This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The objective of this study was to investigate the changes in rice areas in the most recent decade (2000-2009) using remote sensing technique to understand its relation with rice production and rice prices. The study was conducted in Nepal with an area of 14,718,100 hectares in South Asia. The MODIS (MOD09Q1) 500m time-series data and the spectral matching techniques (SMTs) were found ideal for agricultural cropland change detection over large area and provided fuzzy classification accuracies between 63.3-100% for various rice classes and accuracy is 79.1 % field plot data. The MOD09Q1 derived rice areas for the districts were highly correlated with the national statistics data with R2-values is 0.9832. Dynamic mapping rice area for the most recent decade (2000-2009) has been done by identifying 8 main classes of rice (irrigated, rainfed, upland etc.). There is no significant change in the rice area and proportionate area under different rice classes except in the year 2006. During the year 2006, rice area in Nepal declined by 14% from 2005 and 2007 figures. Area under rainfed rice class continues to be the the predominant one from 2000 to 2009.
A multi-level analysis was carried out by combining remote sensing techniques with socioeconomic information and national statistics. The analysis highlights the dynamics of rice areas, production and increasing rice prices in Nepal over the years (2000-2008). The findings infer that future rice research program in Nepal needs to increasingly focus on proper rice production planning and targeting of technologies along the 8 main rice classes (ecosystems and production systems) identified by this study.
Key words: Rice maps, rice production, land use change, rice classes, rice price, and yield.
Nepal is an agrarian country where agriculture is still the single largest sector in the economy, accounting for 32 percent of GDP (MoF, 2008). The country's main agricultural production includes paddy rice, maize, wheat, sugar cane, vegetables, potatoes, pulses and tea, milk, meat and other livestock products. The country also produces cash crops like tea, ginger, citrus, non-timber forest products in the Hill region both for domestic consumption and exports. Cropping pattern is dominated by three crops, rice, maize and wheat accounting for over 90 % of the area and food grain production. Growth in agriculture has direct impact on the national economy and livelihood of the poor as more than two-thirds of the labour force is currently employed in agriculture.
Rice is the most important food crop in the country in terms of area, production and livelihood of the people. It is currently grown in half of the total cropped area and contributes more than half of the total food grain production in the country (MoAC, 2008). It also supplies about 40% of the food calorie intake for the people of Nepal. The crop is grown extensively under a wide range of agroecological conditions from lowland in Terai (50 msl) to high mountain valley and mountain slopes (2830 msl) in Jumla- the highest altitude of rice growing location in the world. The crop is cultivated in all agro-ecological regions (Mountains, Hills and Terai) covering mountain slopes, hill terraces, intermountain basins, river valleys, and flat lowland plains bordering to India (Gauchan et al, 2008). About two-third of (74 %) of paddy is produced in the flat lowland of Terai and the rest (26%) in the hills and mountains (Gauchan and Pandey, 2010). Rice is mainly cultivated during wet monsoon season (June-November) in major part of the country. However, in some parts of the lowland plains and valley floors, they are also grown during spring season (Chaite rice) as an irrigated crop. Transplanting is the major method of rice establishment in both the irrigated and rainfed lowland conditions. Direct seeding (broadcasting) is mainly practiced for the upland rice (Ghaiya) in upland fields.
Rice has special significance and economic importance in agricultural development and poverty reduction in Nepal. However, we lack adequate understanding and information about recent changes in rice area over the years to design appropriate production planning and technology targeting in the country. Therefore, this study aims to analyze changes in rice area in the most recent decade (2000-2009) using remote sensing technique to understand its relation with rice production and prices.
Studies reporting advantages of MODIS satellite imagery to map agriculture change response to water availability in the large areas between water-surplus year water-deficit years and to understand the change dynamics of irrigated areas due to fluctuating water availability (Gumma et al. 2009, Gaur et al. 2008). Satellite imagery can provide detailed maps of where cropping patterns change significantly in response to water availability (Thiruvengadachari et al. 1997). Satellite imagery has been increasingly used to quantify water use and productivity in irrigation systems (Bastiaanssen and Bos 1999; Thiruvengadachari and Sakthivadivel 1997), but has less frequently been used to identify how rainfed rice areas change in response to variations in climate change.
