LMB Drought Monitor


The amount of rainfall and its spatial distribution are important for water resources assessment, flood and drought prediction. In many developing countries the availability of ground measuring stations is very limited and unevenly distributed, making the assessment of water resources and flood forecasting difficult. The availability of several high resolution global satellite based rainfall products by various operating agencies, such as Global Satellite Mapping of Precipitation (GSMaP), CMORPH, TRMM Multi satellite Precipitation Analysis, can provide information on the amount of precipitation and its spatial distribution in such data sparse regions. Among these, GSMaP provides the highest spatial resolution satellite-based products at a fine temporal scale.

Initially GSMaP was promoted as a research study for producing high resolution and high precision global precipitation maps using satellite data sponsored by JST (Japan Science and Technology). Since 2007, this project has been promoted and initiated by the Japan Aerospace Exploration Agency (JAXA) as Precipitation Measuring Mission.

Drought is an extended period when a region receives a deficiency in its water supply. This can happen anywhere in the world. Drought indices summarize associated data like rainfall, stream flow, etc. into a single number which can be used to obtain a comprehensible overview. Some well-known drought indices include: Palmer’s Index (Palmer, 1965), Keetch–Byram Drought Index (Keetch and Byram 1968), Standardized Precipitation Index (McKee et al. 1993, 1995), Vegetation Condition Index (Kogan 1998).

The development of the Remote Sensing Technology enabled scientists to formulate drought indices on the basis of remote sensing data with better spatial resolution and wider extent. One of theses indices is the Keetch–Byram Drought Index (KBDI). The KBDI was developed by John L. Keetch and George Bryam in 1968 (Keetch, John J; Byram, George, 1968, "A drought index for forest fire control"). Typically it is applied in potential wild-fire assessment in the US. KBDI reflects water gain or loss within the upper soil column using mean annual rainfall, daily rainfall and daily maximum temperature as input parameters. It is ranging from 0 (no drought) to 800 (extreme drought). The formulas to determine the KBDI are shown below,(where dF: drought factor [-]; dt: time step [d] (1 day); T: maximum temperature of previous day[C]; R: rainfall of previous day [mm])

Drought is not just row rainfall deficiency. It implies deviation from Normal Condition. Because of this, we have to compare Drought Indices such as KBDI in temporal dimension, which means Anomaly of Drought Indexes (KBDI). KBDI Anomaly map is produces from comparison of current KBDI map with respect to long term average KBDI map. This variation is given as a percentage in KBDI anomaly map.

The chlorophyll in plant leaves absorbs visible light (VIS) (from 0.4 to 0.7 μm) for the photosynthesis process. The cell structures of the leaves reflect near-infrared (NIR) light (from 0.7 to 1.1 μm). These properties can be presented as a ratio which is widely known as Normalized Difference Vegetation Index (NDVI). The NDVI varies between -1.0 and +1.0:


MODIS NDVI which is produced from MODIS data is one of the most popular NDVI data sets that is available for free with a 250m resolution. Band 1 and band 2 of the MODIS sensor are used as visible (red, VIS) and near-infrared (NIR) reflecting bands. In order to deal with the cloud cover problem, standard MODIS products are based on 16 day composite data.

Several remotely sensed indices such as Normalized Difference Vegetation Index (NDVI) and Vegetation Crop Index (VCI) are being used extensively for agronomic monitoring, in particular for vegetation stress and crop yield assessment. These satellite based indices cover large areas at high frequencies, they are economically superior to ground based measurements. The technology allows to easily monitor the spatial and temporal variation of drought related vegetation stress at regional, continental and even the global level.

VCI is an indicator of the vigor of vegetation covers as a function of NDVI values for a given area. It assesses changes in the NDVI signal through time due to weather conditions, reducing the influence of ‘geographic’ (Kogan, 1990) or ‘ecosystem’ (Kogan, 1995c) variables i.e. climate, soils, vegetation type and topography.

The VCI is expressed in percentage and provides a measure of the present observed value against the long term variability of values. Low and high values indicate bad and good vegetation state conditions, respectively. The formula for computing the Vegetation Crop Index is (where VCIj is the image of vegetation condition index values for date j; NDVIj is the image of NDVI values for date j; NDVImax and NDVImin are images of maximum and minimum NDVI values from all images within the data set):

The near-infrared band (NIR, MODIS band 2) provides the option for water and land boundary discrimination. Thresh-holding of NIR band images can be used to separate water (relatively dark pixels) from land (relatively light pixels).

Cloud shadows and terrain shadows possess a similar spectral signature as water. Due to this phenomenon, cloud shadows and terrain shadows will be misclassified as water if simple thresh-holding is employed.

MODIS spectral band 1, band 2 and band 7 which correspond to red, near-infrared (NIR) and mid-infrared (MIR) respectively, are used for this product. In order to reduce the cloud shadow miss-classification effect, several images that have been acquired in consecutive days are being used with the assumption that the clouds on these consecutive images do not coincide. The algorithm only identifies water bodies if these have been identified in consecutive images as water bodies. Even though, this technique cannot completely remove cloud shadows, it can remove a considerable amount of the cloud shadows.