CN103353353B - Method for detecting near-surface average temperature based on MODIS data - Google Patents

Abstract

The present invention provides a method for detecting a near-surface average temperature based on MODIS data. The method comprises: A. obtaining an EVI' based on the MODIS data, the EVI' being an EVI of eight days; B. obtaining an EVIc according to a near-surface average temperature actually measured by a weather station and a corresponding LST and a corresponding EVI in the MODIS data, the EVIc being an EVI of full vegetation coverage; and C. calculating a near-surface temperature Ta according to a formula 8 which is defined in the description, wherein the Ts<day> and the Ts<night> are obtained based on the MODIS data. Compared with statistical methods, split-window algorithms and thermodynamic methods, the algorithm of the invention requires only one undetermined parameter EVIc, and the module is simple and easy to use, providing good expansibility. Besides, compared with the temperature-vegetation index method, the algorithm of the invention needs not to open up a calculation window, so that the self-similarity of near-surface temperature inversion results is reduced, and the calculation efficiency is improved; what's more, the influence caused by poor effects of Ts-NDVI linear fitting in the calculation window is prevented.

Description

A kind of method based on MODIS data snooping near surface temperature on average

Technical field

The present invention relates to the method utilizing satellite remote sensing to estimate near surface temperature, particularly relate to the near surface temperature evaluation method based on MODIS data.

Background technology

Air themperature (the T near surface 2 meters of _a) be an important meteorologic parameter, it affects energy circulation and the Water Cycle of ground vapour interactive system, plays very important effect to the change in animals and plants habitat.Conventional observation T _amode be utilize fixed or portable weather station, for spatially continuous print parameter, there is a lot of shortcoming in this metering system, the density of such as meteorological site, the deficiency of uniformity coefficient and spatial representative etc., and can cause cannot the spatial-temporal characteristics of accurate description air temperature field.The feature that remote sensing technology was observed on a large scale because of its moment, is expected to the deficiency that can make up conventional observation mode.There are four large classes both at home and abroad based on the T of remote sensing technology at present _aevaluation method:

(1) statistical method.Set up T _aand the statistical relationship between its major influence factors, such as T _awith surface temperature, vegetation index, landform altitude, solar zenith angle etc.The statistical relationship of this method establishment depends on training sample and training area, and Generalization Capability is lower.

(2) Split-window algorithm.Be similar to remote-sensing inversion surface temperature, the atmospheric envelope top radiation brightness utilizing remote sensing to obtain, by Split-window algorithm, inverting near surface temperature.The difficulty of Split-window algorithm is to be difficult to distinguish the contribution of near surface atmospheric temperature and the contribution of surface temperature in the radiation brightness of atmospheric envelope top, and in addition on thermal radiation transmission path, the impact of steam, gasoloid etc. is also difficult to eliminate.

(3) thermodynamic methods, based on Land surface energy budget equation, the Parameter Expression scheme of derivation near surface temperature.When the method calculates near surface temperature, need the auxiliary parameter inputting a lot of non-remote sensing.

(4) temperature-vegetation index.When the method supposes full vegetative coverage, the surface temperature of Vegetation canopy is identical with temperature, and then in a computing window, set up the relation of temperature and vegetation index, and when this relation is extended to full vegetative coverage, namely surface temperature is now approximately temperature.But this method supposes that soil moisture conditions in computing window, atmospheric conditions are homogeneous, if window humidity condition is inconsistent, so will be difficult to the linear equation simulating temperature and vegetation index.

Summary of the invention

The present invention is intended to a kind of method proposing detection near surface temperature on average based on MODIS data.

The method of the detection near surface temperature on average based on MODIS data of the present invention, comprising:

A, be 8 days EVI by MODIS data acquisition EVI ', EVI ';

LST and EVI corresponding in B, the near surface temperature on average of being surveyed by weather station and MODIS data obtains EVI _c; EVI _cfor EVI during full vegetative coverage;

C, through type 8 calculate near surface temperature T _a;

$T_a=\frac{1-{\mathrm{EVI}}_c}{1-{\mathrm{EVI}}^{\&prime;}}({T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}})+{T_s}^{\mathrm{night}}$ (formula 8);

Wherein, T _s ^dayand T _s ^nightby MODIS data acquisition.

Preferably, by MODIS data acquisition EVI ' time, first the inferior quality in LST and the EVI data in MODIS data or invalid data are filtered.

