Identification of Arid and Drought Prone areas from Remote Sensing based Land Surface Parameters – A study from a Tropical Region, Maharashtra, India. C.V.Srinias1, K.P.R.Vittal Murty2 and Y.V.N.Krishna Murty3 1. Radiological Safety Division, Radiological Safety & Environment Group, Indira Gandhi Centre for Atomic Research, Kalpakkam, India 2. Department of Meteorology & Oceanography, Andhra University, Visakhapatnam India. 3. Regional Centres, National Remote Sensing Centre (NRSC), Hyderabad, India Email of corresponding author: cvsri@igcar.gov.in 1 Abstract In this study the Land Surface Parameters comprising Normalized Difference Vegetation Index (NDVI), surface temperature, albedo, vegetation growing period and dryness quotient were estimated using the NOAA-AVHRR data from a tropical region in central Maharasthra, India to identify the arid and drought regions. The climatic conditions in the drought prone districts were analyzed from the rainfall, water balance, growing period parameters determined from long term mean meteorological data in the study region. A comparative analysis of remote sensing based land surface parameters with the climatic parameters revealed that the drought prone areas were associated with typic arid to dry semi-arid climate, very low NDVI of 0.15 to 0.30, low span of growing season of 60 to 120 days, a high surface albedo of 16.7 to 23.5 per cent, dryness quotient 2.2 to 4.7 and high land surface temperature of 300 K in contrast to the high rainfall zones in the state. These parameters are found to vary between 1992 and 1995 as per the variation in the rainfall. The study revealed that the remote sensing based parameters were highly useful in the detection of arid and drought prone regions and can be operationally used for the purpose with a prior long term analysis of the data. The study also indicated the influence of the the global ElNino / Southern Oscillation phenomena for the aridity and drought events in the region apart from regional parameters. [Keywords: Aridity, Droughts, NDVI, albedo, surface temperature, growing season.] 1. Introduction Arid and semi-arid zones are important components of earth’s environment. They are fragile ecosystems, where vegetation is absent or scarse due to insufficient moisture. They are characterized with frequent droughts where moisture is a major limiting factor. Low rainfall, sparse vegetation and high surface temperature are the prominent characteristics of an arid/ semiarid region. Aridity and droughts are possibly caused by positive feedback mechanisms involving land surface factors (Charney, 1975). Increase in surface albedo due to altered land use/ land cover or over grazing increases radiative losses over an area enhancing the negative net radiation balance. This may affect the local Hadley circulation leading to reduction of rainfall causing further aridity. Study of arid zones and drought prone areas is essential from scientific as well as social and economic points of view. Drought can be defined according to meteorological, hydrological and agricultural criteria. Aridity and drought events are generally associated with anomalous weather patterns, which are reflected in poor quantities and distribution of rainfall over an area. The meteorological causes of drought vary according to the climate. In tropical regions droughts are associated with changes in summer monsoonal circulation causing a delay or failure of the monsoon system leading to break monsoon conditions and prolonged dry phase. In India, the southwest monsoon system describes the rainy season, failure of which leads to aridity and droughts. Forecasting of the droughts in India is linked with the forecasting of monsoon which is influenced by several global and regional factors such as the El Nino and Southern Oscillation (ENSO) events (Ropelewski and Halpert, 1986, Nicholls, 1993). The affected areas are the arid and semi-arid regions where the annual rainfall is not only less but its coefficient of variation is also very high. Droughts in these areas are associated with the absence of significant rainfall for a longer period which is sufficient 3 enough to cause moisture deficits in the soil due to evapotranspiration and decrease in stream flow, disrupting the normal biological and human activities (McNab and Karl, 1991). Several methods have been proposed for the analysis and identification of arid and drought conditions. The conventional methods are based on analysis of rainfall, radiation, temperature (Koppen, 1936), water balance / hydrological approach involving comparison of mean annual precipitation with mean annual potential evapotranspiration and computation of climatic indices (Thorthwaite, 1948; Palmer, 1965; Subramanian et al., 1965; Subrahmanyam, 1983; Rao et al., 1972; George, 1972; Appa Rao et al., 1979; Krishnan, 1988), and energy balance approach involving comparison of mean annual net radiation with the heat flux required to evaporate mean annual precipitation (Budyko, 1958; Lettau, 1969). Arid and semi-arid regions are characterized by low mean annual precipitation of about <250 mm and <800 mm, high surface temperature (Koppen, 1936) and moisture index (Im) <=-61 and -60 to -21 respectively (Thornthwaite, 1948). Conventional methods of climate studies suffer in delineation of spatial extent of aridity/ droughts, because they are based on point observations and analysis of discrete meteorological data. More so it depends on the network of stations for collection of data which is often sparse. Application of Remote sensing techniques in the delineation of vegetation and water on the earth's surface and in determination of weather parameters in atmosphere is unique. Generally the observed changes in vegetation lag in time relative to the conditions described by meteorological indices (e.g., crop moisture index, Palmer drought index) based on observations of rainfall and temperature. Droughts are associated with a decline in biological productivity, increased surface albedo and increased surface temperatures. Remote Sensing techniques have been widely employed for the delineation of vegetation covers. The deficient rainfall of a region 4 would be reflected in the status of crops / vegetation and the surface effects as observed from the vegetation index, surface temperature and albedo. These parameters, singly or combined also provide vital information about the arid conditions and drought. Since remote sensing based methods analyze the data over an extensive area they are more useful compared to classical approaches. Vegetation is the only component known to strongly absorb visible light, and highly reflect and transmit in the near infrared region. These characteristics help in monitoring photosynthetically active vegetation through remotesensing techniques. The Normalized Difference Vegetation Index (NDVI) is a transformation of reflected radiation in the visible (0.58-0.68 m) and near IR (0.725-1.10 m) bands of NOAA