System Performance Evaluation
and Diagnostic Analysis of Canal Irrigation Projects S. T. Chari, S. Jonna, P.
V. Raju, C. S. Murthy and K. A. Hakeem AbstractWater Resources Group, National Remote Sensing Agency Dept of Space, Govt. of India Hyderabad - 500 03 Satellite remote sensing techniques can efectively complement and supplement ground data collection towards objective and reliable evaluation of irigation system performance and diagnosti analysis. Multidate satellite data during the rabi seasons prior to and after implementation of National Water Management Project NWMP have been analysed to provide spatial information on irrigated area and paddy productivity right up to distributary command level add through the years since 1986-87 rabi season. The Bhadra project performance is seen to have significatntly improved after NWMP in terms of increased irrigated area, improved paddy yield, decreased depth of water applied and improved water use efficiency with respect to paddy production. Diagnostic analysis has indicated problem distributaries of gap in irrigation utilisation, less nad stagnant paddy yield and poor water use efficiency. Equity in water application has also been diagnosed. Distributaries with large gap in paddy yield between head reach ad Satellite remote sensing applications are thus seen to be effective tools for irrigation water management. Introduction The Bhadra rrigation Project in Karnataka State is one of the subprojects of National Water Management Project NWMP of India. The NWMP objective is to increase the agricultural productivity and farms income inexisting irrigation schemes by providing predictable, reliable and equitable irrigation service. The productivity would be measured in regard to agricultural yield and area irigated. The project incorporates monitoring and evaluation at various levels to provide information on system performance evaluation and diagnostice analysis, necessary for corrective measures. Satellite remote sensing techniques have been employed to effectively complement and supplement the ground mechanism in this project. System performance indicators are depth of water applied irrigation intensity, major crop yield and water use efficiency. Equity of irrigation service has also been evaluated. Diagnostic analysis is carried out using temporl and spatial information at distributor level, Problem distributaries have been identified in regard to irrigation gap, low crop yield, low/no yield improvement through years and low water use efficiency. While many earlier studies have demonstrated the utility of satellite data in providing crop area statistics and crop yield assessment (Groten 1993, Rasmussen 1992) Tennakoon et all 1992, Thiruvengadachari 1981, Teng 1990 ) Over large administrative units, the investigation in Bhadra Project perhaps is the first attempt to provide informaton on spatal and tempra varability of performance indcatiors at distributary comand leve (Thiruvengadachari et a 1994.) Basic Data and Methodology Multidate Indian Remote Sensing Satellite (IRS) LISS I data during the rabi seasons since 1986-87 have been analysed (LANDSAT MSS/TM data to a limited extent has been used to fill tbe gaps in temporation coverage/non availability of IRS data) Kharif season could not be analysea in view of continuous cloud cover preciuaing satelite vrew of the ground. Liss I sensor data was used since even at the spatial resolution of 72.5 meters this data could provide distributary level information on major crop groups and is also cost/time effective. The satellite data has been digitally classified into paddy and non paddy crop areas. Though the latter has been further clasificed into individual crop types such as groundnut, sugarcane and garden crops, these were later lumped into one category, in view of reduced classification accuracy resulting from non contiguous smaller patchiness of such crops. Different classification schemes, ranging from single date to multidate data analysis and use of vegetation index along with raw bands, were attempted to provide the best delineation of irrigated crop area. A base map of the command area, defining the distributary command boundaries, is prepared from survey of India topographic map in 1:50000 scale and revenue survey map in 1:7920 scale, taking the aid of topography, drainage network and other features. The base map is digitised and overlaid on the classified crop map, to enable crop statistices extraction under each distributary. The crop condition is indicated by the normalised difference vegetation index (NDVI) which is an arithmetic combination of reflected radiation in the visible and near-infrared wavelengths of IRS/LANDSAT sensor. The satellite data analysis is supported by field visits and ground data. The paddy yield model has been developed based on crop cutting experiments conducted during 1992-93 rabi season in the command area