Preferably, the decision condition of inferior quality or invalid LST data is formula 1;

$\left\{\begin{array}{c}b1=(\mathrm{QA}/2)\mathrm{mod}2\\ b2=(\mathrm{QA}/2^2)\mathrm{mod}2\\ b3=(\mathrm{QA}/2^3)\mathrm{mod}2\\ b5=(\mathrm{QA}/2^5)\mathrm{mod}2\\ b7=(\mathrm{QA}/2^7)\mathrm{mod}2\\ p=\mathrm{where}\left((b1=1)\mathrm{or}(b2=1)\mathrm{or}(b3=1)\mathrm{or}(b5=1)\mathrm{or}(b7=1)\right)\end{array}\right.$ (formula 1);

Wherein, QA is MOD11A2 quality document, and p is the Data Position of inferior quality or invalid LST; B1 is QA data point the 2nd, and b2 is QA data point the 3rd; B3 is QA data point the 4th; B5 is QA data point the 6th; B7 is QA data point the 8th.

Preferably, the decision condition of inferior quality or invalid EVI data is formula 2;

$\left\{\begin{array}{c}{b2}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^2)\mathrm{mod}2\\ {b3}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^3)\mathrm{mod}2\\ {b4}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^4)\mathrm{mod}2\\ {b5}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^5)\mathrm{mod}2\\ p^{\&prime;}=\mathrm{where}\left(({b2}^{\&prime;}=1)\mathrm{and}({b3}^{\&prime;}=1)\mathrm{and}({b4}^{\&prime;}=1)\mathrm{or}({b5}^{\&prime;}=1)\right)\end{array}\right.$ (formula 2);

Wherein, QA ' is MOD13A2 or MYD13A2 quality document, the Data Position that p ' is inferior quality or invalid EVI; B2 ' is QA ' data point the 3rd, and b3 ' is QA ' data point the 4th; B4 ' is QA data point the 5th; B5 ' is QA ' data point the 6th.

Preferably, the 16 days synthesis EVI provided respectively according to Terra and Aqua satellite have 8 days overlap periods and obtain EVI '.

Preferably, EVI _cthrough type 3 calculates and obtains,

${\mathrm{EVI}}_c=1-\frac{{T_a}^{\mathrm{avg}}-{T_s}^{\mathrm{night}}}{{T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}}}\&times;(1-{\mathrm{EVI}}^{\&prime;})$ (formula 3);

T _a ^avgfor the near surface temperature on average of weather station actual measurement.

Preferably, EVI _cthrough type 3 calculates and obtains,

${\mathrm{EVI}}_c=a\&times;\sin (\frac{2\mathrm{\&pi;}}{365.25}\&times;\mathrm{DOY}+b)+c$ (formula 4);

Wherein, a, b and c are the EVIs corresponding according to different DOY _c, obtain according to least square fitting; DOY is the number of days of certain date in 1 year.

Preferably, for the urban area within five rings, Beijing, a=-0.13, b=1.41, c=0.70.

Algorithm of the present invention, compared with statistical method, Split-window algorithm, thermodynamic methods, only needs a undetermined parameter EVI _c, model is simple and easy to use, and scalability is better.Compared with temperature-vegetation index, algorithm of the present invention does not need to open up computing window, because this reducing the self-similarity of near surface temperature inversion result, improve operation efficiency, being priorly that of avoiding the impact that in computing window, Ts-NDVI linear fit effect is poor.

Accompanying drawing explanation

Fig. 1 is the schematic diagram of the method for the detection near surface temperature on average based on MODIS data of the present invention;

Fig. 2 a is EVI _cparametrization value and the comparison diagram of measured value; Fig. 2 b is EVI _cparametrization value and the deviation profile frequency plot of measured value;

Fig. 3 is the result of detection figure (2009-6-18 ~ 2009-6-25) of the method for the detection near surface temperature on average based on MODIS data of the present invention;

Fig. 4 a is T _ameasured value and the scatter diagram of estimated value; Fig. 4 b is T _ameasured value and the probability distribution graph of estimated value deviation; Fig. 4 c is T _ameasured value and the time variation diagram of estimated value deviation;

The process flow diagram of the method for Fig. 5 detection near surface temperature on average based on MODIS data of the present invention.

Embodiment

Below, by reference to the accompanying drawings the present invention is specifically described.

In the present invention, span near the ground is 1.5-2 rice overhead.

The invention process regional choice is the urban area within five rings, Beijing, and remotely-sensed data is the MOD11A2 of MODIS2000-2008, MOD13A2, and MYD13A2 data, and actual measurement temperature is within five rings, Beijing and the weather station measured value of neighboring area.