AVHRR data and is a function of green leaf area and biomass. Numerous studies showed that NDVI was highly useful in monitoring vegetation and drought (Justice et al , 1985; Kogan, 1995; Jeyseelan and Thiruvengadachari, 1986; Nicholson and Farrar, 1994). In arid and semi-arid regions, the incidence of droughts can be monitored using NDVI which correlates well with percent vegetation cover, biomass, and biological productivity (Nicholson, 1994) and hence can be used to identify vegetation stress zones. National Agricultural Drought Assessment and Monitoring System (NADMS) was initiated in India to monitor the drought conditions periodically based on AVHRR NDVI and supplemented by the ground based rainfall data (Jeyaseelan and Thiruvengadachari, 1986). Land surfaces are typically associated with spatial homogeneity of surface cover. The seasonality characteristics of vegetation canopy can be derived from vegetation indices. Several studies reveal the association of soil moisture and vegetation activity (e.g., Gutman, 1990). NDVI is an integrated function of photosynthesis, leaf area and evapotranspiration (Running and Ramakrishna, 1988). It has been found that NDVI of active vegetation lies typically between 0.1 to 0.6, values at higher end indicating increased 5 photosynthetic activity and greater canopy density (Tarpley et al., 1984). Development of vegetation during the growing season can be monitored by detecting the changes in vegetation indices. Critical thresholds of NDVI can be identified to distinguish vegetated land from nonvegetated land in the non-growing season (Loveland et al., 1991). Thus the land surface parameters (vegetation index, surface temperature, albedo, growing period etc) derived from remote sensing data can be used to detect the land surface effects and as potential parameters for the identification of arid and drought regions in monsoon climatic environment. Maharashtra state, in tropical India, experiences semi-arid climate over a major area. Droughts are common phenomena in various parts of this state. The aridity and the droughts in central Maharashtra are attributed to the rain-shadow effect of the long chain of mountain ranges called Western Ghats (also called Sahyadri ranges), which run along the west coast of India. Drought prone areas in Maharasthra were delineated following a soil climate analysis approach (Challa and Wadodkar, 1996) and were classified into slight, moderate and highly drought prone considering the water retention characteristics of soils, rainfall distribution and its variability (Anonymous, 1993). In the present study an attempt was made to examine the climatic conditions in some of the drought prone districts of central Maharashtra using the remote sensing derived land surface parameters of vegetation index, albedo, land surface temperature and crop growing season derived from NOAA AVHRR environmental satellite data. The above parameters were analysed in relation to the rainfall, water balance components, climatic indices determined from conventional methods using climate / meteorological data. The intent of this study was to examine how well the satellite derived land surface parameters correlate with the climatic conditions determined from conventional data and to study how such parameters can be used for identification of arid and drought prone regions in a tropical region. 6 2. Description of Study Region The study area, Maharashtra lies between 15 44’ and 21 40’ latitudes and 73 15’ and 80 33’ E longitudes and forms a major part of Indian peninsula with an area of about 304391 sq km (Figure 1). It has diverse physiographic, soils, vegetation and climatic conditions. The rainfall received during the south west monsoon from June to September varies from 450 mm in rain shadow areas to 6000-7000 mm in Western Ghats (Raman, 1974) (Figure 1a). It increases from 700 mm in central parts to 1500 mm in eastern parts. The state has four major climatic zones (Figure 1b), i.e., Sub-humid to Semi-arid in the Eastern Maharashtra, Semi-arid in the Central Maharashtra and Perhumid along the Konkan coast plains, Humid to Perhumid transition zone along the Western Ghats (IMD, 1986) which is a long chain of highlands running along the western edge of the Indian subcontinent. The land use of the state comprises forests, croplands, pastures and scrub lands. The forests occupy about 17.3% of the total geographical area (TGA), and are distributed in the western, northern and eastern parts associated with relatively high rainfall. About 58.1 per cent of the land is cropland and the rest is either covered with forest or scrub vegetation and pastures (Challa et al ., 1885). 3. Data and Methods Analysis of climatic data was carried out for nine frequently drought prone areas in semiarid central Maharashtra covering the districts of Dule, Jalgeon, Nasik, Pune, Satara, Sangli, Sholapur, Ahmednagar and Beed. Mean monthly rainfall of over 30 years (India meteorological Department, 1973), mean weekly rainfall data of the years from 1946 to 1980 and evapotranspiration of various locations in the study area (Rao et al., 1971) are used in the water 7 balance analysis (Thornthwaite and Mather, 1955; Subrahmanyam, 1983). The water balance equation (1) was solved in monthly and weekly time steps to evaluate the water deficit (WD) or water surplus (WS) components. The water balance is given by P ET S R 0 (1) where P is the rainfall, ET is the evapotranspiration, S is the change in the soil moisture due to depletion by evapotranspiration or infiltration of water by precipitation and R is the runoff. The release of soil moisture ‘S’ is an exponential function. S AWCe ( P ET ) / AW C (2) AWC is the storage capacity of the soil moisture, (P-ET) are accumulated values of (P-ET)<0 up to the month or week considered, which is always negative. The actual evapotranspiration (AET) is estimated as S+P where S is positive and ET when S is negative. The water deficit is treated as ET-AET. The water surplus is treated as (P-ET)-S which is the water available above the storage capacity of soil that percolates to the water table. The climatological indices viz., humidity index (Ih), ariditiy index (Ia) and the moisture index (Im) on annual basis were computed as Ih WS x100 ET (3) Ia WD x100 ET (4) Im Ih Ia (5) The data on water holding capacity of the soils in the study area were obtained from the Soil Resource Inventory of Maharashtra (Challa et al., 1995). Analysis of water balance was done using the mean climatic data as well as with the yearly data. The dry years were identified based on the annual aridity index, its mean and departure from standard deviation (Subrahmanyam, 8 1983). Vegetation Growing period (GP) was also considered as an indicator of the prevailing climatic conditions as the vegetation would reflect the availability of soil moisture, the antecedent rainfall and temperature regimes (Mandal et al., 1996). The growing period was computed from the mean