Study of Temporal Behaviour
of Homogeneity Maps for Estimating
Representative Area of a Ground Sample
Using Remote Sensing
Prasad J. Deshpande, Anudeep Sure, Onkar Dikshit, and Shivam Tripathi
Abstract Modelling and prediction of hydro-meteorological variables over land and
atmosphere involve ground sampling at selected locations over the study area. Optimally selecting the number and location of sampling points is important for making
reliable predictions without escalating project costs. This study proposes an approach
for selecting sampling locations by considering inter-dependency of predictor variables and the prediction variable using remote sensing data. A homogeneity map,
i.e., a thematic map representing areas with the same expected value of the prediction
variable, with a given level of uncertainty and spatial resolution, is generated. The
homogeneity maps can be different at different times for the same location. Thus,
along with the spatial variability of the prediction variable, its temporal variability
is also obtained. Depending on the obtained variability, a decision on the number
and location of sampling points can be taken prudently. In this paper, the proposed
methodology is demonstrated by considering soil moisture over an experimental
watershed as the prediction variable.
Keywords Hydro-meteorological variable · Regionalisation · Spatio-temporal
clustering · Heterogeneity · Google Earth Engine
1 Introduction
Prediction of hydro-meteorological variables is the process of estimating the variable
at unsampled locations by using observations at a few sampling locations. Hydrometeorological variables change their value spatially and temporally, and hence,
the selection of sampling locations is difficult. This paper proposes to use remotely
sensed variables that are related to prediction variable (i.e., auxiliary variables) for
selecting sampling locations. These auxiliary variables are used to find a representative area for each sample location. The representative area of a sample is the area
surrounding the sample having similar value as the sample. The representative area
P. J. Deshpande (B) · A. Sure · O. Dikshit · S. Tripathi
Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur, India
e-mail: prasadj@iitk.ac.in
© Springer Nature Singapore Pte Ltd. 2021
C. Bhuiyan et al. (eds.), Water Security and Sustainability,
Lecture Notes in Civil Engineering 115,
https://doi.org/10.1007/978-981-15-9805-0_9
93
of Homogeneity Maps for Estimating
Representative Area of a Ground Sample
Using Remote Sensing
Prasad J. Deshpande, Anudeep Sure, Onkar Dikshit, and Shivam Tripathi
Abstract Modelling and prediction of hydro-meteorological variables over land and
atmosphere involve ground sampling at selected locations over the study area. Optimally selecting the number and location of sampling points is important for making
reliable predictions without escalating project costs. This study proposes an approach
for selecting sampling locations by considering inter-dependency of predictor variables and the prediction variable using remote sensing data. A homogeneity map,
i.e., a thematic map representing areas with the same expected value of the prediction
variable, with a given level of uncertainty and spatial resolution, is generated. The
homogeneity maps can be different at different times for the same location. Thus,
along with the spatial variability of the prediction variable, its temporal variability
is also obtained. Depending on the obtained variability, a decision on the number
and location of sampling points can be taken prudently. In this paper, the proposed
methodology is demonstrated by considering soil moisture over an experimental
watershed as the prediction variable.
Keywords Hydro-meteorological variable · Regionalisation · Spatio-temporal
clustering · Heterogeneity · Google Earth Engine
1 Introduction
Prediction of hydro-meteorological variables is the process of estimating the variable
at unsampled locations by using observations at a few sampling locations. Hydrometeorological variables change their value spatially and temporally, and hence,
the selection of sampling locations is difficult. This paper proposes to use remotely
sensed variables that are related to prediction variable (i.e., auxiliary variables) for
selecting sampling locations. These auxiliary variables are used to find a representative area for each sample location. The representative area of a sample is the area
surrounding the sample having similar value as the sample. The representative area
P. J. Deshpande (B) · A. Sure · O. Dikshit · S. Tripathi
Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur, India
e-mail: prasadj@iitk.ac.in
© Springer Nature Singapore Pte Ltd. 2021
C. Bhuiyan et al. (eds.), Water Security and Sustainability,
Lecture Notes in Civil Engineering 115,
https://doi.org/10.1007/978-981-15-9805-0_9
93