Given the above background, the main objective of this research was to study the rice growing areas change response to over last decade and to understand the change dynamics of rice growing areas due to climate change. A multi-level analysis (farm level surveys and regional rice areas and production assessment through remote sensing techniques and national statistics) was carried out to study the dynamics of rice areas, production and rice prices in the study area over the most recent decade (2000-2009).
The paper is organized as follows. Following introduction, it provides sources of data and discusses about methods for analyzing changes in rice area using remote sensing techniques. A general overview of rice production planning and targeting is presented combining remote sensing technique with socioeconomic data and rice production and growth rate analysis. Finally, the paper presents about results, discussion and conclusion.
2.0 Study area:
Nepal is a landlocked Himalayan country located in South Asia, ranging from 800 E to 920 E and 260 N to 300 N (Figure 1). It is bordered to north by China (Tibet) and to south, east and west by India with total geographical area of 147,181 km and population of 2.7 million. For a small territory, the Nepali landscape is uncommonly diverse, ranging from the humid Terai in the south to the lofty Himalayas in the north. Nepal boasts eight of the world's top ten highest mountains, including Mount Everest on the border with China. Geographically, the country is divided into three ecological regions: Mountain, Hill and Terai accommodating 7.3%, 44.3% and 48.4 % of the population (CBS, 2001). Administratively, the country is divided into14 zones and 75 districts, grouped into 5 development regions which are eastern, central, western, mid-western and far-western regions (Fig 1). The country has three major rivers viz; Koshi, Gandaki and Karnali that are originated in the northern mountainous part directly from Himalayas and flow to the south. In addition, Mahakali river also flows from the far-western part of Nepal to the south and get into the India.These rivers are the major source of Ganges in India.
Figure1: Study area: Map of Nepal, showing development regions with major rivers across Nepal. River network delineated from SRTM DEM.
Insert Figure 1 here
3.1 MODIS surface reflectance data
The MODIS 8-day composite surface reflectance product from the Terra platform (MOD09A1) is ideal for monitoring vegetation at a continental scale (Thenkabail et al. 2005). The seven bands of reflectance data (Table 1) at a resolution 500m, coupled with a high-repeat frequency, can capture the seasonal variations in vegetation vigour, soil and vegetation moisture, and surface water that characterize key stages of rice cultivation. The reflectance data undergo several pre-processing steps, including algorithms for atmospheric correction. Furthermore, the rate of observation coverage, the viewing angle, cloud or cloud shadow coverage, and aerosol loading are all assessed on a pixel-by-pixel basis to ensure that each pixel contains the best observation during that 8-day period. MOD09A1 also includes two quality assessment datasets at the pixel and band level, which are vital for user post-processing to identify and remove areas of persistent cloud and snow cover. The MOD09A1 (Version 005) data are available in a tile system, where each tile covers 10 degrees by 10 degrees (1111.2 x 1111.2 km at the equator, again often rounded up to 1200 x 1200 km in the literature). We downloaded 12 tiles, for every 8 days, from http://modis-land.gsfc.nasa.gov covering South Asia for all dates start from June 2000 to May 2010, which includes the rice crops in two seasons (kharif and rabi) for each year. Analysis has done for whole South Asia, presents study focus on Nepal.
Table 1: MODIS data sets (7 bands): MODIS Terra 7-band reflectance data characteristics used in this study.
Band width (nmÂ³)
Band center (nmÂ³)
Absolute Land Cover Transformation, Vegetation Chlorophyll
Cloud Amount, Vegetation Land Cover Transformation
Cloud Properties, Land Properties
Note: 1 = of the 36 MODIS bands, the 7 bands reported here are specially processed for Land studies.
2 = MODIS bands are re-arranged to follow the electromagnetic spectrum (e.g., blue band 3 followed by green band 4).
3 = nanometers.
4 = taken from MODIS web site (http://modis-land.gsfc.nasa.gov/)
Insert Table 1 here
Two field-level surveys were conducted during October 11-26, 2003 and August 30- September 28, 2005 across 1,004 locations covering the major cropland areas in South Aisa (Gumma et al, 2010). The locations were chosen based on the knowledge of local agricultural extension officers to ensure that the same crops were grown during the 2000-01 as were observed during the survey. The local experts also provided information on crop calendars, cropping intensity (single or double crop), and percentage canopy cover for these locations from their recorded data for the years 2000-01. Overall, 1,004 spatially well-distributed data points were collected. Of this, 75% of the points (743) were used for call identification and labeling and the rest of the points (261) were used for accuracy assessment.