meteorological data using the Food and Agriculture Organization (FAO) method (Higgins and Kassam, 1981; FAO, 1983) based on comparison of precipitation (P) and potential evapotranspiration (PE) taking into account of the soil water storage. It was estimated as the sum of climatic moist period (CMP) and Soil moisture supported moist period (SMP) from the mean climatic data. The climatic moist period was calculated as the period during which precipitation was greater than or equal to 0.5 PE called vegetation moist phase plus a period in which precipitation was greater than evapotranspiration called the humid phase. The CMP was calculated for 240 stations in the state using data on mean monthly rainfall data and potential evapotranspiraton obtained from IMD scientific reports (IMD, 1973; Rao et al., 1971). The spatial analysis of point data for various parameters was made using SPANS (Spatial Analysis System) Geographical Information System (GIS) (PCI, 1998). Weather stations with their location coordinates (longitude, latitude) were digitized and attributed with the CMP values which were then interpolated using spherical Krigging in GIS. The dryness factor (Lettau, 1969) was calculated according to the relation D Q * /( LxP) (6) where Q* is the net radiation, L is the latent heat of vaporization of water and P is the precipitation. The net radiation term in (6) was evaluated from the procedure outlined by (Budyko, 1958) Q* ( R q)(1 ) ( LWu LWd ) H LE G 9 (7) where LWu and LWd are upward and downward longwave emissions by the earth-atmosphere system, ‘R’ and ‘q’ are the direct and diffuse solar radiation received at the earth’s surface, ‘’ is the albedo of the land surface. The net radiation Q* is estimated from sensible heat flux (H), latent heat flux (LE) and Soil Heat flux (H) terms from the surface energy balance procedure (SEB) (Tarpley, 1994). Surface meteorological data, albedo, surface temperature and roughness coefficient derived from NDVI were used to evaluate the net shortwave radiation, net longwave radiation and soil heat flux components of the SEB. The land surface parameters NDVI, albedo and surface temperature (Ts) and vegetation growing period were derived from NOAA-AVHRR 1 km Global land 10-day composites (USGS, 1993) for the visible, near infra-red and thermal infrared bands. This is a preprocessed data for cloud screening, atmospheric corrections and is available as composite over 10 days periods for perceivable changes in biophysical quantifies. The AVHRR imagery were georeferenced and processed for the estimation of various land surface parameters using the EASI/PACE image processing software (PCI, 1988). The vegetation index NDVI was quantified using a normalized difference of the atmospheric corrected reflectance R1 and R2 of the visible and infrared bands of AVHRR. NDVI was estimated at 10 day intervals at a spatial resolution of 1km. The albedo () was estimated from the spectral radiance (Li) summed over reflective bands (visible, near IR), exo-atmosphere solar irradiance of spectral bands (Ei), solar zenith angle () and distance correction factor (D) for instantaneous sun-earth distance using the procedure outlined by Robinove,(1981) and Saunders (1990). Ts was derived using an empirical relationship between the emissivity and vegetation cover (Van de Griend and Owe, 1993). The roughness coefficient and turbulent transfer coefficient were determined from the NDVI and surface meteorological data (Srinivas et al., 2001a) using aerodynamic turbulent transfer theory. The GP was also 10 estimated from NDVI time series (Loveland 1981; Running et al, 1995) as a period of greenness in which NDVI is above a threshold value of 0.1 to distinguish vegetation from non-vegetated land in the non-growing season. The NDVI time series was compared with the pattern of growing period at each district to find their relative variation and correlation trends. Comparative analysis was made between GP estimated using water balance principle and that obtained from NDVI to determine how the GP obtained from NDVI time series correlates with that obtained from climatic data. The analysis of remote sensing data was selectively made for the years 1992 and 1995 which have different rainfall characteristics as per the availability of the data for limited number of years from USGS in open source. The year 1995 was a La Nina year with above normal rainfall while 1992 has relatively low rainfall compared to 1995. Interpretation of remote sensing derived land surface parameters over these years with contrasting monsoon rainfall characteristics would enable to assess the utility of remote sensing data for climate studies in the study region. 4. Results As per the objectives of the study the climate of the drought prone region was first examined by making long term rainfall data analysis, water balance computation and by vegetation growing period analysis. Subsequently the remote sensing based parameters of vegetation index, NDVI based growing period, albedo, surface temperature, and dryness quotient were analysed to examine their association with the climatic parameters. 4.1 Rainfall in central Maharashtra and ENSO events The drought years and the degree of drought identified from the analysis of rainfall are presented in Table. 1. The annual rainfall fluctuations of the region were expressed as percent 11 standard departure from the long-term climatic mean (Figure 2a). The periodicities in rainfall departures and other parameters are analysed from simple quantitative and qualitative examination of results. Positive and negative departures indicate excess and deficit rainfall conditions respectively. The rainfall followed a cycle of decreasing and increasing phases with a period ranging from 4 to 7 years. Rainfall averaged across the drought prone areas in Maharashtra seems to have association with El Nino events for some years. Most of these years were identified to be El Nino years (Quinn et.al., 1987; Rasmusson and Carpenter,1983; Khan,1994) and significant departures of rainfall below normal are observed (Table 1). Mean annual rainfall, aridity index in the study region were plotted with the Southern Oscillation Index (SOI) to examine their relative variation. Rainfall had similar trend as that of SOI with alternating increasing and decreasing magnitudes but with slight lead in time. It is noted that the rainfall followed the same periodicity as with SOI, except for few deviations, thus indicating the influence of the El Nino phenomena on regional rainfall (Figures 2b). The aridity index determined from climatic water balance procedure and averaged for these areas had an opposite trend with SOI (Figure 2c) i.e, higher aridity index corresponded with lower SOI and vice versa. Years of extreme drought coincided with years when the Southern Oscillation Index (SOI) was strongly to moderately negative. Conversely major floods occurred when the SOI was strongly positive. The years marked with El Nino were characterized by higher aridity index values ranging from 52 to 81 and lower SOI values ranging from -0.15 to -1.8. The rainfall departures from normal in El Nino years ranged between -40% and +10.1%. The La Nina years were characterized with aridity index ranging between 59.3 to 71.8 % and SOI ranging from -1.0 to 1.0. The analysis indicated the possible existence of a teleconnection between the drought events in Maharashtra and the ElNino-Southern Oscillation phenomena, in general, which is well 12 documented in several studies on Indian monsoon rainfall and other studies on major drought and flood episodes in the dry land areas of the world (Williams and Balling, 1993). However, the present analysis showed that some normal years were also marked with SOI anomalies, thereby indicating that ENSO events might be one among other factors responsible for regional rainfall variation in drought prone areas of Maharashtra.. 