1. method flow

The process flow diagram of the method for Fig. 5 detection near surface temperature on average based on MODIS data of the present invention, comprising:

A, MODIS quality of data is filtered

According to the quality document of MODIS product, by the decimal system to binary translation function, realize the automatic fitration of inferior quality and invalid data, and the automatic extraction of the data that conform to quality requirements.

Inferior quality or invalid LST decision condition as follows:

$\left\{\begin{array}{c}b1=(\mathrm{QA}/2)\mathrm{mod}2\\ b2=(\mathrm{QA}/2^2)\mathrm{mod}2\\ b3=(\mathrm{QA}/2^3)\mathrm{mod}2\\ b5=(\mathrm{QA}/2^5)\mathrm{mod}2\\ b7=(\mathrm{QA}/2^7)\mathrm{mod}2\\ p=\mathrm{where}\left((b1=1)\mathrm{or}(b2=1)\mathrm{or}(b3=1)\mathrm{or}(b5=1)\mathrm{or}(b7=1)\right)\end{array}\right.$ (formula 1)

Wherein, QA is MOD11A2 quality document, and p is the Data Position of inferior quality or invalid LST; B1 is QA data point the 2nd, and b2 is QA data point the 3rd; B3 is QA data point the 4th; B5 is QA data point the 6th; B7 is QA data point the 8th.

The decision condition of inferior quality or invalid EVI is as follows:

$\left\{\begin{array}{c}{b2}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^2)\mathrm{mod}2\\ {b3}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^3)\mathrm{mod}2\\ {b4}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^4)\mathrm{mod}2\\ {b5}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^5)\mathrm{mod}2\\ p^{\&prime;}=\mathrm{where}\left(({b2}^{\&prime;}=1)\mathrm{and}({b3}^{\&prime;}=1)\mathrm{and}({b4}^{\&prime;}=1)\mathrm{or}({b5}^{\&prime;}=1)\right)\end{array}\right.$ (formula 2)

Wherein, QA ' is MOD13A2 or MYD13A2 quality document, the Data Position that p ' is inferior quality or invalid EVI; B2 ' is QA ' data point the 3rd, and b3 ' is QA ' data point the 4th; B4 ' is QA ' data point the 5th; B5 ' is QA ' data point the 6th.

B, generate EVI ', EVI ' be 8 days EVI

Terra and Aqua two satellites each provide the EVI of one group of synthesis in 16 days, and these two groups of EVI have the overlap period of 8 days.Two overlapping 16 days EVI are averaged, obtain the EVI of synthesis in approximate 8 days, i.e. EVI '.

MOD13A2, from the annual the 001st day, obtained the EVI of single sintering every 16 days.MYD13A2, from the annual the 009th day, obtained the EVI of single sintering every 16 days.MOD13A2-001 and MYD13A2-009 averages and obtains 8 days EVI that synthesis phase is 009 to 016.MYD13A2-009 and MOD13A2-017 is averaged, and obtains 8 days EVI that synthesis phase is 017-024.The like, obtain 8 days annual EVI.

C, EVI _cdetermination, EVI _cbe full vegetative coverage EVI

According to the near surface temperature on average of weather station actual measurement, to EVI (full vegetation cover EVI, the EVI of full vegetative coverage _c) carry out parametrization, EVI is provided _cestimation equation.

First, according to the weather station observed reading of near surface temperature on average, and the remote sensing of correspondence surface temperature and EVI round the clock, calculate EVI according to following formula _c:

${\mathrm{EVI}}_c=1-\frac{{T_a}^{\mathrm{avg}}-{T_s}^{\mathrm{night}}}{{T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}}}\&times;(1-{\mathrm{EVI}}^{\&prime;})$ (formula 3)

EVI in formula _cfor EVI during full vegetative coverage, T _a ^avgfor the near surface temperature on average of weather station actual measurement, T _s ^dayand T _s ^nightbe respectively surface temperature round the clock.

Then, according to EVI _ctime dependent trend, builds following EVI _cparametrization estimation scheme:

${\mathrm{EVI}}_c=a\&times;\sin (\frac{2\mathrm{\&pi;}}{365.25}\&times;\mathrm{DOY}+b)+c$ (formula 4)

Wherein, a, b and c are the EVIs corresponding according to different DOY _c, obtain according to least square fitting; DOY is the number of days of certain date in 1 year.For the urban area within five rings, Beijing, a=-0.13, b=1.41, c=0.70.

