An application of RS and GIS for hydrological modeling in continental scale

An application of RS and GIS for hydrological modeling in continental scale


Ai Nakagawa, Shiro Ochi and Ryosuke Shibaski
Institute of Industrial Science, University of Tokyo
7-22-1, Roppongi, Minato-ku, Tokyo 106-8558, Japan
Tel: +81-3-3402-6231 Fax: +81-3-3479-2762
E-mail-: [email protected]

When thinking about food production to sustain world population, water resource is an important factor to define it. The quantitative analysis for available water is necessary to evaluate the cultivation potential which is enhanced by irrigation system. A hydrological model based on “Bucket Model” using Remote Sensing data and GIS is proposed in this paper to understand monthly distribution of water resource in continental map scale. When a precipitation is given to a pixel, the runoff from the pixel is calculated using “Bucket Model” in which the balance of Ep versus E (Evapotranspiration), and Sp versus soil wetness are considered. Monthly river runoff and soil wetness of each pixel with 0.5° grid are computed and simulated.

1. Introduction
The subject that securing enough food production becomes a very serious problem is pointed out from a global viewpoint. It is imagined that it is maybe impossible to expand agricultural area simply, so it is an important focus of discussion how far the expansion of agricultural productivity can be. Expected agricultural productivity depends on progress of technology; expansion of crop area or cultivation period by irrigation, increase of crops by improvement of species and wide advancement of agricultural land control or water control. Moreover, it is much influenced by deterioration of the soil etc. It is said that not only an artificial condition but also a natural condition of deterioration of the soil or availability of water resource etc. has much influence on the productivity. In this research, the availability of water resource of these condition is listened up and monthly available water is calculated in continental scale. Concretely, first, Ep(potential evapotranspiration) is calculated by using the temperature data presumed from remotely sensed data. And by using this, Et(evapotranspiration) and St(soil wetness in a pixel) are computed by hydrological modeling based on “Bucket Model”. . In addition, the behavior of the runoff water is modeled by Distribution Model, and monthly river runoff and monthly available water of each pixel.

2. Construction of the Model

2.1 Outline of the Model
When it rains, the part of rain water is intercepted by leaves of the trees or the crop grasses while the other part falls to the ground directly. A part of the rain water which falls to the ground floods to the earth and the remainder gathers to the lower land. Rain water stocked in the low ground increases it depth because of the continuance of the rainfall and finally flows out to lakes and marshes and oceans as a stream. The rain water which flows out to the ground infiltrates in perpendicular direction. When it reaches the layer which water easily permeates, it flows horizontally along the layer and flows out to the ground or the river on the way. Moreover, water evaporates from leaves, the ground, and the surface of the water and spreads into the air. In this research, the mechanism is simplified as follows. The behavior of water from falling to the ground to runoff is represented by “Bucket Model” in which the balance of Ep versus Et, and Sp(potential soil wetness in a pixel) versus St are considered. And the behavior of runoff water is represented by distributed model in which inclination of the ground is considered.

2.2 Bucket Model
This model supposes that water runs off only when the amount of the rain fall of one month and the water stock unil last month having subtracted the evapotranspiration exceed the potential water stock. (Figure 1).

Figure 1. Bucket Model
The relation between Ep (potential evapotranspiration), Et (evapotranspiration), Sp (potential soil wetness) and St(soil wetness) is assumed such as Figure 2. This shows, that evapotranspiration occurs only when some ground water exists, and that evapotranspiration from ground becomes constant and he surplus runs off when the wetness of the land exceeds the potential water stock.

Figure 2. the relation between Et/ Ep and Et /Sp
2.3 Distribution Model
Distribution model used at this time assumed that the runoff from a certain pixel flows only to the pixel in the surface flow direction which is made from digital elevation model (DEM). Further, the amount of the runoff from each pixel is by the Bucket described in the foregoing paragraph.

3. Method of Estimation of Parameters

3.1 Precipitation
IIASA database for mean monthly which UN/GRID offered is used. The data is 0.5°grid.

3.2 Potential Soil Wetness
Soil Water Holding Capacity of Documentation Summary for Data Set which Un/GRID offered is used. The data is 1.0°grid.

