Because the statistical unit of catch production is fishing grid whose size is 30’×30’or 10’×10’ in marine fishery

the horizontal temperature gradient of fishing grid need to be calculated in order to analyze fishing ground environment by scientists. But there are the questions of real distance

irregular grid （after fishing grid is projected to Mercator projection）

value of grid points （ not cell value） if current commercial GIS software is used to calculated temperature gradients of fishing grid. For example

the max relative error of temperature gradient of fishing grid reaches 26. 48%

and the max absolute error of horizontal temperature gradient of fishing grid is 0. 124℃/10km

which is caused by the question of real distance and calculated from SST isolines on January 1-4

2001 in Northwest Pacific Ocean. A new algorithm to calculate horizontal temperature gradient of fishing grid from SST isoline is presented in this paper. The basic principle of this new algorithm is as follow: Interpolating temperature of fishing grid points in geographical coordinate system in order to build regular grid and index lines which angle is 45 degrees. The multi-section algorithm is used to interpolate temperature of fishing grid points

and exceptional temperature values are dealt with single point change surface interpolation containing orientation. The coordinate of fishing grid points with interpolating temperature is projected using Mercator projection

and the regular fishing grid becomes irregular fishing grid. The eight direction horizontal temperature gradients are computed with temperature values of nine points on each fishing grid

and the length of any two points

which is an important parameter

uses geodetic distance of elliptic earth. The methods of validating results are as follows: （1） The interpolating temperature of grid points is validated with two methods. One is manual estimation with temperature values of fishing grid points overlaying original SST isoline. Another is to compare SST isoline by overlaying original SST isoline on SST isoline generated with interpolating temperature of grid points. The same methods are used to validate the interpolating temperature of grid points by ArcGIS TIN. The result by algorithm of this paper is better than results by ArcGIS TIN. （ 2 ） The horizontal temperature gradient of fishing grid is validated by manual estimation with horizontal temperature gradient values of fishing grid overlaying original SST isoline. The same methods are used to validate results by ArcGIS GRID the interpolating temperature of grid points by ArcGIS. The result by algorithm of this paper is better than the results by ArcGIS.