3.3 Secondary data sets
3.3.1 Socio economic data
Socioeconomic data from various sources in Nepal and international agencies were also used. National level rice price data was obtained from published government sources (Economic survey 2008-09) and rice production and yield data were obtained from Statistical Information on Nepalese Agriculture from Ministry of Agriculture and Cooperatives (MoAC, 2009) Government of Nepal. The rice price in US $ was for rice was derived from FAOSTAT database 2010.
3.3.2 SRTM 90 m elevation
The Shuttle Radar Topography Mission (SRTM) obtained elevation data on a near-global scale to generate the most complete high-resolution digital topographic database of Earth (Farr and Kobrick, 2000; Rabus et al., 2003; Rodriguez et al., 2005; Farr et al., 2007). Since the topography of the river basin under investigation is highly diverse, the SRTM elevation data set is useful in separating irrigated areas within the command areas and deltas with low elevations and high elevated areas with forest vegetation. The SRTM data (90 m resampled to 30 m) were also used to perform image segmentation based on elevation values in the basin.
3.3.3 National statistics for rice areas:
Rice area statistics were obtained at the sub-national level (development regions, zones and districts), and represent the total cropland area sown to rice. The data adopted by the Statistical Information on Nepalese Agriculture from Ministry of Agriculture and Cooperatives, Nepal (MoAC, 2009) and the Central Bureau of Statistics, Government of Nepal (CBS, 2008). The information was also supplemented by Economic Survey of Nepal (MoF, 2009).
4.1 Remote Sensing methodology for historical rice maps
MODIS (MOD09Q1) data were used to map rice areas from 2000 to 2009 according to the methodology adopted from Gumma et al, (2010).
The basic process begins with downloading eight-day composites of MOD09QNDVI with 500m resolution were stacked into a single data set for 2000-01 (43 cloud free NDVI images), 2001-02 (43 cloud free NDVI images), 2002-2003 (42 cloud free NDVI images), 2003-04 (36 NDVI images), 2004-05 (42 cloud free NDVI images) and 2005-2006 (45 cloud free NDVI images), 2006-07 (42 cloud free NDVI images), 2007-08 (41 cloud free NDVI images), 2008-09 (44 cloud free NDVI images) and 2009-10 (37 cloud free NDVI images.
The MODIS mega-file was divided into three distinct zones, one was major irrigation command areas zone using India's central board of irrigation and power (CBIP) command area. The idea behind the segmentation process is to focus more on the segments having higher amount of informal and fragmented irrigated classes such as the CBIP command areas. Such segments would be classified in to more numbers of classes then the other for better delineation of different irrigated classes (irrigated rice single crop, irrigated rice double crop, irrigated-conjunctive use and etc) using protocols explained in Gumma et al., 2010. Second was SRTM- derived slope zones, Such segmentation allows for easier class spectrum separation and identification of minor irrigated areas (informal irrigated areas) other than command areas and deltas with low elevations and high elevated areas with forest vegetation.
Each dataset is then classified using unsupervised ISOCLASS cluster K-means classification. Unsupervised classification was used instead of supervised classification in order to capture the range of variability in phenology over the image. At a regional scale and where the NDVI signatures of all potential classes are not known, unsupervised classification captures the range of phenological variability. The number of classes varied from 40 to 100 based on area covered by the segment and complexity of area.
Class identification and labeling is based on bi-spectral plots, NDVI time-series plots, ground truth data and very high resolution images (Google Earth). Grouping class spectra based on class similarities and/or by comparing them with ideal/target spectra, rigorous protocols for class identification and labeling that included use of large volumes of groundtruth data and the use of very high resolution imagery. Resolving mixed classes through specifying GIS spatial analysis/Modeling layers (DEM/Rainfall), and establishing innovative methods for irrigated area calculations and accuracy assessments. The classes generated from the unsupervised classification were aggregated into 11 classes and named based on spectral similarity and intensive field plot information. Spectral matching (Thenkabail et al, 2007) was used to relate the classes for all years. These processes are described in Thenkabail et al, 2007 and Gumma et al, 2008 briefly.
Once the rice areas are mapped, calculate precise areas, however, in coarser resolution MODIS were 500m on a side, which was larger than many agricultural fields in the study area. Many pixels overlapped several land cover types, so any given land cover class had a corresponding value for the areal fraction that was rice crop (Gumma et al., 2010). The rice fractions were determined using intensive field plot information.