4.2 Drought Climatology and Water Balance The drought frequency expressed as a percentage of the drought years for a few drought prone districts was presented in Table 2. The annual average rainfall during the period of study (1946 to 1988) was found to range from 532 mm at Satara district to 780 mm at Jalgaon district. Considering aridity anomaly as an indicator for incidence of drought, the districts Dhule and Pune have 31%, Jalgaon has 21%, Parbani has 20%, Sholapur has 16%, Satara has 17% and Nasik has 12% of years of drought occurrence between 1946 and 1988. The periodicity of drought cycle varies from 3 yrs at Dhule and Pune districts to 8 years at Nasik district. The principal drought class was moderate in Sholapur, Nasik, moderate to severe in Parbhani, severe in Dhule, Jalgaon and Pune and large in Satara district. The coefficient of variation of rainfall in these districts was found to range from 21.6 % to 31.0%. Water balance analysis of the mean climatic data of 45 locations in the drought prone areas in the semiarid central Maharashtra (Table 2) indicated that there was no seasonal water surplus throughout the year. The districts Jalgaon, Sholarpur, Sangli, Satara and parts of Aurangabad, Pune were characterized with severe arid conditions. It was noted that the annual water deficit in various parts of central Maharastra was of the order of 950 mm in moderate drought prone districts, 1100 mm in severe drought prone districts, and 1200 mm in large 13 drought prone districts. The aridity index was about 60.0 in moderate drought districts, 65.0 in severe drought districts and about 70.0 in large drought districts. The climate was dry semi-arid in a major area and typic arid in some parts in these districts. The moisture index (Im) was between -83.3 and -66.7 in parts of Satara, Sholapur and Sangli districts and between -66.7 and 50.0 in major areas in the districts of Dhule, Jalgaon, Nasik, Ahmednagar, Pune, Aurangabad, Sangli and Sholapur (Figure.1b). As per water balance calculation, the soil remained dry in most part of the year and recharge/storage of a very meager amount occured during the rainy season between June to September. The mean soil moisture recharge estimated from soil water budgeting was found to be of the order of 56.1 mm in Aurangabad, 81.2 mm in Dhule, 111.1 in Jalgaon, 18.1 mm in Nasik, 30 mm in Pune, 44.4 mm in Sholapur, 15 mm in Sangli and 16.1 mm in Satara districts. As already noted the annual rainfall in these districts had large increasing and decreasing variations with periodicity of about 4 or 5 years (Figure.2a). These districts were classified into different degrees of drought proneness based on the moisture retention characteristics of soils in earlier studies (e.g, Anonymous, 1993; Challa and Wadodkar, 1996). 4.3 Land Use/ Land cover and NDVI The land use/ land cover of the study area was delineated from field observations collected during the soil resource mapping of Maharastra (Challa et al, 1995) (Figure 1c). This data was digitized and spatially mapped using GIS and used in the interpretation of prevailing vegetation. The principal crops cultivated in the arid to semi-arid drought prone districts were Jowar, Bajra, Redgram, Gram, Cotton, groundnut, Safflower, Soybean and fodder which are basically dry land crops. Rice and Sugarcane are grown in parts of Kolhapur district, the western parts of which were received high rainfall. Fallow and scrublands were noticed in the districts of Satara, Sangli, Beed, Ahmednagar and parts of Pune where fodder, maize, Jowar, cotton and 14 other commercial crops were grown. The average annual rainfall of the study area was low in 1992 relative to that in 1995. Figure 3 gives the relative variations in the average NDVI and the percent crop cover during the growing season in the above districts for the year 1992. The NDVI was noted to be low in the districts Solapur, Pune, Jalgaon, Satara and Sangli which had scrublands and fallow lands and very low percent cropped area. A similar variation between NDVI and the percent cropped area was also noted for the year 1995 (not shown). 4.4 NDVI and Growing period Figure 4 gives the monthly variation of NDVI and rainfall at four stations Satara, Pune, Dhule and Jalgaon in the drought prone region in 1992, which indicated NDVI had positive trend with rainfall at the respective locations. A time lag of about 30 days was observed in the trends of NDVI with rainfall indicating gradual response of vegetation to rainfall. Analysis of seasonal NDVI with mean annual rainfall showed that the NDVI followed a similar trend as that of rainfall in both 1992 and 1995 at various stations in Maharasthra (Figure 5). The NDVI values were relatively low in 1992 compared to 1995 possibly due to the reduction of rainfall in 1992. NDVI was relatively low in semi-arid rain-shadow central region as compared to the humid/ perhumid Western Ghat region and the sub-humid eastern Maharasthra, indicating a low percentage of surface vegetation cover in semi-arid region in both the years. The climatic moist period (Higgins and Kassam, 1981; Srinivas et al., 2001b) in Maharasthra varied from 59 to 182 days (Figure 6d) and followed a similar spatial pattern as that of spatial rainfall distribution in the study region (Figure 1a). The lowest growing period of 60 to 135 was observed in eastern parts of Satara, south eastern part of Pune, northern part of Sangli and western parts of Sholapur. The rain-shadow dry region was found to have a growing period of 135 to 165 days with some pockets having 120-135 days and 150-165 days. Central 15 Maharashtra region had a growing season of 135-180 days. Konkan coast covering the districts of Sindhudurg, Ratnagiri, Raigad and Thane districts was found to have a growing period of 165 to 227 days, the highest was noted in Ratnagiri and coastal strip of Thane and Sindhudurg. A high growing period of 180 to 227 days was also noted in eastern Maharashtra in part of Chandrapur, Gadchchiroli, Bhandara and Nagpur districts, which