3.3 Potential Evapotranspiration
Potential evapotranspiration is calculated by Thornthwaite method which ad only one variable temperature which can be derived from IIASA database for men monthly value offered by UN/GRID.
<Thornthwaite’s Formula>


a=0.000000675J3-0.0000771J2+0.01792J+0.49293         (1)

Ep = : Monthly potential evapotranspiration (mm/day), D0 : possible irradiation hours (12 h/day),
tj : mean temperature in jth month (°C)

3.4 Evapotranspiration
Evapotranspiration can be presumed from the relation explained in Figure 1 and Figure 2 as follows.

. for St-1 + Pt- Et > = S   (2)
Et = Ep    
. for St-1 + Pt- Et < Sp    
Et = (St / Sp-k) * Ep / (1-k)   (3)

3.5 Soil Wetness
Soil wetness can be presumed from the relation explained in Figure 1 and Figure 2 s follows.

. for St-1 + Pt-Et >=Sp    
St=Sp   (4)
. for St-1 + Pt-Et<Sp    
St= St-1+Pt-Et   (5)

3.6 Runoff and Discharge of the River
Runoff from each pixel can be presumed from the relation explained in Figure 1 as follows.

. for ST-1+Pt-Et>=Sp  
Rot = Pt+ St- Ep- Sp (6)
. for St-1 + Pt- Et < Sp  
Rot = 0 (7)

It is supposed that the all runoff from each pixel flows into the river and to flow out to the downstream according to the gradient of ground. Distribution Model is used for calculating the runoff of water body in the river. The monthly water discharge is the sum total of water passed each pixel in one month. Flow velocity is obtained as follows.

3.7 Flow Velocity
Kinematic Wave model is used or the calculation of flow velocity. Flow velocity is obtained by the next formula. However, Manning’s roughness coefficient is constant as 0.05 and the river road section is assumed in the next paragraph.

n : velocity (m/s) , n : Manning’s roughness coefficient, R : depth (m), i : gradient, Q : quantity of flow (m3/s) , A : cross section (m2)

3.8 River Road Cross Section
The river road cross section assumes the following rectangular sections.

Figure 3 river road section 4. Results
Monthly calculation results in 1990 of each of the following items are obtained.

Figure 4 evapotranspiration, Jul

Figure 5 evapotranspiration, Nov

Figure 6 soil wetness, Jul

Figure 7 soil wetness, Nov

Figure 8 runoff, Jul

Figure 9 runoff , Nov
In this research, the evapotranspiration, the soil wetness and the water runoff of the Amazon basin in South America in 1990 are calculated with 0.5° grid. Figure 4-9 shows the result in November and June. Further, rainy season in this area is during October-April, and dry season is during May-September.

5. Discussion
The evapotranspiration, the soil wetness and the water runoff of the Amazon basin in South America in 1990 are calculated above. The behavior of the water runoff from each pixel is grasped by using the gradient data and the monthly amount of available water of each pixel is calculated. In addition, it is verified by using the discharge data by Grobal Runoff Data Center (GRDC).

Moreover, still Thornthwaite Method is used at this time when the potential evapotranspiration is obtained, strong correlation of evapotranspiration and NDVI is pointed out by some current researches. Therefore, it is also possible to calculate the amount of available water which is used by evapotranspiration related to land cover classification. While I used the current data about potential soil wetness at this time, it is possible to use the potential soil wetness which is related to the land cover classification by remotely sensing data. The comparison of those results is to be desired.

This research is funded by “Research for the Future” program of Japan Society for Promotion of Science.


  • Hidetaka Chikamori, Tara Oka, Kaoru Takara and Go Okubo : Study on Application of Geographic Information System to Construction of Rainfall-Runoff Model, Theory and Applications of GIS, 6 No. 1, pp 19-28, 1998.
  • Masaki Sawamoto and Shachul Shin: Water Balance Evaluation in the Korean Peninsula through Vegetation Monitoring, Journal of Remote sensing society of Japan, 17, pp. 4-15, 1997.
  • Mutumi Kadoya: technique of outflow analysis No 1, Journal of the Japanese Society of Irrigation, Drainage and Reclamation Engineering, 47, pp811-821
  • Mutumi Kadoya: technique of outflow analysis No 6, Journal of the Japanese Society of Irrigation, Drainage and Reclamation Engineering, 48, pp419-425
  • Shachul Shin, Tuyoshi Tada, So Kazama and Masaki Sawamoto: Estimation of Spatial distribution of evapotranspiration in Korea, Journal of Japanese Association of Hydrological, 38, pp. 161-166, 1994.