×

A study of rainfall depth in Kerala, India. (English) Zbl 0846.62082

Publication. Centre for Mathematical Sciences, Trivandrum. 29. Trivandrum, Kerala: Centre for Math. Sciences, iv, 53 p. (1995).
Summary: The present study focuses attention on the construction of different types of models for the rainfall data recorded at the Trivandrum meteorological station, Kerala state, India. Using the hourly rainfall data for nine years several types of models are constructed. The first type deals with the total hourly rainfall depth assuming an exponential distribution for hourly rainfall depths. Various parameters are also estimated using the actual data. The second type is where first and second order Markov chain models are constructed for the hourly rainfall data. The initial, first and second order transition probabilities are estimated and the chi-square goodness of fit test is also carried out. All these results are included in chapter 2. In chapter 3, the rainfall data for the years 1973 to 1981 are analysed by using time series techniques. Autoregressive integrated moving average models are constructed for each of the five variates. The other stages of model building, namely estimation of parameters using conditional and unconditional sums of squares, diagnostic checking and forecasting are also done.
In chapter 4 we focus attention on the normalized rainfall curve (NRC) depicting the association between cumulated percentage rain amount \((x)\) and cumulated percentage number of rain days \((y)\) of the rainfall series. We have fitted two nonlinear models to the rainfall data for the years 1973 to 1981, collected from the meteorological station, Trivandrum, and made a comparison between the predicted values and the observed values. Thereafter by utilizing these models, various parameters of the daily rainfall distribution are also studied. All these results are included in chapter 4.
In chapter 5 we present the following new results which are not yet available in the published literature. We introduce a modified nonlinear model for representing the normalized rainfall curve and fitted this model to Trivandrum rainfall series (1973-1981) for the months of June and November, by utilizing an algorithm for nonlinear least squares. Comparison between the observed and analytical values of the model is also made and it is shown that the modified model best fits the daily rainfall series. Moreover by using this modified representation we evaluated the rain intensities at specified points on the normalized rainfall curve. Two alternative models are also suggested for comparison.

MSC:

62P99 Applications of statistics
86A10 Meteorology and atmospheric physics
86A32 Geostatistics
PDFBibTeX XMLCite