Drought Monitoring in India and the Philippines with Satellite Remote Sensing Measurements

Droughts are normal recurring climatic phenomena that vary in space, time, and intensity. The spatial and temporal variability and multiple impacts of droughts provide challenges for mapping and monitoring on regional scales. With the launch of new generation sensors such as Moderate Resolution Imaging Spectroradiometer (MODIS), the monitoring of aberrant climatic events may be explored in an efficient way. An empirical method called Temperature Vegetation Dryness Index (TVDI) was used for drought monitoring in two different countries. In this paper we demonstrate the usefulness of the TVDI approach and of MODIS products for the identification of drought conditions in affected states of India and the Philippines. The satellite derived results reveal that they can detect and monitor the drought accurately. The results were compared with the crop yield for validation of remotely sensed measurements for drought detection. In Gujarat state, the drought years showed a negative yield anomaly as compared to a normal year. The average yield anomaly in 2002 was -11.2 and -35.49 for food grains and oil seeds respectively. This indicates the influence of drought on yields to a greater extent. But in 2003, it gives a positive yield anomaly and indicates no drought effect on crop yields. In Iloilo province, the results revealed that the rice area and the production reduced due to the drought in 2000 as compared to other normal years. Overall, the results imply that the satellite derived drought index (TVDI) is a useful tool for the identification of drought affected areas in real time using satellite measurements. INTRODUCTION Drought is a natural hazard that impacts economic, social, and environmental aspects of society. In the agricultural sector, it is one of the dominant causes of crop loss. Although a drought first appears as below-average rainfall within a normal part of climate, it can develop as an extreme climatic event and turn into a hazardous phenomenon which can have a severe impact on communities and water-dependent sectors. Droughts are a recurrent feature of the climate, varying in intensity, duration, and frequency across the climatic spectrum. Knowledge about the timing, severity, and pattern of droughts on the landscape can be incorporated into effective planning and decision making. To monitor droughts, decision-makers at the administrative and grass-roots levels need timely and accurate information about the spatial and temporal dimensions of droughts. This information helps officials and farmers to be more proactive in managing drought risks (1). Furthermore, drought impacts can be reduced through better understanding of drought and identifying the appropriate drought indicators for an early warning system. This includes providing decisionmakers with timely drought products (e.g., maps and data) that identify the frequency, severity, and spatial extent of droughts. In the past, climate and meteorological data were the primary sources for drought information used to support decision-making. However, satellite observations have recently proved to be a valuable source of timely, spatially continuous data with improved details for monitoring vegetation dynamics over large areas. Many prior studies of vegetation conditions base analyses on numerical transforms known as vegetation indices (VI). These indices have been used for studying vegetation characteristics over large areas since the 1970s (2). Additional studies have presented analyses of droughts in the USA, Africa, South America, and Asia and illustrate how derivatives of the normalEARSeL eProceedings 7, 1/2008 82 ised difference vegetation index (NDVI) can improve the ability to observe droughts in time-series satellite data (3,4). Remote sensing technology is an economical and promising tool for obtaining land surface parameters. Remote sensing technology used to assess or monitor regional drought is mainly based on an index that is a function of spectral vegetation index or land surface temperature. Wang (5) concluded that drought information is not closely related to NDVI data and that a drought index based on NDVI should be insensitive to the soil moisture status. A drought index based on land surface temperature (Ts) should be more efficient than those based on NDVI. A drought index based on normalised difference vegetation index (NDVI) falls short in monitoring a drought because NDVI is a rather conservative indicator of water stress, which means that vegetation remains green after initial water stress (6). In contrast, Ts is more sensitive to water stress (7). The combination of NDVI and Ts provides information on the vegetation and moisture status. The scatter plot of remotely sensed temperature and spectral vegetation index often exhibits a triangular (8) or trapezoidal (9) shape and is called the NDVI-Ts space if a full range of fractional vegetation cover and soil moisture content is represented. In this paper, based on NDVI-Ts space, the temperature vegetation dryness index (TVDI) has been developed for a drought monitoring approach in drought affected states of India and the Philippines using Terra/MODIS measurements. MODIS provides a unique opportunity for global assessment and monitoring of vegetation in every eight days at 1 km spatial resolution. The advantage of MODIS data is to develop a prototype for a near real time drought monitoring system at the scale of a country, state, district or pixel with an 8or 16-day time interval. The results described feed directly into the development of the regional drought monitoring system (10). The objective of MODIS mission is to improve predictions and characterisations of natural disasters like droughts. MODIS is expected to determine the land surface temperature accurately and by integrating MODIS thermal infrared data into land surface monitoring two main problems in current drought monitoring schemes can be addressed. Firstly, accurate temperature observations from remotely sensed data can overcome very coarse spatial resolutions of weather stations at relative low costs. Secondly it can be an appropriate tool for real time drought monitoring, which has not been accomplished successfully by current remotely acquired measures, such as vegetation indices, due to a lagged vegetation response to drought (11). In this study, an empirical method called Temperature Vegetation Dryness Index (TVDI) has been developed for a drought monitoring approach in affected states of India and the Philippines. The sub-objectives of this study are: (a) Mapping drought indices over the western states of India and Western Visayas of the Philippines; (b) to deliver timely geo-referenced information (in the