receive a high annual rainfall of about 1100 to 1500 mm. The NDVI integrated from June to October was considered as an indicator of biomass during the growing season in the study region. To examine the spatio-temporal variation of NDVI and to assess the growing season from its time series data, five monthly data sets were considered over September, October, November, December and February. The average NDVI in the middle 10 day period (11-20th) of each month was considered representative of NDVI for each month for this purpose. The NDVI values were considered in the range from 0.1 to >0.5 at 0.05 intervals. NDVI values higher than 0.5 were considered to characterize the perennial vegetation such as forests, values lower than 0.15 as scrublands and the intermediate values with vegetation comprising various crops. It was noted from the NDVI map of the study area for February (nongrowing season) (Figure 6a), September and October (growing season) (Figures 6b,c) that the districts of Dhule (part), Jalgaon (part), Ahmednagar, Aurangabad, Beed, Pune, Sholapur and Sangli were characterized by relatively low NDVI (<0.1 to 0.25) in the peak of the growing season which indicated the prevalence of unfavorable growing conditions in these areas. The growing period of these districts estimated from NDVI time series was found to be of the order of 75 to 125 days (Srinivas et al., 2001b). During September and October many districts had NDVI in the range of 0.2 to 0.5 (Figures 6b,c) except the eastern parts of Satara, Sangli, western parts of Sholapur, Pune, central parts of Dhule, Nasik and Beed. This indicates the senescent 16 stage of vegetation in those areas and a growing period of 60 to 120 days. NDVI in September in other areas was typically between 0.3 to 0.5 representing vigorous vegetation. NDVI progressively decreased in November, December and February in all districts, the area with NDVI < 0.25 increased initially up to September, decreased gradually thereupon till November and increased again thereafter (Table 3). The area under the NDVI range 0.25-0.35 increased from August to December and thereafter a decreasing trend was observed. The area with NDVI range 0.35-0.40 increased from August to November and decreased subsequently. The NDVI classes 0.40-0.45, 0.45-0.50 and >0.5 are found to occur over a greater percent of the area in August, and reduced abruptly from September onwards. In November NDVI in the districts Dhule, eastern Nasik, Ahmednagar, Pune, Sholapur, eastern portion of Osmanabad, Beed, Aurangabad was noted to be in the range of <0.1 to 0.25 indicating the withered stage of vegetation, limiting the growing season to 135 to 150 days. The area with dormant vegetation was further increased in December as observed from NDVI map. The districts of Buldhana, Akola, Parbhani and parts of Nanded, Latur were characterized with NDVI below 0.20 with a growing period of 165-180 days. The areas occupied by forests in the districts of Chandrapur, Gadchchiroli, parts of Bhandara, Western Ghats region in Sindhudurg, Ratnagiri, Raigad were in the NDVI range of 0.45 - >0.50 even in February (Fig.7), while a major area of the state had NDVI below 0.3. This indicated a growing season of 220 days or more in areas having perennial vegetation. These areas were associated with dense forests and heavy precipitation in the range of 1300 mm to 6000 mm (Challa et al., 1995). The results agree with the FAO model output and a correlation factor of r=0.78 was obtained between the two estimates of growing period from climatic data and NDVI (Table 4, Figure 7). However, 17 deviation in some cases might have been due to using fewer NDVI data sets. Multi temporal NDVI data sets would be helpful in detecting the duration of growing season more accurately. 4.5 Surface albedo Figure 8 shows the surface albedo map of Maharasthra for February during non-growing season and November during growing season. The surface albedo was found to be high in the premonsoon season i.e., April and May, decreased to low values in the rainy season (June to September) and increased to moderate to high values in the post monsoon and winter season (November to December). The albedo at various locations was observed to be characteristic of the local NDVI pattern. Both albedo and NDVI patterns significantly varied spatially depending upon the rainfall and vegetation pattern at each location. The surface albedo was high in central Maharashtra relative to the western and eastern Maharashtra regions (Table 5). Albedo was found to range between 17.6 to 20.3 with an average value of 19.1 in the semi-arid region, and between 15.2 and 19.6 with an average value of 17.3 in the subhumid eastern Maharasthra region in 1992. In 1995, albedo ranged between 18.7 to 20.9 with an average value of 19.6 in humid/ perhumid region, between 17.4 to 20.6 with an average of 18.8 in semi-arid region, and between 16.5 to 22.4 with an average value of 20.4 in sub-humid eastern Maharashtra region. Albedo was not only higher in drought prone areas, but its interannual variation was also lower. The variation of surface albedo during 1992 and 1995 was between 0.3 and 0.4 in arid and semi-arid areas and about 2.0 in humid/ perhumid high rainfall regions. The seasonal average surface albedo from April to December was very low in the subhumid eastern region, and relatively high in the semi-arid dry land areas. 4.6 Land Surface Temperature 18 The land surface temperatures in Maharasthra in the pre-monsoon summer time in April, 1995 and post monsoon time in November 1995 were shown in Figure 9. Generally Land surface temperature was noted to be high in the central Maharashtra and decreased towards the eastern Maharasthra and Western Ghats region. Seasonal average land surface temperature in the study area (Table 6) ranged between 298.7 °K and 301.3 °K with an average value of 299.8 °K in semi-arid region, and between 294.8 °K and 302.0 °K with an average value of 297.8 °K in subhumid eastern region respectively in 1992. In 1995, the surface temperature was noted to vary between 295.4 °K and 301.8 °K with an average value of 298.9 °K in humid/perhumid western Maharashtra, between 298.1 °K and 303.8 °K with an average value of 300.9 °K in semi-arid region, and between 294.3 °K and 301.2 °K with an average value of 297.8 °K in sub-humid eastern region.. The semi-arid central region identified with the drought prone districts was found to experience higher surface temperatures both in 1992 and 1995. The above analysis reveals that humid/ perhumid and subhumid regions had relatively low surface temperatures in both years of observation. 