and NDVI of paddy. Though irrigated crop area and yield as well as depth of water and water use efficiaency have been estimated at distrubutary level, only summary statistics are provided in subsequent sections. System Performance Evaluation Irrigated Crop Area Irrigation is being provided to a significantly larger area after NWMP implementation (Table 1) . The irrigation intensity in the current rabi season is 87 percent compared to 76 percent in the pre NWMP 1986-87 rabi season. Irrigation utilisation is highest in head end Bhadravathy division followed by Malebennur and Davangere deivisions. The tail end Davangere division has registered an increase of 9 percent in irrigation intensity. Aalysis of irrigation intensity at canal subdivision and at distributary level idicates that the irrigation intensity is high in No. 3 subdivision of Bhadravathi division. The intensity is decreasing towards the tail end of Malebennur division, following similar trends before and after NWMP. The trend of decreasing irrigation intensity is seen in Dvangere division also. In general, while irrigation intensity has increased in recent rabi seasons, the spatil, while irigation intensity has increased in recent rabi seasons, the spatial variability is still following the pre NWMP trend. This non uniformity in irrigation intensity needs to be narrowed and equity established. Redistribution of water from distributaries with larger than 100 percent irrigation intensity can lead to greater equity. The change in cropping pattern before and after NWMP is given in Table 2. The percent paddy area in the command area has increased from 56 percent in 1986-87 rabi and stabilised at 69 prcent in the recent rabi seasons. The percentage area under paddy crop is 49 percent in Bhadravathy division and around 76 percent in Malebennur division. Percent paddy area in Davangere division however is showing a significant increase in the recent two rabi seasons, perhaps indicative of improved water availability in this tail end division. Significant acroce is also feen in the tail end most No. 1 sub-division within vangere division. The spatial pattern of percent paddy area in cent rabi seasons is remarkably similar to the variability in 1986-87 rabi season, indicating alround farmer preference for paddy crop with improved irrigation service. Table 1. Irrigted Crop area (ha) Through the years
*As per ground data Note Figure within bracket represents irrigation intensity TABLE 2 COMPARISON OF PERCENT PADDY AREA DURING RABI SEASONS
*As per ground data Paddy Productivity While conventional crop cuttin experiments have been conducted for the major crop of Paddy since 1991-92 the sampling scheme is often inadequate and not representative of the yield variability in the command area. Further the conventional scheme provides yield estimates only at the project level, and thus spatial variability between distrubutaries is not available nor such estimates are available for pre-NWMP years. A bold approach of relating satellite data derived NDVI to paddy yield has been attempted. While earlier studies have demonstrated this relationship (William 1990 , Groten 1993) this is perhaps the first attempt at extending the NDVI yield model through years (after satellite data normalisation) and across distrubutaries in the command area. Yield (kg/ha)= 42.23 TCVI-3439.05 Where TCVI is the time composited NDVI for any paddy pixel. TCVI is the maximum NDVI value for each pixel from the coregistered NDVI images of the three compeass dates. The yield model has been developed using crop cutting experiments in 1992-93 rabi season at 70 plots, and has been validated through farmer surveys and CCE results of 1993-94 rabi season. Spatial yeeld estimates over canal subdivisions and distributary commands have been made usin the model. Temporal variability in paddy yield is obtained through application of radiometrically normalised satellite data in different years. The satellite estimate of paddy yield in 1993-94 rabi season is given in Table 3. The average yield is estimated to be 48.71 Qtls/ha,with highest yield coming from Malebenur division (53.4Qtls/ha) followed by Davangere (46.4 Qtls/ha) and Bhadravathi division (42.9 Qtls/ha). The yield in Bhadravathi division is 12 percent lower than the weighted average yeild over the command area. In Malbennur division, the paddy yield in No. 2 canal subdivision is 16 percent less than the divisional average yield. In Bhadravathi and Davangere divisions the subdivision yields are more uniform (within 10 percent of divisional average yield). There is a general decrease in paddy yield in te tail end under Bhadravathi main canal. The paddy yields are more uniform under Davangere branch canal. The paddy yields under LBC and RBC of D.B. Kere pickup are 45 Qtls/ ha compared to the Malebennur division average of 53 Qtls/ha. Table 3 Paddy Yield (Qtls/Ha) Through The Years