form of maps and data) about areas where the vegetation is impacted by drought. METHODS Study area location and drought propensity The drought affected areas in India and the Philippines have been chosen for the drought analysis using MODIS satellite data. In India, the Gujarat state has been selected for this study which was one of the most drought affected states in recent times. In addition, Assam state in India has also been selected for the drought analysis. In the Philippines, Western Visayas has been selected for timely drought measurements. Gujarat state is located in the north west of India between 20°01'N to 24°07'N latitude and 68°04'E to 74°04'E longitude (Figure 1). The tropic of Cancer passes through the northern border of the state. The two deserts, one north of Kutch and the other between Kutch and the mainland Gujarat are saline wastes. It covers a total geographical area of 196024 km and accounts for 6.19% of the total area of the country. The climate of Gujarat is moist in the southern districts and dry in the northern region. The year can be divided into: the winter season from November to February, the hot season from March to May, the south-west monsoon season from June to September and the intervening month of October. The average rainfall in Gujarat varies from 33 to 152 cm. The region of Kutch can be described as a desert-like area. In the summer the average temperature is between 25° to 43°C and reaches as high as 48°C. Gujarat is a water scarce region under constant threat of drought, and the availability of water is an ongoing issue of struggle for the people. EARSeL eProceedings 7, 1/2008 83 Drought being a common occurrence, agriculture fails to support livelihood solely. The incidence of drought has become a regular feature, and any 5-year cycle has 2-3 years of drought. Assam state is located in the north-eastern part of India between 24°50'N to 28°00'N latitude and 89°42'E to 96°00'E longitude (Figure 1). It is surrounded on all other sides by predominantly hilly or mountainous tracts. During the monsoon the climate is warm and humid. The Brahmaputra river flows through the entire length of the State from east to west. Assam is mainly an agrarian state, 89% of the people live in rural areas. Drought in the northeast of India is an exemption in terms of drought because this area is hit by a flood in nearly every year due to heavy rainfall. The scenario was just opposite in 2006 due to scanty rainfall in most of the districts in Assam. Figure 1: Location of study area Gujarat and Assam state, India. The Philippine Republic’s Region VI, Western Visayas, comprises six provinces: Negros Occidental, Guimaras, Iloilo, Capiz, Antique, and Aklan. It is located in Central Philippines between two inter-island bodies of water: the Sibuyan Sea and Visayas Sea. Geographically, the region is defined by grid coordinates 121°5'E 123°2'E longitude and 9°25'N 12°12'N latitude (Figure 2). This area is under Region VI, which is an agricultural region. In the Philippines, El Niño events are associated with conditions drier than normal which cause dry spells or even drought. The effects of El Niño can be felt in various sectors of the country: agriculture, environment, water resources, energy and health. The agricultural sector is most vulnerable to drought. Climatological studies showed that major drought events in the Philippines are associated with El Niño occurrences or warm episodes in the central and eastern equatorial Pacific. Provinces in the western portions of the country climate are characterised by two pronounced seasons, dry and wet, with maximum rain period from June to September due to the extension of Southwest monsoon. Seasonal aridity is exacerbated by the increasing incidence of El Niño, which is now occurring at a two to three-year cycle from previous five-year intervals. EARSeL eProceedings 7, 1/2008 84 Figure 2: Location of study area western Visayas, Philippines. Satellite data used for remotely sensed measurements The satellite data we used were Terra/MODIS land surface temperature and surface reflectance. The 8-day composites cloud-free data were downloaded from EOS data gateway during 2000, 2002, and 2006. The MODIS products used for this study are as follows: a) MOD11A2-LST 8day composite images (1km resolution) b) MOD09Q1Surface Reflectance 8day composite images (250m resolution) Land surface temperature (LST or Ts) Land surface temperature (LST) is generally defined as the skin temperature of the ground. For the bare soil surface, LST is the soil surface temperature; for dense vegetated ground, LST can be viewed as the canopy surface temperature of the vegetation; and in sparse vegetated ground, LST is determined by the temperature of the vegetation canopy, vegetation body and the soil surface (12). LST is a very useful input for modelling energy balance components and mapping evapotranspiration (ET). Retrieval of LST using thermal IR bands of satellite images is the most effective way to derive energy balance and ET on a regional basis. The land surface temperature is an important factor controlling most physical, chemical and biological processes in the Earth. Vegetation Index It is a measure of the amount and vigor of the vegetation at the surface. The magnitude of NDVI is related to the level of photosynthetic activity in the observed vegetation. In general, higher values of NDVI indicate greater vigor and amounts of vegetation. It is defined as: 2 1 2 1 band band NDVI band band − = + , where and band . 2 858 nm band = 1 645 nm = NDVI is a good indicator of green biomass, leaf area index, and patterns of production (10). NDVI was computed using two bands of a surface reflectance image, which varies from -1 to +1. The MODIS-LST and surface reflectance products were geocorrected using HEGTools and the scale factors were multiplied to all the satellite images. The corrected surface reflectance and LST images were then used for NDVI and TVDI computation. EARSeL eProceedings 7, 1/2008 85 Temperature Vegetation Dryness Index (TVDI) TVDI is a simple and effective method for regional drought monitoring. In this study, the statistic characteristics of MODIS-NDVI, LST at different times and locations have been analysed and compared. In the current study, we used MODIS-NDVI as vegetation index for TVDI. Figure 3: Schematic plot of Temperature vegetation dryness index proposed by Sandholt (6). Following the concept in Figure 3, isolines can be drawn in the triangle defining the Ts/NDVI space. As a first iteration to obtain information on the surface soil moisture content, a dryness index (TVDI) having the values of 1 at the ‘‘dry edge’’ (limited water availability) and 0 at the ‘‘wet edge’’ (maximum evapotranspiration and thereby unlimited water access) can be defined: