International Water Management Institute
Colombo, Sri Lanka
Email: [email protected]
Since availability of fresh water is limited, managing water among different uses and users is becoming a central issue. A common problem is a lack of information on water utilization. This information should include water consumption through actual evapotranspiration and the relationship between water consumption and biomass production in different parts of agricultural lands. Using this information, water and land use can be linked in order to get optimal and sustainable use of natural resources.
Satellite remote sensing gives opportunities to monitor land surface conditions and the status of water resources on different spatial and temporal resolutions. The NOAA-AVHRR sensor supplies daily data at 1km spatial resolution. The low cost satellite data coupled with physically based models enables the estimation of parameters such as vegetation growth and actual evapotranspiration that are paramount information to water managers.
Since vegetation reflects the climatic, soil and terrain conditions, vegetation growth can be used as a single indicator for land use classification. Conventional classification systems use different types of data in different scales. For example, soil maps give homogeneous soil units, but the climate data on the same regions, based on point data, may not be homogeneously distributed over the whole area. The currently available Sri Lanka agro-ecological zones map based on rainfall, soils and elevation—divides the country into 24 regions (Panabokke 1996). But by looking at vegetation growth patterns it is possible to identify different areas with different potentials of vegetation growth inside a single zone of this map. Therefore, spatial and temporal variation of vegetation growth, which gives the result of all the parameters responsible for vegetation growth, enables a more refined classification rather than one based on a few parameters.
This paper aims at discussing demarcation of Sri Lanka into different zones in terms of spatial and temporal variation of vegetation growth and demonstrates the feasibility of applying such a technique using a case study.
Sri Lanka is situated within latitudes North 90 50’N, South 50 55’N and longitudes East 810 53’E and West 790 42’ E. The aerial extent of the country is 65,610 km2. The mean annual rainfall in Sri Lanka varies from 2,500mm to over 5,000 mm in the southwest of the island. This rainfall is mainly from two monsoons: the southwest monsoon from May to September and the northeast monsoon from December to February. The rainfall during the two inter-monsoonal periods, march to April and Oct. to Nov. is mainly convectional and occurs over the whole island. The regional difference in temperature is mainly due to altitude only and shows no dependence on latitude. The mean monthly temperature differs slightly throughout the year depending on the seasonal movement of the sun, with some modifying influence caused by rainfall. In the lowlands, the mean annual temperature is 27.5 0C and at Nuwara Eliya at an altitude of 1800 m above mean sea level, the mean temperature is 15.9 0C. Sri Lanka is categorized into three elevation classes: up country, mid country and low country. Low country is demarcated as land below an elevation of 300 meters, up country is above 900 meters and mid country is between 300 and 900 meters in elevation.
Materials and Method
In this study, NOAA-AVHRR images acquired and processed by the department of meteorology, Sri Lanka, were used. The study is based on a series of 10-day composites of NOAA images, which cover entire Sri Lanka during a complete annual cycle started in July 1999. The actual evapotranspiration and evaporative fraction were estimated with the surface energy balance algorithm for land (SEBAL) (Bastiaanssen 1998). In addition to that, indicators such as vegetation growth, soil moisture, potential evapotranspiration and surface temperature were derived (Chandrapala 2002). The accumulation of biomass or primary production is according to the Monteith model (1972) related to accumulated absorbed photosynthetical active radiation (APAR). Net primary biomass production was calculated with a minimum of ground data by considering physical models. Soil moisture was estimated using the statistical relationship between evaporative fraction and soil moisture (Bastiaanssen et.al. 2000).
Based on this dataset, monthly vegetation growth was estimated over the country during the study period (Samarasinghe et al. 2002). Using these estimates, the whole country was divided into 91 vegetation growth zones with respect to the spatial and temporal variations of vegetation growth (Muthuwatta et al. 2002). These zones were named and described by the degree of wetness and elevation. Wetness was determined, in addition to soil moisture, using the moisture availability index (MAI) (Heargreaves, 1975). Areas where annual MAI is below 1.0, from 1.0 to 1.4 and above 1.4 were classified as dry (D), intermediate (I) and wet (W) regions, respectively. Of these zones, 60 came under low country, 8 under low country and 7 under up country. In addition to that, 10 zones were distributed between the low country and the mid country, and 6 zones were distributed between the mid country and the up country. Two letters and a number were assigned to each zone: the first letter refers to the MAI zone and the second to the elevation class (U for up country, M for mid country and L for low country), and the number refers to the MAI in ascending order. For example, WL1 gives the area with the lowest MAI in the wet areas of the low country (Muthuwatta et al. 2002).
A case study carried out for the DL1 region is indicated in the existing agro-ecological zone map to present the usefulness of the new zoning technique (figure 1). This DL1 zone belongs to the dry part of the country. The dry zone of the country is classified into 5 regions namely DL1, DL2, DL3, DL4 and DL5. DL1 has a 75 percent probability annual rainfall amounting to 775 mm. The soils in DL1 are predominantly reddish brown earths and low humid grey soils. More than 50 percent of the total irrigated lands of the dry zone are situated in this region. Rice is the main crop in the irrigated lowland. There are 29 major vegetation growth zones identified inside the DL1 zone. In the existing agro-ecological zone map the whole area is classified as a dry zone region. However, in the new classification system, there are 3 zones identified within the intermediate zone out of these 29 vegetation growth zones (1<MAI <1.4). The P75 histograms for these 3 zones also significantly deviate from the histogram associated with the existing DL1 zone.
Figure 1: Agro-ecological regions and Vegetation growth
zones within the DL1 zone
Figure 2: Annual actual evapotranspiration
(a) and annual accumulated biomass (b)
across the DL1 region from June 1999 to May 2000.
During this one-year cycle study, the accumulated biomass production calculated ranges from 31255 kg/ha/year to 19741 kg/ha/year for the vegetation growth zones within DL1. The average biomass production for the dry monsoon forest is 27,820 kg/ha/year. The actual evapotranspiration values vary between 1,123mm/year to 1,472 mm/year. In the paddy cultivation areas, the average biomass development is 8,501 kg/ha/season and 7,546 kg/ha/season for the two cropping seasons of Maha (Oct. – Feb. major cultivation season) and Yala (Mar. – Aug.minor cultivation season), respectively (Fig. 2).
Figure 3 shows the relationships between vegetation growth and P75 for all vegetation growth zones as delineated using vegetation growth in DL1 zone. Graphs (a), (b), (c) represent the relationships in Dec., Aug. and Jan respectively. Graph (d) based in the annual averages of vegetation growth and P75. No clear relatio relationship can be identified between vegetation growth and P75. In all four graphs overall relation ships between these two variables are negative. But the correlation coefficients are very low. One to one relationship would imply a complete dependency of water from rainfall to create vegetation growth. But it is well known that soil moisture and ground water are playing a very important buffer role in vegetation growth. The non-linearity of relationships in graphs (a), (b), (c) and (d) show exactly that process and its geographical dependency. This implies that there is no proper relationship between vegetation growth and P75 in different vegetation growth zones inside DL1 agro-ecological zone. In existing agro-ecological zone approach for Sri Lanka, P75 is the most predominantly spatially and temporally varying parameter in DL1 region. The new approach is based on spatial and temporal variations of vegetation growth. Therefore, the vegetation growth zones derived, estimated using NOAA-AVHRR data, identified different sets of zones representing the combined effect of all the bio-physical parameters which are different from existing agro-ecological zones. Further, the new approach identified relatively small areas with respect to the existing agro-ecological zones enabling assessment of local activities such as crop selection and agronomic practice.
Figure 3: Relationship between P75 and vegetation growth in
(a) January, (b) August, (c) December and (d) Annual
This study based on information derived by NOAA-AVHRR images provides an example of using free public satellite data for agricultural planning. The major innovation of this study is that the actual biological activities on the ground are considered as major factors in demarcating land areas into different zones. Even though the spatial resolution of NOAA-AVHRR is low, it is useful for macro level analysis such as demarcating a country into vegetation growth zones. In conventional zoning methods, only rainfall, soils and elevation data have been used. But the actual ground situation is a result of the combined effect of such parameters, subject to spatial and temporal variations. Therefore, vegetation growth is an appropriate indicator in monitoring the combined effect of these parameters. As mentioned in introduction, a lack of data on water utilization among different uses and users was identified as a major problem in water management practices. The case study of the DL1 zone shows the usefulness of such techniques in extracting information for different parts of the domain. In agricultural practices, demarcation of areas into different agro-ecological zones is vitally necessary. This technique provides a zoning method that is more accurate than the existing ones. It also furnishes the data to establishes relationships between water use and biomass production. Based on these relationships and other bio-physical information, best agricultural practices for different areas can be determined.
- Bastiaanssen, W.G.M; M. Menenti; R.A. Feddes; and,A.A.M. Holtslag. 1998. A remote sensing surface energy balance algorithm for land (SEBAL). Formulation 1. Journal of.Hydrology, 212-213,198-212.
- Bastiaanssen, W.G.M; D.J. Molden; and I.W. Makin.2000. Remote sensing for irrigated agriculture : example from research and possible applications. Agricultural Water Management, 46:137-155.
- Chandrapala, L.; and M. Wimalasuriya., 2002, Satellite measurement supplemented with meteorological data to operationally estimate actual evapotranspiration over Sri Lanka. Agricultural Water Management (in press).
- Hargreaves, G.H. 1975. Moisture availability and crop production. Transaction of the American society of Agricultural engineering, vol.18.
- Monteith, J.L. 1972, Solar radiation and productivity in tropical ecosystems. Journal of.Applied Ecology,9:747-766.
- Muthuwatta, L. ; and Y.Chemin. 2002. Vegetation growth zonation of Sri Lanka interpreted from satellite data. Agricultural Water Management (in press
- Panabokke, C.R. 1996. Soils and agro–ecological environment of Sri Lanka. Natural resources, energy and science authority of Sri Lanka.
- Samarasinghe, G. B. W. Bastiaanssen. 2002. The annual crop growth cycle of Sri Lanka and some yield estimates. Agricultural Water Management (in press).