D, near surface temperature are estimated

Input 8 days average surface temperatures round the clock and 8 days vegetation indexs, calculate and export near surface 8 days temperature on average.

Theoretical according to day and night temperature-vegetation index feature space, derive and draw the evaluation method of near surface temperature.By certain extensive scheme, the LST round the clock and 8 day EVI that the method only needs input 8 days average, can calculate and export near surface 8 days temperature on average.

Fig. 1 is the schematic diagram of the method for the detection near surface temperature on average based on MODIS data of the present invention, wherein ABCD is day and night temperature-vegetation index feature space, P is any point in feature space, suppose the temperature difference of earth's surface round the clock that P point position is corresponding and round the clock air Temperature Difference be respectively Δ T _swith Δ T _a, so theoretical according to feature space, have following formula to set up:

$\frac{{\mathrm{\&Delta;T}}_s-{{\mathrm{\&Delta;T}}_s}^{\mathrm{wet}}}{{{\mathrm{\&Delta;T}}_s}^{\mathrm{dry}}-{{\mathrm{\&Delta;T}}_s}^{\mathrm{wet}}}=\frac{{\mathrm{\&Delta;T}}_a-{{\mathrm{\&Delta;T}}_a}^{\mathrm{wet}}}{{{\mathrm{\&Delta;T}}_a}^{\mathrm{dry}}-{{\mathrm{\&Delta;T}}_a}^{\mathrm{wet}}}$ (formula 5)

In formula,

$\left\{\begin{array}{c}{\mathrm{\&Delta;T}}_s={T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}}\\ {\mathrm{\&Delta;T}}_a={T_a}^{\mathrm{day}}-{T_a}^{\mathrm{night}}\\ {{\mathrm{\&Delta;T}}_s}^{\mathrm{dry}}=e\&times;{\mathrm{EVI}}^{\&prime;}+f\\ {{\mathrm{\&Delta;T}}_a}^{\mathrm{dry}}=e\&times;{\mathrm{EVI}}_c^{\&prime;}+f\end{array}\right.$ (formula 6)

When supposing that humidity is very big round the clock the earth's surface temperature difference and round the clock air Temperature Difference be 0, and EVI greatly (equaling 1) time round the clock the earth's surface temperature difference be 0, so e=-f, Δ T _s ^night=Δ T _a ^wet=0.Therefore obtain

${T_a}^{\mathrm{day}}=\frac{1-{\mathrm{EVI}}_c}{1-{\mathrm{EVI}}^{\&prime;}}({T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}})+{T_a}^{\mathrm{night}}$ (formula 7)

Because nighttime surface temperature and temperature have the correlativity of highly significant, by the T in above formula _a ^nightuse T _s ^nightreplace, and suppose the error that this replacement causes, by regulating EVI _cand reduce.Thus the evaluation method obtaining near surface temperature of the present invention is as follows:

$T_a=\frac{1-{\mathrm{EVI}}_c}{1-{\mathrm{EVI}}^{\&prime;}}({T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}})+{T_s}^{\mathrm{night}}$ (formula 8)

In formula, T _afor near surface temperature, EVI _cfor EVI during full vegetative coverage, T _s ^dayand T _s ^nightbe respectively surface temperature round the clock, T _s ^dayand T _s ^nightby MODIS data acquisition.

2. the superiority of interpretation of result and this method

Utilize the MOD11A2 of 2000-2008, MOD13A2, and the temperature of MYD13A2 data to region, Beijing city is estimated.Fig. 2 a is EVI _cparametrization value and the comparison diagram of measured value; Fig. 2 b is EVI _cparametrization value and the deviation profile frequency plot of measured value;

Fig. 3 is the result of detection figure (2009-6-18 ~ 2009-6-25) of the method for the detection near surface temperature on average based on MODIS data of the present invention;

Fig. 4 a is T _ameasured value and the scatter diagram of estimated value; Fig. 4 b is T _ameasured value and the probability distribution graph of estimated value deviation; Fig. 4 c is T _ameasured value and the time variation diagram of estimated value deviation.

Test result shows, the precision for test zone algorithm of the present invention is higher, and RMSE reaches 1.41K, and mean absolute deviation is the coefficient R of 1.1K, estimated value and measured value ²reach 0.98.Algorithm of the present invention, compared with statistical method, Split-window algorithm, thermodynamic methods, only needs a undetermined parameter EVI _c, model is simple and easy to use, and scalability is better.Compared with temperature-vegetation index, algorithm of the present invention does not need to open up computing window, because this reducing the self-similarity of near surface temperature inversion result, improve operation efficiency, being priorly that of avoiding the impact that in computing window, Ts-NDVI linear fit effect is poor.

