology (CREST), the Japan Science and mean dataset is analyzed to present new Technology Corporation (JST), Special insights in global water cycle. Global mean water Coordination Funds for Promoting Science and balance explicitly including the interception-loss Technology, rant-in-Aid for Scientific Research and transpiration, surface and sub-surface runoff, (B) from JSPS, and the Research Institute for snow ratio in precipitation and snow Humanity and Nature (RIHN). accumulation, etc., is presented with their inter-annual variations in 10 years and References: uncertainties inferred from the variance among Dirmeyer, P., X. Gao, and T. Oki, 2002: GSWP-2: estimates of various LSMs. Such a process The Second Global Soil Wetness Project based estimates did not existed in the past. Even Science and Implementation Plan. though they are derived from numerical models, International GEWEX Project Office, IGPO some of them are well validated, and it is Publication Series, No. 37, 65 pp. expected that the model bias is somewhat Falkenmark, M., J. Lundqvist, and C. Widstrand, eliminated by the averaging multiple model 1989: Macro-scale water scarcity requires estimates. micro-scale approaches; Aspects of The current results are preliminary, and more vulnerability in semi-arid development, detailed analysis and validation should be done. Natural Resources Forum, 14, No.11, Budyko’s diagram, which adopts net-radiation 258-267. over precipitation and evapotranspiration over Oki, T., Y. Agata, S. Kanae, T. Saruhashi, D. precipitation in annual basis as for major Yang, and K. Musiake, 2001: Global variables to explain water and energy balance, Assessment of Current Water Resources will be used in accordance with vegetation cover using Total Runoff Integrating Pathways, to classify the global lands into several climatic Hydrol. Sci. J., 46, 983-996. types. Representative water balance in these Oki, T., Y. Agata, S. Kanae, T. Saruhashi, and K. land use types is also presented. The global Musiake, 2003: Global Water Resources water balance figures shown here will be shown Assessment under Climatic Change in 2050 for particular land surface types and climatic using TRIP, IAHS Publication, 280, 124-133. zones. Water balance will be validated using Vörösmarty, C. J., Green, P., Salisbury, J. & river discharge and other observable quantities Lammers, R. B., 2000: Global water such as total water storage change derived from resources: vulnerability from climate change atmospheric water balance. and population growth. Science 289, 284–288. Acknowledgements: Parts of this study were supported by the Core Research for Evolutional Science and