4.7 Dryness Quotient The dryness quotient represents the net radiation available for evaporation from a wet surface to the heat required to evaporate the mean annual precipitation. It is less than unity in humid areas and greater than unity in dry areas. Increasing values represent the trend towards an arid climate. The dryness quotient evaluated from the surface energy balance was greater than 2.0 for the districts of Jalgaon, Dhule, Nasik, Ahmednagar, Parbhani, Pune, Beed, Satara and Sholapur in both 1992 and 1995 (Table 7). The dryness quotient for these districts ranged between 2.2 and 4.7, the districts with high values being Nasik, Ahmednagar, Dhule, Pune, Satara and Jalgaon. The dryness quotient for the subhumid eastern Maharasthra region ranged 19 between 1.55 to 2.56, and for the humid/ perhumid western ghat region between 0.31 to 1.55 (Figures 10 a, b). 5. Conclusions In this study the satellite remote sensing derived NDVI, albedo, surface temperature, dryness ratio and vegetation growing period from NOAA-AVHRR data over the tropical region, Maharashtra, in India were analysed in association with the climatic data and the water balance components estimated from climatic data in an attempt to determine the utility of remote sensing based land surface parameters in climate analysis and to delineate the arid / drought affected areas in the region. The analysis was made with special emphasis over the arid to semiarid central Maharahstra. The analysis of rainfall and water balance components in the semi-arid central Maharashtra showed that the districts of Sholapur, Parbhani, Dhule, Jalgaon, Nasik, Pune, Ahmednagar, Beed, Sangli and parts of Aurangabad experience low mean annual rainfall with a coefficient of variation of 20 to 31% and high aridity index (between 51% to 74%). The drought cycle had a frequency between 3 to 8 years in these parts of the state. The vegetation in these districts consisted mainly of dry land crops with fallow and scrublands. The rain shadow region was characterized with typic arid to semi-arid conditions with a low vegetation index (0.15 to 0.30), short growing season (60 to 120 days), high surface albedo (16.7 to 23.5 per cent), high dryness quotient (2.0 to 0.47) and high surface temperatures (>299.8 °K). NDVI analysis revealed the presence of dense vegetative cover in areas with high amount of mean annual rainfall, particularly in humid/ perhumid Western Ghats region and the subhumid eastern Maharasthra region, and very low NDVI values in the rain shadow semi-arid area in central Maharasthra. The growing season mean NDVI varied between 0.09 to 0.28 with a 20 mean of 0.17 in 1992 and from 0.23 to 0.35 with a mean of 0.28 in 1995 in the above districts. The NDVI pattern reflected the span of growing periods and the prevailing land cover in the above districts. The relatively low NDVI during the growing season reflects the typically poor vegetation conditions and the prevailing aridity in these parts. The growing period derived from NDVI increased from deficit monsoon year 1992 to good monsoon year 1995 by 10 to 30 days in the sub-humid region, 30 to 60 days in the semi-arid region and 10 to 40 days in the west coast following the NDVI variation between the two years in the respective regions. While Albedo slightly increased (~1-2%) in the sub-humid and perhumid west coast regions it decreased by ~1% in the arid/semi-arid central region from 1992 to 1995. The land surface temperatures increased by ~1C in central region from 1992 to 1995. A long term NDVI time series analysis of the study area would help in the spatial delineation of arid/ drought prone areas by assessing the deviation of NDVI from its long term normal value (Nicholson et al, 1988). The present study showed NDVI was low in sparsely vegetated areas with low rainfall and high in regions with perennial vegetation and high rainfall. A threshold value of NDVI during the growing season can be used to distinguish the semi-arid and arid regions. The study reveals that the land surface parameters viz., NDVI, albedo, land surface temperature, dryness quotient and vegetation growing period derived from remote sensing data correlate well with the climatic data (rainfall, water balance components etc) and that they can be used in the identification and delineation of aird/ semi-arid and drought prone areas. This method based on analysis of satellite imagery has the unique advantages of obtaining more accurate spatial estimation of aridity and drought over the conventional methods based on traditional meteorological point information. However, a time series analysis of parameters NDVI, albedo, surface temperature, crop growing season over climatically significant long period over a given 21 region would be necessary to determine the threshold limits of each parameter for their application in the detection of arid and drought conditions. 6. Acknowledgements Authors are thankful to the India Meteorological Department, Pune for providing the meteorological data used in the study. The NOAA AVHRR data was obtained from the USGS EROS data center from their online achieves. Sri S. Rajasekhar, RRSSC, Nagpur is acknowledged for his help in the processing of NOAA AVHRR data. Thanks are due to anonymous reviewers for their suggestions to improve the manuscript. 22 References: Appa Rao, G., Abhiyankar, V.P. and Mahajan, A.V. (1979). Analysis of 1979 Kharif agricultural drought over Pre-Published Sci. Report No. 81/2, IMD, Meteorological office, Pune. Anonymous (1993). Districtwise General Statistical Information 1991-92 & 1992-93, Part II – Epitome of Agriculture in Maharashtra. Department of Agric., Pune. M.S. Budyko, M.I (1958) The Heat Balance of the Earth's surface. Translated by N.Stepanava under US Dept. of Commerce, Washington, D.C., 1958, 259 pp. Challa,.O., Vadivelu, S and Sehgal, J. (1995). Soils of Maharashtra for optimising Land Use. NBSS Publ.54b (Soils of India Series)., NBSS&LUP, Nagpur, 112pp. Challa,O., and Wadodkar, M.R. (1996). Delineation of Drought Prone Areas in Maharashtra : A Soil-Climate Based Approach. Agropedology., 6(2), pp.21-27. Charney, J.G. (1975). Dynamics of deserts and drought in Sahel. Quart. J. Roy. Meteor. Soc., 101, pp.193-202. FAO., (1983). Guidelines : Land Evaluation of Rainfed Agriculture. Soils Bulletin, 52, FAO, Rome, Italy, 237 pp. 23 George, C.J. (1972). Proc. Symp. On Droughts in Asiatic Monsoon Area, Dec. 14-16, Pune, INSA Bulletin No.54. Gutman Garik, G.,(1990). Towards monitoring droughts from Space. J.Climate., 3, pp. 282-295. Higgins, G.M. and Kassam, A.H.(1981). The FAO agro-ecological zone approach to determination of land potential, Pedologie, 31(2), pp.147-168. IMD (1973). Climate of Maharashtra State. India Meteorological Department, New Delhi, 129 pp. IMD (1986). Agroclimatic Atlas of India. India Meteorological Department, Division of Agricultural Meteorology, Pune, 114 pp. Jeyseelan, A.T. and Thiruvengadachari, S (1986). Current Drought Monitoring System in Andhra Pradesh State. Report No. IRS-UP-NRSA-DRM-TR-03-1986, NRSA, Hyderabad, India. Justice, C.O., Townshend, J.R.G.., Holben, B.N., and Tucker, C.J. (1985). Analysis of the phenology of global vegetation using meteorological satellite data. Int. J. Remote Sens., 6, pp.1271-1318. Khan, H.J., 1994..Prophets of Rain. Science Reporter., 31(8), pp.9-37. 