Claims (8)

1., based on a method for the detection near surface temperature on average of MODIS data, it comprises:

A, be 8 days EVI by MODIS data acquisition EVI ', EVI ';

LST and EVI corresponding in B, the near surface temperature on average of being surveyed by weather station and MODIS data obtains EVI _c; EVI _cfor EVI during full vegetative coverage;

C, through type 8 calculate near surface temperature T _a;

$T_a=\frac{1-{\mathrm{EVI}}_c}{1-{\mathrm{EVI}}^{\&prime;}}({T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}})+{T_s}^{\mathrm{night}}$ (formula 8);

Wherein, T _s ^dayand T _s ^nightbe respectively daytime, night surface temperature, T _s ^dayand T _s ^nightby MODIS data acquisition.

2., as claimed in claim 1 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

By MODIS data acquisition EVI ' time, first the inferior quality in LST and the EVI data in MODIS data or invalid data are filtered.

3., as claimed in claim 2 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

The decision condition of inferior quality or invalid LST data is formula 1;

$\left\{\begin{array}{c}b1=(\mathrm{QA}/2)\mathrm{mod}2\\ b2=(\mathrm{QA}/2^2)\mathrm{mod}2\\ b3=(\mathrm{QA}/2^3)\mathrm{mod}2\\ b5=(\mathrm{QA}/2^5)\mathrm{mod}2\\ b7=(\mathrm{QA}/2^7)\mathrm{mod}2\\ p=\mathrm{where}\left((b1=1)\mathrm{or}(b2=1)\mathrm{or}(b3=1)\mathrm{or}(b5=1)\mathrm{or}(b7=1)\right)\end{array}\right.$ (formula 1);

Wherein, QA is MOD11A2 quality document, and p is the Data Position of inferior quality or invalid LST; B1 is QA data point the 2nd, and b2 is QA data point the 3rd; B3 is QA data point the 4th; B5 is QA data point the 6th; B7 is QA data point the 8th.

4., as claimed in claim 2 or claim 3 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

The decision condition of inferior quality or invalid EVI data is formula 2;

$\left\{\begin{array}{c}{b2}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^2)\mathrm{mod}2\\ {b3}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^3)\mathrm{mod}2\\ {b4}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^4)\mathrm{mod}2\\ {b5}^{\&prime;}=({\mathrm{QA}}^{\&prime;}/2^5)\mathrm{mod}2\\ p^{\&prime;}=\mathrm{where}\left(({b2}^{\&prime;}=1)\mathrm{and}({b3}^{\&prime;}=1)\mathrm{and}({b4}^{\&prime;}=1)\mathrm{or}({b5}^{\&prime;}=1)\right)\end{array}\right.$ (formula 2);

Wherein, QA ' is MOD13A2 or MYD13A2 quality document, the Data Position that p ' is inferior quality or invalid EVI; B2 ' is QA ' data point the 3rd, and b3 ' is QA ' data point the 4th; B4 ' is QA ' data point the 5th; B5 ' is QA ' data point the 6th.

5., as claimed in claim 1 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

According to 16 days synthesis EVI that Terra and Aqua satellite provides respectively, there is 8 days overlap periods, obtain described EVI '.

6., as claimed in claim 1 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

EVI _cthrough type 3 calculates and obtains,

${\mathrm{EVI}}_c=1-\frac{{T_a}^{\mathrm{avg}}-{T_s}^{\mathrm{night}}}{{T_s}^{\mathrm{day}}-{T_s}^{\mathrm{night}}}\&times;(1-{\mathrm{EVI}}^{\&prime;})$ (formula 3);

T _a ^avgfor the near surface temperature on average of weather station actual measurement.

7., as claimed in claim 1 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

EVI _cthrough type 4 calculates and obtains,

${\mathrm{EVI}}_c=a\&times;\sin (\frac{2\mathrm{\&pi;}}{365.25}\&times;\mathrm{DOY}+b)+c$ (formula 4);

Wherein, a, b and c are the EVIs corresponding according to different DOY _c, obtain according to least square fitting; DOY is the number of days of certain date in 1 year.

8., as claimed in claim 7 based on the method for the detection near surface temperature on average of MODIS data, it is characterized in that:

For the urban area within five rings, Beijing, a=-0.13, b=1.41, c=0.70.