Figure 2: Overview of the methodology for rice production planning and targeting.
Insert Figure 2 here
4.2 Accuracy assessment base on field-plot data
A fuzzy accuracy assessment was performed using 25% (261 data points) of field-plot data to derive robust understanding of the accuracies of the datasets used in this study. The field-plot data were based on an extensive field campaign conducted throughout India during kharif season by the International Water Management Institute researchers and consisted of 1,004 points. Fuzzy accuracy assessment provides realistic class accuracies where land cover is heterogeneous and pixel sizes exceed the size of uniform land cover units (see Gopal and Woodcock 1994; Thenkabail et al. 2005, 2009; Gumma et al. 2009). For this study, we had assigned 3 x 3 cells of MODIS pixels around each of the field-plot points to one of six categories: absolutely correct (100% correct), largely correct (75% or more correct), correct (50% or more correct), incorrect (50% or more incorrect), mostly incorrect (75% or more incorrect), and absolutely incorrect (100% incorrect). Class areas were tabulated for a 3 x 3-pixel (nine-pixel) window around each field-plot point.
Based on the theoretical description given by Congalton and Green (1999), equations 1, 2, and 3 were used to estimate accuracies and errors. Field-plot information was used to determine robust accuracies, using equations 1, 2, and 3.
where RFPCIA = rice field-plots classified as rice areas (number),Â TRFP = total rice field-plots (number), NRFPRA = non-rice field-plot points classified as rice area (number), TNRFP = total non-rice field-plots (number)Â , and RFPNRA = rice field-plots classified as non-rice areas (number).
4.3 Accuracy assessment based on correlations between census and satellite data
The final classified map of rice areas was compared against rice area statistics for Nepal country at the highest available spatial resolution. The MODIS rice area fractions were aggregated to get a summed rice area per district or state and these were compared against the reported planted areas for the kharif season in 2000-01.
4.4 Growth rate analysis for area, production and yield
The growth rate was computed for each period by the least-squares regression technique in STATA statistical package by transferring area, production and yield values into natural logarithm.The least-squares growth rate, r, is estimated by fitting a a least-squares trend regression to the logarithmic annual values of the variable in the relevant period (WDR, 2008). The regression equation takes the form log Xt = a + bt + e and and represents a logarithmic transformation of the compound growth rate equation: X t. In these equations, X is the variable of interest, t is the time (year), and a = log X t = X0 (1+r) 0 and b = log (1 + r) are the parameters to be estimated; e is the error term. If b* is the least-squares estimate of b, then the average annual percentage growth rate, r, is obtained as [antilog (b*)] - 1 and is multiplied by 100 to express it as a percentage.
4.5 Rice price, area and yield comparison
National average rice price and rice yield obtained from national official sources and FAOSTAT was presented graphically to understand their trend, variability over the years and relationship between price and yield. Rice area obtained from remote sensing was also compared with rice price for the last one decade to observe pattern of rice area and price variation and trend over the years.
In this section, we focus on the resulting rice classification over years, the classification accuracy assessment based on field-plot data, comparison between the MODIS rice areas estimates, sub national statistics, growth rate in rice area with yield and production, rice area and production by developmental regions, trend in rice area and prices and Relationship between paddy yield and price.
5.1 Rice map and area statistics
Altogether, 11 classes were identified and labeled (Figure 3). Almost 1.7 million hectares of arable land were labeled as containing some degree of rice cultivation based on FPAs. However, when RAFs (last column in Table 4) were used, the actual (sub-pixel) areas were 1.6 million hectares (last column in Table 3) for the year 2000-01. The final class name or label (Figure 3, Table 3) is based on the predominance of a particular rice class (e.g., single- or double-season rice), and the dominant water source. For example, the name for class 1 is "03. Irrigated 100 percent - Rice/other". This rice class is dominated by rice cultivation in the wet and summer seasons and the area is predominantly irrigated from surface water source. This class occurs scattered in the different regions both in the lower Hills and Terai. The largest area under this class are found in central and western Hills, Terai and inner Terai valleys. Similarly, class 9 is labeled "09. Rainfed 100 percent - Rice" since this is an intensely cropped rice class, but heavily dependent on seasonal rains. This class is predominantly scattered across east to far-wetern region in heavy rainfall areas including in higher elevation such as hilly terrace. Similarly, class 2 is "02. Irrigated 100 percent-rice-rice or rice-other crop", predominantly in the Terai region from east to far-western region around Indo-Gangetic s basins (along India boarder). The spectral separability in the temporal NDVI signatures for each of the rice classes is shown in (Gumma et al. (prep)).