24 Kogan, F. (1995). Drought of the late 1980s in the United States as derived from NOAA polar orbiting satellite data. Bull. Amer. Meteorol. Soc., 76, pp.655-668. Koppen, W. (1936). In Hardluichdu Klimatologie (eds Koppen, W.and Geiger); Gehruder Bomtraenger, Berlin, 1, Pt. C. Krishnan, A. (1988). Delineation of climatic zones of India and its practical application in agriculture. Fertilizer News 33(4): 11-19. Lettau, H.. (1969). Evapotranspiration Climatology I. A new approach to numerical prediction of monthly Evapotrasnpiration, Runoff, and Soil Moisture. Mon. Wea. Rev., 97, 691–699. Loveland, T., Merchant., J., Ohlen, D..and Brown, J. (1991). Development of a Land cover characteristics database for the conterminous U.S., Photogramm. Eng. Remote Sens., 57, pp.1453-1463. Mandal, C., Mandal, D.K., and Srinivas, C.V. (1996). Defining Climatic map of India following FAO Growing Period Concept. Geographical Review of India, Vol.58, 243-250. Mc Nab, A.L. and Karl, T.R.(1991). Climate and Droughts. National Water Summary 1988-89, Hydrologic Events and Floods and Droughts. US Geological Survey Water- Supply Paper, 2375, pp. 89-98. 25 Nicholson.S.E, Tucker, C.J., and Ba, M.B.(1998). Desertification, Drought and Surface Vegetation : An example from the west African Sahel. Bull. Amer. Meteorol. Soc., 79(5), pp.815-829. Nicholls, N.(1993). ENSO, drought and flooding rain in Southeast Asia. pp.154-175 in : Southeast Asia’s environmental future. The search for sustainability. Oxford Univ. Press., 456 pp. Nicholson, S.L. (1994): On the use of the Normalized Difference Vegetation Index as an indicator of rainfall. Global Precipitations and Climate change, NATO ASI Series 1, 26, Springer, pp.293-306. Nicholson, S.E and Farrar, T.J.(1994). The influence of soil type on the response between NDVI, rainfall and soil moisture in semi-arid Botswana. Part I. Response to rainfall. Remote. Sens. Environ., 50, pp.107-120. Noy-Meir, I. (1985). Desert ecosystem structure and function. Hot deserts and Arid shrublands, (Eds. M.Evenari I.Noy-Meir and D.W.Goodball). Ecosystems of the world. 12 A, Elsevier, pp.93-103. Palmer, W.C. (1965). Meteorological drought. U.S.Weather Bureau, Research paper No.45, U.S. Department of Commerce, Washington. 26 PCI (1988) Engineering Analysis and Scientific Interface. User's manual of EASI/PACE Ver. 6.3, 221 pp., PCI Geomatics, Richmond Hill, Ontario, Canada PCI., (1998). Spatial Analysis System. Manual of SPANS. Ver.7.0, Language Reference, pp.1107,Tydac Research Inc., Ontario, Canada. Quinn, W.H., Neal, V.T.., Santiago, E., and De Mayolo, 1987. El Nino occurrences over the past four and a half centuries. J. Geophys. Res., 92(C13), pp.14449-14461. Raman, C.R.V. (1974). Analysis of commencement of Monsoon Rains over Maharashtra State for Agriculture planning. Pre-published Scientific Report No.216, India Meteorological Department, Meteorological Office, Pune., 19 p. Rao, K.N., George, C..J. and Ramasastri, K.S.(1971). Potential Evapotranspiration (PE) over India. Pre-published Scientific Report No.136, IMD, Meteorological Office, Pune., 12p. Rasmusson, E.M. and Carpenter, T.H..(1983). The relationship between eastern equatorial Pacific sea surface temperature and rainfall over India and Sri Lanka. Mon. Wea. Rev., 111, pp.517 - 528. Robinove, C.J., Charvez, P.S., Gehring, D., and Holmgren, R., (1981). Arid land monitoring using albedo difference images. Rem. Sens. Environ., 1, pp.133-156. 27 Ropelewski, C.F. and Halpert, M.S.(1986). Global and Regional Scale Precipitation Patterns Associated with the ElNino / Southern Oscillation. Mon. Wea. Rev., 115, pp.1606-1626. Running, S.W and Ramakrishna, N.R.(1988). Relating Seasonal Patterns of the AVHRR Vegetation Index to Simulated Photosynthesis and Transpiration of Forests in Different Climates. Rem. Sens. Environ., 24, pp.347-367 Running, S.W, Loveland, T.R., Lpierce, L., Nemani, R.R., and Hunt, E.R.(1995). A Remote Sensing Based Vegetation Classification Logic for Global Land Cover Analysis. Rem. Sens. Environ., 51, pp.39-48. Saunders, R.W. (1990). The determination of broad band surface albedo from AVHRR visible and near-infrared radiances. Int. J. Remote Sens., 11, pp.49-67 Srinivas, C.V., Vittal Murty, K.P.R.., Krishna Murty, Y.V.N., Krishna, N.D.R., and Obireddy, G.P. (2001a). Land Surface Climatic Parameters over Maharashtra derived from NOAAAVHRR, Proceedings of International Conference on Remote Sensing and GIS/ GPS (ICORG 2001), Volume no.II, BPB Publications, pp 306-311. Srinivas, C.V., Vittal Murty, K.P.R., Krishna Murty, Y.V.N., Maji, A.K., Krishna, N.D.R. and Obireddy, G.P. (2001b). Evaluation of growing period over Maharashtra in GIS and its validation using NDVI derived from NOAA AVHRR, Proceedings of International 28 Conference on Remote Sensing and GIS/ GPS (ICORG 2001), Volume no.II, BPB Publications, pp. 460-466. Subrahmanyam, V.P. (1983). Water Balance and Its Applications in Agriculture. Meteorological monogarpah., Andhra University Press, Visakhapatnam, 133 pp. Subrahmanyam, A.R. (1983). Agro-ecological zones of India. Arc. Met. Geoph. Boocl. Ser B. 32, 329-333. Tarpley, J.D., Schneider, S.R. and Money, R.L. (1984). Global Vegetation Indices from NOAA– 7 meteorological Satellite. J.Clim. Appl. Met., 23, pp.491-494. Tarpley, J.D. (1994). Monthly Evapotranspiration form satellite and conventional meteorological observations. J. Clim. Appl. Met., 23, pp.491-494. Thornthwaite, C.W. (1948). An approach toward a rational classification of climate. Geog. Rev., 38, pp.55 - 94. Thornthwaite, C.W. and Mather, J.R. (1955). The Water Balance. Climatology 8(1), Drexel Institue of Technology, Laboratory of Climatology, Centerton, N.J, 104p. U.S.Geological Survey. (1993). Global Land 1-km AVHRR Data Set Project. Electronic version., http://edcwww/landdaac/1km/1kmhomepage.html. 29 Van de Griend and Owe, A.A.(1993). On the relationship between thermal emissivity and the Normalized Difference Vegetation Index for natural surfaces. Int. J. Remote. Sens., 14, pp.1119-1131. Williams, M.A.J and R.C.Balling, 1993 : Interactions of Desertification and Climate. Chapter 6, Impact of Climatic variability on the hydrological cycle, pp.1-4, Preliminary Abstracts Prepared for the World Meteorological Organization and the United Nations Environment Program . 