Water Managment Analysis of satellite data has also helped in evaluating improvement under NWMP in water management aspects such as depth of water applied and sater use eficiency for paddy production. The depth of water application has improved to 0.859 meters in 1993-94 rabi season over the area commanded by Bhadra Right Bank Canal (RBC) compared to 1.059 meters in 1986-87 rabi rabi season (Table-4) . The depth of water applied on paddy area is 1.018 meter in the 1993-94 rabi season against 1.343 meter prior to NWMP. Similarly water use efficiency for paddy crop improved to 0.478 kg/cu.m in 1993-94 rabi season against 1.343 meter prior to NWMP. Similarly water use efficiency for paddy crop improved to 0.478 rabi season. TABLE 4. IRRIGATION WATER MANAGEMENT IN BHADRA COMMAND
Equity in Water Distribution The equity in water distribution during 1993-94 rabi season is estimated by the Christiansen's Uniformity Coefficient (UCC) defined as UCC= 1-S(|xi-x|)ai/(Sai) x Where x/x i= depth of irrigation water delivered to incremental area ai d\weighted mean depth of irrigation water delivered The UCC value closer to 1 indicates better uniformity in water distribution. Water supply among subdivisions during 1993-94 rabi seasn is more quitable in Malebennur and Davangere divisions compared to badravathi division (Table 5) . The equity in Water supply among the distributaries within a subdivision is given. Subdivisions with low uniformity are given in bold. There is a need to improve and ensure equitable canal diliverty jpropotionate to the area, starting from these distributaries. Table 5 Equity In Water Suply During 1993-94 Rabi Season
*UCC = 1 indicates 100% uniformity in water application Diagnostic Analysis Spatial and temporal variability in irrigated area, and crop productivity derived from satellite data provides the capability to diagnose problem distributaries, with low irrigation intensity (Table 6) and low water use efficiency for paddy production (Table 7). Table 6 Distributaries With Low Irrigation Intensity In Rabi 1993-94
Table 7. Distributaries with low water use efficiency (less than 90% of branch canal average)
Equity within distributary command has been evaluated in reardd to gap in yield between head reach area (HRA) and tail reach area (TRA). Problem distributaries with significant inequity have been identified (Table 8) in Malebennur and Davngere canal divisions. TABLE 8 DISTRIBUTARIES WITH NON-UNIFORMITY IN PADDY YIELD
Such diagnostic analysis and identification of problem pockets wihtin the command area, made possible by satellite data, should lead to corrective management in the subsequent season. Conclusions Multidate satellite data during the rabi season of pre and post-NWMP years have indicated significant increase in irrigation intensity and also in area under paddy crop. since canal discharge is the limiting factor in water delivery rather than reservoir storage, water management is seen to have substantially improved. This is all the more notewoethy since the yield of major rabi crop, paddy, has significantly increased. The depth of water application has decreased under NWMP and water use efficiency for paddy production has improved. Thus there is significant improvement in system performance after NWMP implementation. Diagnostic analysis of irrigated area and paddy productivity at distributary level, has identified problem distributaries of low irrigation intensity and paddy yield and low/no yield improvement in recent years. Distributaries with large yield gaps between head reach area and tail reach area have been targeted. Canal sub divisions/divisions with low equity in water distribution and distributaries with low water use efficiency for paddy production have been identified. corrective management in subsequent seasons will help in overall and equitable improvement of Bhadra Project. Acknowledgments The authors gratefully acknowledge the guidance and support received from Director and Associate Director of National Remote Snesing Agency. Thanks are also due to Mr. M. L. Lath, Commissioner [Water Management], Ministry of Water Resources, Govt. of India, and Mr. A. A. Pai, then Irrigation Engineer and Mr. Bandyopadyay, Irrigation Engineer, World Bank, New Delhi for their unstinted support and constructive discussions which helped to shape the methodology. Support received from Govt. of Karnataka towards project funding and cooperation during field visits/ data collection is gratefully acknowledged. Various colleagues in Water Resources Group and other functional areas of NRSA have significantly contributed to the success of this study. Thanks are due to Mr. Sanjay who extended support in many ways right through the project progress and in the preparation of manuscript. Secretarial support received from Ms. Sudharani and Mr. T. Subba Rao is acknowledged. References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||