The classification accuracies from the 261 field-plot observation data points in the validation dataset are summarized in Table 3. The fuzzy classification accuracy varied from 63.3% to and 100% across 12 classes with an overall accuracy of 79.1%î ºnearly four out of five rice pixels have been correctly classified as rice. It has to be noted that the uncertainty of about 20% is due to the inter-mix among the various rice classes. So, that the rice versus no-rice class accuracy will be very high. The irrigated classes, generally, have higher classification accuracies than the rainfed or mixed irrigated/rainfed classes of various classes (Table 3). Of the 261 points, 61% of the points were classified as absolutely correct.
Insert Figure 3 here
Insert Table 2 here
Insert Table 3 here
Figure 3: Rice classification with source wise across the Nepal over the years (start from 2000-01 to 2009-10).
Table 2: Rice areas region wise for Nepal
Rice area (ha)
Total rice area
Table 3: Fuzzy accuracy assessment from field-plot data. Values in the table indicate the % of field-plot windows in each class with a given correctness %age.
Fuzzy classification accuracy
Rice class number and class name
(100 percent correct)
(75 percent and above correct)
(51 percent and above correct)
(51 percent and above incorrect)
(75 percent and above incorrect)
(100 percent incorrect)
01. Irrigated 100 percent - Rice/Rice
02. Irrigated 100 percent - Rice/Rice or Rice/Other
03. Irrigated 100 percent - Rice
04. Irrigated 60 percent / Rainfed 40 percent - Rice/Rice
05. Irrigated 30 percent / Rainfed 70 percent - Rice/Rice or Rice/Other
06. Upland 80 percent / Rainfed 10 percent / Irrigated 10 percent - Rice
07. Rainfed 60 percent / Irrigated 40 percent - Rice/Rice
08. Rainfed 90 percent / Irrigated 10 percent - Rice
09. Rainfed 100 percent - Rice
10. Deepwater 100 percent - Rice/Rice
11. Deepwater 100 percent - Water/Rice
12. Wetlands 100 percent - Rice/Rice
5.2 Accuracy assessment based on comparison with sub-national statistics
Figure 4 compares the summed rice areas across all classes against the published rice statistics. We performed the comparison at the district level (75 spatial units) across study area, but, for reasons of space, we only report the tabulated areas at the development regions level (5 spatial units) table 2 and the district level comparison is shown in the figure 4. Figure 4 shows the relationship between MODIS area summarized at district against the subnational rice areas at the same level of spatial detail. The level of agreement between the MODIS area estimates and the published statistics is very goodî º 98.3% at the district level (Figure 4).
Figure 4: Accuracy assessment and validation. The district-wise rice areas derived using MODIS 500m compared with agricultural census data.
Insert Figure 4 here
5.3 Growth rate in rice area, yield and production
The analysis of annual growth rate for rice area, yield and production for the last one decade (2000-2008) is presented in Table 4. Overall analysis of the growth rate for the last 9 years indicated that there was no statistically significant yield and production growth in the country. The area growth was negative but non-significant. However, when analysis was made by dividing the total period into three sub-periods for each of three years, it showed variation over the different sub periods. Yield and production growth was negative in the first three year period (2000-20002). Area growth was also negative but statistically non-significant. In the second three year period (2003-2005), yield growth was negative and statistically significant. In the last and most recent three year period (2006-2008), per annum yield growth was high, positive and statistically significant but area and production growth was not significant despite their positive growth. Overall, in the last decade, the growth in rice production and yield is stagnated and not growing in a rate of population growth (2.2% per year).
Table 4: Growth rates (% per year) of rice area, yield and production in Nepal
***, **, * indicate statistical significance at 1%, 5% and 10% probability level respectively
Insert Table 4 here
5.4 Rice area and production by Developmental Regions
Rice is the main source of livelihood in all the five development regions of Nepal. The largest area and production of rice takes place in eastern region followed by central and western region (Table 2). Mid-western and far-western regions have relatively smaller area and production. Western region is intermediate among them accounting for 20% of the area and production shares. Eastern and central region each account for 30% of the area and production shares in the country, whilst mid-western and far-western regions account for less than 10 % of area and production shares. Yield levels are also higher in eastern and centeral regions and lowest in far-western region.