30 Table 1 Rainfall, aridity index, southern oscillation index and rainfall departures in the drought prone area of Maharashtra Year Rainfall Ia SOI Rainfall El Nino / La Nina Drought (mm) departure Year status 1946 762.2 59.3 0.16 6.0 Normal N 1947 646.9 64.2 0.06 -10.0 Normal N 1948 830.4 57.3 0.04 15.5 Normal N 1949 802.7 60.3 -0.02 11.6 La Nina N 1950 642.9 65.1 1.85 -10.6 Normal M 1951 592.2 65.7 -1.10 -17.6 El Nino (Q) S 1952 461.0 73.4 0.20 -35.9 Normal S 1953 616.9 67.8 -1.10 -14.2 El Nino (Q&R) S 1954 719.9 61.9 -0.15 0.1 La Nina N 1955 848.3 59.2 1.75 18.0 Normal N 1956 1000.4 52.0 1.20 39.1 Normal N 1957 791.8 61.9 -0.75 10.1 El Nino (Q&R) S 1958 743.4 61.9 -0.75 3.4 Normal N 1959 864.1 55.7 -0.20 20.2 Normal N 1960 690.8 62.5 0.25 -3.9 Normal N 1961 843.5 60.2 -0.20 17.3 Normal N 1962 698.9 60.6 0.25 -2.8 Normal N 1963 736.5 63.2 -1.10 2.4 Normal M 1964 716.4 61.2 1.00 -0.4 La Nina N 1965 568.1 70.0 -1.15 -21.0 El Nino (Q&R) S 1966 615.0 71.8 -1.00 -14.5 La Nina M 1967 683.0 61.7 0.10 -5.0 Normal N 1968 560.5 69.9 0.20 -22.0 Normal M 1969 724.3 62.8 -0.80 0.7 El Nino (R ) L 1970 708.9 63.6 0.00 -1.4 La Nina N 1971 506.6 70.9 1.00 -29.5 Normal M 1972 424.0 81.4 -1.50 -41.0 El Nino (Q&R) D 1973 723.1 60.2 0.15 0.6 La Nina N 1974 732.1 63.1 0.80 1.8 Normal N 1975 918.8 59.3 1.00 27.8 La Nina N 1976 680.9 64.8 -0.70 -5.3 El Nino (Q&R) L 1977 720.1 63.5 -1.20 0.2 Normal N 1978 716.4 62.9 0.20 -0.4 La Nina N 1979 745.6 61.6 0.10 3.7 Normal N 1980 652.9 66.2 -0.50 -9.2 Normal N 1987 882.4 56.0 -1.80 22.7 Normal N 1988 1031.5 64.8 0.00 43.5 La Nina N N- No drought, M - Moderate drought, L- Large drought, S - Severe drought, D - Disastrous drought. Q - El Nino (Quinn et al., 1987), R - El Nino (Rasmusson & Carpenter, 1983) 31 Table 2 Rainfall characteristics and drought periodicity in the drought affected districts of Maharashtra District Annual Monsoon rainfall rainfall (mm) (mm) Solapur 750.6 558.6 Parbhani 920 780.4 Dhule 622.5 537.3 Jalgaon 780.6 663.7 Nasik 591.6 466.3 Satara 532.8 335.3 Pune 691.8 510.9 Aurangabad 615.4 500.0 Kolhapur 670.0 515.0 Sangli 555.0 485.0 Coeff. % of Variation drought years 31 16 28.4 20 25 31 21.6 21 26.3 12 24.2 17 23.9 31 24.5 30 23.7 22 25.6 25 Water deficit Aridity index 1161.0 1050.0 890.7 1175.0 1064.5 1190.5 972.5 1004.3 989.5 1135.6 63.2 56 68.3 63.5 64.1 71 59.1 62.2 59.0 67.1 32 Climate Typic arid Dry Sei-arid Dry Semi-arid Dry Semi-arid Dry Semi-arid Typic arid Dry semi-arid Dry Semi-arid Typic arid Typic arid Drought periodici ty 5 3 6 8 6 5 3 5 5 5 Drought class Severe Moderate Severe Severe Moderate Large Severe Severe Moderate Severe Table 3 Area (in %) in each NDVI range in different months during growing season NDVI Range % of Area September October November December February 4.28 17.71 15.3 0.09 0.23 1.8 3.98 11.06 5.96 0.54 2.4 6.74 5.91 12.16 7.73 3.46 8.48 14.51 8.14 12.92 12.2 9.49 19.08 18.88 10.29 12.54 14.87 17.83 23.58 19.71 12.47 11.48 13.4 20.58 20.36 16.06 13.31 9.33 10.51 18.98 12.12 9.75 12.82 5.94 7.88 13.55 6.1 5.2 10.52 3.74 6.21 6.8 3.59 3.01 18.28 3.13 5.94 8.69 4.06 2.48 August 1 < 0.10 2 0.10 - 0.15 3 0.15 - 0.20 4 0.20 - 0.25 5 0.25 - 0.30 6 0.30 - 0.35 7 0.35 - 0.40 8 0.40 - 0.45 9 0.45 - 0.50 10 > 0.50 Table 4. Growing period estimated using FAO method and form NDVI data in Mahrashtra Location Growing period from NDVI 1992 Akola Chandrapur Jalgaon Dhule Nasik Ahmednagar Parbhani Raigarh Pune Beed Satara Ratnagiri Sindhudurg Sholapur Thane Nagpur Kolhapur 1995 150 215 180 120 150 180 180 180 150 120 90 180 180 180 180 180 180 180 220 180 120 180 180 180 210 180 180 150 210 220 90 190 190 210 33 Growing period by FAO method 1992 1995 176 160 196 203 180 164 130 132 147 125 150 171 180 165 --230 160 150 120 175 90 145 198 225 173 232 133 150 180 174 169 164 196 203 Table.5 Seasonal Surface Albedo in different regions of Maharashtra in 1992 and 1995 Region Eastern Maharashtra (R1) Semi-arid rainshadow central (R2) Western Ghats & Konkan coast (R3) Location Nagpur Akola Chandrapur Parbhani Jalgaon Dhule Rahuri Pune Ashti Satara Sholapur Nasik Karjat Dapoli Wakawali Borgaon Kolhapur Shirala Kundewadi Kalyani 1992 Alebdo Regional average 19.19 20.32 19.81 17.63 17.90 16.65 19.80 18.38 23.46 17.42 20.03 15.16 19.46 15.77 16.53 18.52 19.62 18.69 16.18 15.74 19.24 19.09 17.30 34 1995 Albedo Regional average 18.92 20.93 19.96 18.69 18.53 19.81 17.53 20.55 17.44 19.34 18.38 16.53 22.36 21.67 22.31 17.78 21.31 21.58 19.95 20.29 19.63 18.80 20.42 Table 6.Seasonal Land Surface Temperature in different regions of Maharashtra in 1992 and 1995. Land Surface Temperature LST (K) in Region Location 1992 1995 LST Regional LST Regional average average Eastern Nagpur 298.92 299.5 297.05 298.93 Maharashtra Akola 298.70 301.39 (R1) Chandrapur 298.81 295.44 Parbhani 301.34 301.82 Semi-arid Jalgaon 303.45 299.8 303.75 300.90 rainshadow Dhule 298.93 300.64 central Rahuri 295.91 298.08 (R2) Pune 302.54 301.70 Ashti 298.77 299.78 Satara 298.94 299.14 Sholapur 299.87 303.55 Western Ghats Nasik 300.04 297.83 301.17 297.80 & Konkan coast Karjat 296.89 295.92 (R3) Dapoli 297.17 297.59 Wakawali 298.08 298.45 Borgaon 294.75 294.34 Kolhapur 297.17 298.59 Shirala 295.20 296.74 Kundewadi 299.22 296.85 Kalyani 301.96 300.84 35 Table.7 Net radiation, Rainfall and Dryness Quotient in 1992 & 1995 from selected locations in Maharashtra. Station/ Year Nagpur Akola Chandrapur Jalgaon Dhule Nasik Ahmednagar Parbhani Raigad Pune Beed Satara Ratnagiri Wakawli Borgaon Kolhapur Shirla Kundewadi Kalyan Sholapur Q* 1424.70 1530.30 1422.10 1561.40 1487.55 1631.30 1458.30 1597.80 1498.20 1419.80 1591.10 1546.40 1552.80 1517.90 1535.60 1564.75 1598.62 1482.80 1992 R 716.40 843.80 1181.70 814.40 728.60 537.80 411.20 805.92 668.80 610.60 414.20 3001.10 3692.20 735.00 1273.80 6435.40 273.30 751.90 Q*/L.R 2.56 2.33 1.55 2.47 2.63 3.90 4.57 2.55 2.88 2.99 4.94 0.66 0.54 2.66 1.55 0.31 7.53 2.54 Q* 1300.10 1499.40 1414.70 1530.20 1566.40 1584.10 1470.80 1531.10 1336.60 1443.40 1518.63 1523.58 1350.10 1326.90 1508.70 1431.10 1361.65 1448.30 1450.20 1442.10 1995 R 946.40 574.00 1036.00 556.70 479.20 436.70 587.30 789.50 3984.20 618.30 868.00 541.50 2956.70 3046.70 887.80 926.30 4387.30 926.30 795.60 843.80 Q*/L.R 1.77 3.36 1.76 3.54 4.20 4.67 3.22 2.50 0.43 3.00 2.25 3.62 0.59 0.56 2.19 1.99 0.40 2.01 2.35 2.20 Legend : Q* Net radiation (Wm-2), R Rainfall (mm), Q/L.R Dryness ratio. 36 Figure 1. The study area in Maharashtra, India with a) Rainfall b) climatic zones c) and land use / land cover. 37 Figure 2. Departure of annual rainfall from the climatic mean (a), variation of rainfall with southern oscillation index (SOI) (b), and variation of aridity index (Ia) with SOI in Maharashtra 38 Figure 3. Seasonal mean NDVI and per cent under dry land crops grown in the drought prone districts in Maharashtra for the year 1992 . Figure 4. Monthly variations of NDVI and rainfall at four stations Satara, Pune, Dhule and Jalgaon in central Maharasthra 39 Figure 5. Seasonal integrated NDVI and Rainfall in a few stations in Maharasthra in a) 1992 and b) 1995 40 Figure 6. Monthly trends of NDVI in 1995 for a) February , b) September c) October and d) spatial pattern of climatic moist period in Maharasthra. 41 Figure 7. Trend of Growing period estimated from NDVI with the Growing period estimated using FAO method in the study region. 42 Figure 8. Surface albedo in Maharasthra in 1995 for a) April and b) November months. Figure 9. Land surface temperatures in Maharasthra in 1995 for a) April and b) November months. 43 Figure 10. Dryness Quotient in different districts in Maharasthra for a) 1992 and b) 1995. 44