Table 5: Average of rice area, yield, production and percent shares by regions (2000-2008)
*Source :Area from remote sensing estimates and production and yield from MoAC(2008)
Insert Table 5 here
Rice area and production are lowest in Mid-Westerna nd Far-Western development region. Yield levels are also lower than national average in these region.A recent national living standard survey of Nepal (NLSS, 2004) also indicate that poverty rate is highest in these Mid and Far-Western regions. There is positive relation between poverty incidence and rice area and production level indicating a need of targeting rice technologies development and promotion in these poverty-stricken regions of Nepal..
5.5 Trend in rice area and prices
The relationship between trend in rice area and paddy price (US $ ton/ha) is presented in Fig 5. The rice area remained almost constant over the years. However paddy price (rough rice) is showing increasing trend over the recent years particularly after 2001.This indicates the low supply and increasing demand for rice in the recent years due to stagnant production and increasing population pressure and urbanization. Increase price of rice has negative impact on the food security, nutrition and the livelihood of the low income groups residing both in rural and urban areas. Land less laborers, marginal farmers, women and those who have limited income opportunities are becoming most vulnerable from increasing paddy prices as they spend large share of their income on rice food alone. Poorer people tend to reduce intake of food as a result of increase of food prices, particularly coarse rice which is a major staple of the poorer group (WFP, 2008).
Figure 5: Trend in rice area and paddy price in Nepal
Insert Figure 5 here
5.6 Relationship between paddy yield and price
Paddy price is steadily increasing in the recent years whilst the yield of rice remained variable over the years (Fig 6). A high variability in rice yield over recent years is caused by variability of the rainfall pattern. The yield level (2.5 t /ha) was lowest in 2006 due to drought conditions prevailing particularly in major rice producing regions of eastern and central Terai (FAO, 2007). In recent years (in 2008-09) the yield has regained reaching historic high at 2.9 t/ha for Nepal. However, paddy price is steadily increasing over the recent years indicating a need for increasing production and productivity of the crop by proper planning and targeting of the technologies..
Figure 6: Trend in paddy price and yield in Nepal
Insert Figure 6 here
6.0 Discussion and Conclusion
This study demonstrated a suite of methods and approaches for rice areas changes over the years and its impact on production and prices. First, a baseline rice map of Nepal was produced for last 10 years (start from 2000 to 2009) and their areas calculated. A fuzzy classification accuracy showed that the 9 rice classes were mapped with absolute accuracy ranging from 63.2% to 92.6% and an overall classification accuracy of 79.1% for all the 9 classes. Almost all of the inter-mixing was between two rice classes. Second, the accuracy was also determined by correlating the MODIS-derived rice areas with sub-national statistics obtained from Nepal agriculture department. For this, the R2 values were 96.8% at the district level and 98.32% at the state level for 2000-01. Dynamic mapping rice area for the most recent decade (2000-2009) has been done by identifying 8 main classes of rice (irrigated, rainfed, upland etc.). There is no significant change in the rice area and proportionate area under different rice classes over the years except in 2006. During the year 2006, rice area declined by 14% from the 2005 and 2007 figures. Area under rainfed rice class continues to be the the predominant one from 2000 to 2009. Hence the growth in rice production and productivity has stagnated over the recent decade.
Rice price is increasing steadily since last one decade, however, area and production growth remains stagnant. Moreover, rice yields are variable over the years and are not growing recently in similar pace with rice prices and country's population. Consequently, the country is facing increasing deficit of rice for meeting its food security needs and reducing poverty. Poor consumers and small holder farmers (who are also consumer of rice) are increasingly spending higher share of their income to meet their food needs as a result of increase prices of rice in the market in recent years. This necessitates scientific rice production planning and targeting of suitable new technologies to enhance rice production and productivity in the different rice classes (ecosystems and production systems) as generated from remote sensing analysis and indicated in this paper. Rice research program in Nepal needs to increasingly focus on development and dissemination of stress tolerant rice technologies and associated crop management practices for rainfed environments to reduce production and yield variability over the years and also reduce the negative impact of drought and other abiotic and biotic stresses. Differences in target domain characteristics and technological needs should be accounted for rice technology development and promotion.